Visualizing regional myocardial oxygenation changes with statistically optimal colormaps

Tsaftaris, Sotirios A; Tang, Richard; Klein, Rachel; Li, Debiao; Dharmakumar, Rohan

doi:10.1186/1532-429X-11-S1-P276

Visualizing regional myocardial oxygenation changes with statistically optimal colormaps

Richard Tang

2009, Journal of Cardiovascular Magnetic Resonance

https://doi.org/10.1186/1532-429X-11-S1-P276

visibility

…

description

316 pages

link

1 file

Abstract

Introduction: Prophylactic implantation of a cardioverter/ defibrillator (ICD) has been shown to reduce mortality in patients with chronic myocardial infarction (CMI) and an increased risk for life threatening ventricular arrhythmia (VA). The use of ICDs in this large patient population is still limited by high costs and possible adverse events including inappropriate discharges and progression of heart failure. VA is related to infarct size and seems to be related to infarct morphology. Contrast enhanced cardiovascular magnetic resonance imaging (ceCMR) can detect and quantify myocardial fibrosis in the setting of CMI and might therefore be a valuable tool for a more accurate risk stratification in this setting. Hypothesis: ceCMR can identify the subgroup developing VA in patients with prophylactic ICD implantation following MADIT criteria. Methods: We prospectively enrolled 52 patients (49 males, age 69 ± 10 years) with CMI and clinical indication for ICD therapy following MADIT criteria. Prior to implantation (36 ± 78 days) patients were investigated on a 1.5 T clinical scanner (Siemens Avanto © , Germany) to assess left ventricular function (LVEF), LV end-diastolic volume (LVEDV) and LV mass (sequence parameters: GRE SSFP, matrix 256 × 192, short axis stack; full LV coverage, no gap; slice thickness 6 mm). For quantitative assessment of infarct morphology late gadolinium enhancement (LGE) was performed including measurement of total and relative infarct mass (related to LV mass) and the degree of transmurality (DT) as defined by the percentage of transmurality in each scar. (sequence parameters: inversion recovery gradient echo; matrix 256 × 148, imaging 10 min after 0.2 μg/kg gadolinium DTPA; slice orientation equal to SSFP). MRI images were analysed using dedicated software (MASS © , Medis,

Figures (490)

Arrhythmia: For the 100 episodes of arrhythmia, 79 episodes were atrial fibrillation (AF), 14 episodes were supraventricular tachycardia (SVT), 6 episodes were ventricular tachycardia (VT),

Figure 2 (abstract O2) and | episode was ventricular fibrillation (VF). The mean cardiac T2* was 13.5 + 9 ms, mean liver T2* 6.0 + 6.4 ms, mean serum ferritin 2140 + 1540 ug/L, and the mean ejection fraction was 60.7 + 9.3%. In comparison with cardiac T2* values >20 ms, there was a significantly increased risk of arrhythmia associated with cardiac T2* values <6 ms (RR 8.65, P < 0.001) and T2*<20 (RR 4.6, P < 0.001). There was no significant predictive value using the conventional thresholds of ferritin (ferritin >2500 ug/L, RR 0.90, P = 0.66) or liver T2* (T2* < 0.96 ms, RR 0.78, P = 0.68). (See Figure 2.).

*p < 0.05 baseline compared to visit with common cold. Table | (abstract O03) LV volume and CMR markers for edema and inflammation at baseline, with a common cold and at a 4-week follow-up. Volumetric data are presented as mean standard deviation

Reduced contractile function in an elite professional swimmer with a common cold (LVEF cig 47%, LVEFpaseline 24%) (A, B). Visually apparent increased early enhancement post-contrast in Tl-weighted images (C, D) suggestive of myocardial inflammation, with an early enhancement ratio of 5.7. Evidence of global myocardial edema (E, ratio 2.6).

Table | (abstract O4) Imaging parameters edematous myocardium (T1/T2 = 1119/101.8 ms) [2]. Phantom T| and T2 values were verified using a standard spin echo sequence. as and T2 values were verified using a standard spin echo sequence. In-vivo: T2 Maps were acquired in 9 healthy subjects to determine the normal range of T2 values. Three short-axis and two long-axis views were imaged during breath hold (duration ~5 HB) and in free breathing using navigator gating. In 5 subjects, four averages were acquired with navigator gating to test the benefits of increased SNR. Average T2 values were calculated in 16 myocardial segments using both methods and compared. Measurements were pooled to obtain global mean and standard deviation to investigate inter-subject and inter-segment variability.

Table 2 (abstract O4) Results from pig studies. The values are mean, SD of T2 (ms) in mid ventricular SAX slice.

T2 Weighted image (DB-STIR-TSE) (I) showing high signal intensity due to stagnant blood in the apical regions. The corresponding T2 Map clearly distinguishes myocardium from stagnant blood. Signal intensity variability: Signal in T2W DB-STIR-TSE showed high variability (36.7%) while T2 maps showed no such variation (3%) as seen in Figure 2.

T2 Maps from a pig showing enhanced T2 in anterior segment (arrow). In this pig, T2 in anterior segment was 83.5 + || ms vs. 58.6 + 4.5 ms in inferior segment.

T2W image (left) shows a lot of signal variability due to surface coil sensitivity variations and motion as compared to the corresponding T2 Map (right). Similar color map used in both images for comparison.

3 T Coronary MR image of a 50-year-old patient. Reformatted images (a, b) and volume rendering image demonstrates normal RCA, LM, LAD. Better visualization of the entire coronary artery tree after removing the background of myocardium, long segments of all major coronary arteries are well depicted and correlate well with X-ray angiography (d, e).

leading to such changes are unknown. We have examined the impact of sustained exposure to environmental hypobaric hypoxia, on cardiac function and energetics.

In vito DynaCT and '°F MR images of PFOB Caps phantoms. (A, B) DynaCT images of PFOM Caps phantoms demonstrated the ability to detect as few as 2 capsules. (C) '°F MRI (3D-TrueFISP, BW = 1500 Hz/ px, TR/TE = 3.0/1.5 ms, 2.0 x 2.0 x 5.0 mm?, 24 partitions, 4 avgs, 62 s acquisition) of the same phantom as B showed as few as 25 PFOB Caps were identifiable.

Figure | (abstract O8) ECG- and respiratory-gated T2-weighted cardiac MRI of mice post-Ml (A, 2-chamber view) and post-DOX (B, short axis view), 30 minutes after tail vein infection of 100 ul ANX-SPIO. Note focal T2 signal void (white arrows) of ANX-SPIO in antero-apex of post-MI heart, and diffuse septal, anterior and inferior T2 signal void in post-DOX heart (LA, left atrium; LY, left ventricle; RV, right ventricle; AV, anterior wall; IW, inferior wall). Conclusion: Cardiac MRI using ANX-SPIO can accurately detect myocardial apoptosis in vivo. Distinct MRI signal distributions were noted following ischemic (MI) versus oxidative (DOX) injury. This molecular imaging strategy may help identify ‘at risk’ cardiac cell populations.

Summary of the PS-DIR Coronary wall signed-image reconstruction algorithm. A ring-shaped region-of-interest is selected around the coronary wall (a). Phase points with high magnitude SNR are selected for region-growing (b). Map of local phase inhomogeneity is created using bi-linear interpolation (c). Inhomogeneity is removed from the phase image (d) and a final signed image is calculated (e).

DIR pulse sequence. TI* is the time when blood magnetization is nulled. In PS-DIR, imaging starts at Tl < TI*. Multiple PS-DIR black-blood slices can be acquired with CNR 2 CNR of DIR.

PS-DIR Coronary vessel wall imaging at different Tl < TI* in two separate subjects.

Left: CNR (mean + stdev) at different TI values in PS-DIR signed images (red) and in DIR magnitude images at TI* (green). Right: Wall thickness at different Tl using PS-DIR images (red) and using DIR images at TI*. Note the agreement with the measurements using TI*. Dual-slice single breath-hold PS-DIR Coronary vessel wall black-blook imaging Experiments: Anatomical slices perpendicular to the proximal part of the right coronary artery (RCA) at end-systole were planned similar to a previously published methodology [7]. First, serial single-slice multi-phase PS-DIR images were acquired with incremental Tl ranging from 50 ms—500 ms in |5 healthy adult

Results: Using Tl less than TI*, Fig. 3 shows that PS-DIR enables delineation of the coronary artery vessel wall and supports an increased wall-lumen contrast when compared with

(a) CT scan showing calcification at the edge of the popliteal artery just behind the knee. (b) Magniture gradient echo image showing the signal loss from the calcification of the same area. (c) Phase image showing the diamagnetic effect from the calcification. Note the simliar shape and extent of the calcification in both the CT and MR results. Inserts are zoomed images of the cross-section of the vessels in reformatted transverse images.

Table | (abstract O15) Multivariable regressions analysis of predictors of diastolic LV dysfunction

Kaplan-Meier curves illustrating. A) the prognostic implications of UMI alone and B) the complementary prediction of MACE by RevPD and UMI, in patients without any history of MI. and vasodilator (adenosine or dipyridamole) stress first-pass CMRMPI images were obtained and followed by LGE imaging. All CMRMPI images were interpreted for reversible perfusion defects (RevPD) using the 16-segment nomenclature and graded segmental LGE in a separate session. The readers were blinded to any clinical outcome in either session. At a median follow up of 15 months, |3 cardiac deaths and 26 nonfatal events occurred. RevPD was the strongest multivariable predictor to MACE, demonstrating a > 8-fold hazard increase to MACE (P < 0.0001) Methods and Results: We performed CMR on 254 patients referred with symptoms suspicious of myocardial ischemia. Rest Background: Recent studies have demonstrated the prognos- tic implication of CMR myocardial perfusion imaging (CMRMPI) in a clinical setting. Apart from detecting reversible perfusion defect from flow-limiting coronary stenosis, CMR late enhance- ment imaging (LGE) is currently the most sensitive method in detecting clinically unrecognized subendocardial infarction (UMI) from prior ischemic injury. We therefore tested the hypothesis that, characterization of these 2 processes from coronary artery disease (CAD) by CMR can provide complementary patient prognostic values.

Annual event rates of MACE and cardiac death by RevPD, UMI, and both RevPD and UMI.

*p < 0.04 vs. Z group. Table | (abstract O19) Changes in plaque and vessel wall volume over time

Left — Representative black blood image of SFA from R group patient at baseline. Right — same location imaged one year later. Notice plaque regression in the SFA.

Transmural lateral wall infarction (a) and Inferoapical infarction (b). Results: Three (8%) out of 39 patients scanned had to be excluded due to poor image quality. There were seven women (18%) and thirty two men (82%). Fifteen (45%) patients had grade IR cellular rejection rejection, and two (6%) had grade 2R

Frequency of any delayed enhancement by myocardial segment. Conclusion: DE is a common feature in the transplant population. Most DE observed is in a non-ischemic pattern;

Example of non-ischemic fibrosis. Subepicardial delayed enhance- ment of the anteroseptal and anterior walls (a) and inferior wall (b).

Multi-slice MDCT (top), MR delayed enhancement images (middle) and TTC slices (bottom) from a representative animal show good correspondence between modalities in defining patchy microinfarction embedded in viable myocardium (indicated by arrows).

An example patient showing regions of >50% scar tissue, regional mechanical activation delay, and regions of overlap. In this example, 32% of the region of delayed mechanical activation was also a region of >50% scar tissue. Endocardial and epicardial borders were traced on the short axis DCE images. A separate region of interest was drawn around the area of infarct. Each slice of the myocardium was resampled at 360 chords around the heart, and each chord was assigned a value between O and | depending on the scar burden at that

PM involvement in a subject with RCA territorial infarction. Pre-contrast SSFP image (A) depicted the morphology of PM. IR-FGRE image (B) did not show the PM-MI, while MCDE (C and D) clearly demonstrated the infarcted PM (arrows).

Table | (abstract O30) Results of multiple linear regression of left ventricular remodeling well as lower systolic wall thickening in the infarcted myocardium (all p-values < 0.001). Infarct size, size of area at risk and size of MVO were significantly larger in patients with hermorrhagic MI. At 4 months, a significant improvement in LV ejection fraction in patients with non-hemorrhagic MI was seen (baseline: 49.3 + 7.9% vs. 4 months: 52.9 + 8.1%; p < 0.01). LV ejection fraction did, however, not improve in patients with hemorrhagic MI (baseline: 42.8 + 6.5% vs. 4 months: 41.9 + 8.5%; p = 0.68). Multivariate analysis showed myocardial hemorrhage to be an independent predictor of adverse LV remodeling at 4 months (defined as an increase in LV end-systolic volume). This pattern was independent of initial infarct size (See Table 1).

Figure | (abstract O31) Colour coded SENC-image of a patient with transmural myocardial infarction with corresponding strain-curve. Purpose: To evaluate the predictive value of regional systolic and diastolic function for improvement of regional myocardial function in patients after AMI. Methods: MRI (1.5 T, Achieva, Philips, the Netherlands) was performed to 23 consecutive patients (mean age 57 + 10) 3 + | days after successfully reperfused ST-elevation-myocardial infarc- tion and at a follow-up of 6+2 month. 1I0 age-matched volunteers served as controls. True cine sequences of 3 long- axis views (2-,3- and 4-chamber) and a short-axis (SA) view covering the ventricle from apex to basis were acquired using a Steady State Free Precession (SSFP) sequence. After that, SENC cine images were acquired on the same long-axis planes to measure circumferential strains. Finally, using the same plane

Receiver-operating characteristic (ROC) curve demonstrates that diastolic strain rate assessed with Strain-Encoded Imaging is more sensitive than peak systolic strain for prediction of functional recovery (diastolic strain rate AUC 0.77 (0.67—0.86); peak systolic strain AUC 0.64 (0.53-0.74); p < 0.05).

(A) T2* map acquired 2 days post-PCl. Pixels with a T2* < 20 ms are shown in red and demonstrate the region of post-reperfusion hemorrhage (arrow). Susceptibility artifact (arrowhead). (B) The hemorrhage corresponds to the area of MVO (red contour) shown on the early enhancement image. (C) Myocardial necrosis (red contour), and residual MVO (black core), is shown on the late enhancement image.

T2-STIR images in the same patient as in Figure |. Boundary detection identifies the myocardial edema (green line) which represents the AAR. The AAR determined at each signal threshold is shown in red. At low thresholds non-specific signal noise results in bright pixels in non- ischemic territories causing an overestimation of the AAR. At higher thresholds the signal from the myocardial edema is masked by the presence of hemorrhage in the core of the infarct (arrow) and leads to an underestimation of the AAR. The area of myocardial edema on the T2 STIR images was measured with a boundary detection tool. This was compared to a conventional signal intensity threshold method using 2, 3 and 5 standard deviations (sd) above the mean of remote normal myocardium. A salvage index was calculated as the proportion of the AAR that did not show late enhancement. T2*-mapping of

Figure | (abstract 033) Gd-enhanced mid-ventricular image of the mouse heart | day after MI, showing enhanced infarct region (A) with corresponding thresholded infarct region (D). Same slice T2w spin echo image (B) with corresponding thresholded edema (E). Same slice T2w T2prep image (C) with corresponding thresholded edema (F). Comparison of detected infarct and edema region areas over Days | through 4 post-MI for each imaging method.

Table | (abstract O34) DCE-MRI sequence parameters

Representative dynamic frames from ten accelerated DCE-MRI datasets.

Examples of parametric maps of enhancement ratios in single-vessel disease (corresponding to images 2, 4 and 7 in Figure |). Dark coloured voxels belong to the lowest part of the individual study's frequency distribution.

Rest perfusion MRI acquired with 3 T MR imager, 32 channel cardiac coils and k-t SENSE in a patient with triple vessel disease. Rest perfusion MRI is normal and no endocardial banding artifact is observed.

Stress perfusion MRI in the same patient with triple vessel disease. Subendocardial ischemia is clearly demonstrated in the anteroseptal well, lateral wall and inferior wall on high resolution images.

Data are presented as means + standard deviation (95% confidence intervals) or median (interquartile range) as appropriate Rest MBF corrected: res MBF corrected for rate-pressure product (RPP), an index of myocardial oxygen consumption: MBF = (MBF/RPP) x 10*. Stenosed: myocardial segment subtended by a >50% stenosed coronary artery. Remote: myocardial segments subtended by arteries with minimal or no CAD. Normal: myocardia segments in normal volunteers. * p < 0.05 for comparison between stenosed and remote segments. * p < 0.05 for comparison between remote t ischemia and normal segments. * p < 0.05 for comparison between stenosed and normal segments. Table | (abstract O37) Rest MBF, stress MBF, coronary flow reserve (CFR) and BOLD-signal intensity (SI) change of stenosed, remote to ischemia and normal segments

An example of a patient with significant disease in the right coronary artery. A SI drop was noted in the inferior wall (black arrow). All other myocardial segments showed a rise in SI during stress.

Table | (abstract O39) Semiquantitative parameters in native vessels and grafts

contrast kinetics in areas supplied by native coronaries and different bypass grafts.

Percent changes in peak diastolic flow velocity during cold pressor test. *p < 0.01 compared to baseline and 10 minutes recovery. Methods: Healthy, pre-menopausal women were recruited for this study and provided informed consent approved by the institutional review board. Subjects were placed supine in a 3 T MRI scanner (Achieva, Philips, Best, NL) using a 6 element cardiac receive coil. Scout scans were performed to determine the imaging plane orthogonal to the proximal right coronary

Cross-sectional images of the RCA at rest in a subject using velocity- encoded MRI. (A) Phase image and accompanying (B) Magnitude image. invasively assess coronary vasoreactivity would be especially useful in the early diagnosis and management of women with coronary artery disease.

regional (aortic distensibility) aortic stiffness. Assessment can be combined with precise measurement of carotid atheroma burden, a marker of early sub-clinical atherosclerosis. We investigated whether aortic stiffness in early adult life is already associated with early changes in carotid structure.

Conclusion: |n established carotid artery disease, regions with high WSS are associated with increased necrotic core, but not

Figure | (abstract 045) the manually identified mask (Panel F, G). ROC curve analysis showed that PCA (AUC: 0.96) was significantly better than the single best approach (T2 immediately post-SPIO; AUC: 0.90) (p = 0.016) (Panel H).

Figure 4 (abstract 046) (a) ECG outside the magnet; (b) ECG inside the magnet.

segmented k-space gradient echo sequence was used for scout scanning. Multi-slice cine scans were used for coronary artery localization and for the visual identification of the time period (Td) of minimal coronary motion. Scan plane localization parallel to the right coronary artery (RCA) was facilitated using a three- point planscan tool. Double-oblique free-breathing 3D coronary MRA (segmented k-space gradient-echo imaging, TR = 4 ms, TE = 1.5 ms, RF excitation angle = 15°, field-of-view = 320 x 291 mm?, scan matrix = 392 x 373, |5 slices, slice thickness = 2 mm, acquisition window ~100 ms, scan time ~5 min) was performed using prospective navigator gating with the 2D selective navigator localized at the heart-lung interface. Image data were collected in mid-diastole at the predetermined Td. An adiabatic spectrally selective inversion recovery pre-pulse (Tl = 200 ms) was used for fat suppression and enhanced contrast between the coronary blood-pool and epicardial fat. Coronary MRAs were reformatted and length measurements were performed using the “Soapbubble” software tool.

Table | (abstract O47) Comparison between EPI-FLASH and TrueFISP with identical resolution and imaging time Volunteer imaging: 7 volunteers were scanned on a 1.5 T Espree scanner (Siemens Medical Solutions). Scan parameters were: TR = 11.3, TE = 4.27, flip angle = 25, 66 lines per heartbeat in a window of 124 ms, acquired k-space lines = 132, readout bandwidth = 977 Hz/pixel, Tl = 300 ms, matrix: 256 x 189 x 60, interpolated voxel size: 0.5 x 0.55 x | mm?. 0.2 mmol/kg body weight of Gd-DTPA was injected at 0.5 cc/sec [5]. The total imaging time for the whole-heart scan was

and flip angle, using simulations of the Bloch equations and phantom studies.

Coronary artery images using the EPI-FLASH sequence and a TrueFISP sequence with longer imaging time.

Coronary artery images using the EPI-FLASH and TrueFISP sequences with identical imaging times.

Figure | (abstract 048) Kaplan-Meier survival estimates by RVEF.

Arrows identify patterns of MDE. a) transmural; b) subendocardial; c) EFE; d) subepicardial/intramural. Results: Of the 85 subjects included (65% male; median age at Fontan 4.5 [2.0, | 1.2] years; mean age at CMR 23.1 + | 1.2 years), 21 (25%) had positive MDE in the ventricular myocardium. MDE was seen in the following locations: dominant ventricle free wall (n = 13, 62%), secondary ventricle free wall (n = 9, 43%), septal insertion (n = 5, 24%), ventricular septum (n = 3, 14%), apex (n = 2, 10%), previous surgical sites (n = 2, 10%), and papillary muscle (n = 2, 10%). MDE was seen in the following patterns: transmural lesion

Table | (abstract O50) Univariate analysis of MDE location and pattern (n = 9, 43%), subepicardial/intramural (n = 5, 24%), subendocardial (n = 5, 24%), circumferential endocardial fibroelastosis (n = 4, 19%), and speckled (n = 2, 10%). Results of univariate analysis comparing patients with and without MDE, stratified by location and pattern are summarized in Table |. Multivariate linear regression analysis demonstrated that higher MDE % (slope: —1.7, Cl: —2.5 to —1.0, p < 0.0001) and age at CMR (slope: —0.6, Cl: —1.0 to—0.3, p = 0.002) were independently and inversely associated with EF (R* = 0.59).

The T2 maps of a normal (a) and a DMD subject (b) showing the left ventricle in the short axis view. Values are mean + SD, median [25%, 75%], or n (%); ‘Student t test, *Mann-Whitney U test, *Fisher exact test, *Chi-squared test of independence.

Association between FWWHM/T2mean and €,, in group A and B. Pearson correlation coefficient r = .51.

Figure 2 (abstract O52) Healthy subject pulmonary flow.

of 2D phase contrast flow planes difficult, time consuming, and labor intensive. As a result, studies may be completed without collection of the data necessary to determine flow distribution. 4D flow techniques permit the collection of temporally-resolved 3D data sets of the central systemic and pulmonary vasculature in a single acquisition. Complete data acquisition is guaranteed and appropriate planes for flow quantification can be chosen during post processing.

cardiac cycle, the in-plane phase encode value was held constant while 4 slice-encoding phase encodes are acquired to encode all flow directions. Parallel imaging (GRAPPA) with an acceleration factor of 2 was used. Scan times ranged from 12-16 minutes depending on heart rate. Retrospective EKG gating was used to resolve 20 time frames through the cardiac cycle yielding a temporal resolution of 50-80 msec. Respiratory compensation was employed. The raw data was reconfigured for EnSight visualization (CEI Inc., Apex, NC). Navigation within the 3D data set allows retrospective placement of planes perpendicular to the vessel of interest. 10 subjects with TOF were prospectively enrolled. For each subject, a 4D flow data set was obtained and an attempt was made to obtain 2D flow in 4 locations; the ascending aorta (AsAo), main pulmonary artery (MPA), and branch PA’s (RPA and LPA).

Figure | (abstract O53) Normalized flow vs imaging angle.

Methods: Fifty elderly subjects [mean age = 65 (+ 8), 16/ 50 = 32% women] with at least one major cardiovascular risk factor (smoking, diabetes, hypertension, hyperlipidemia) under- went MRI (1.5 T Siemens Sonata) for abdominal and peri-aortic adipose tissue quantification and aortic atherosclerosis asses-

Proton density weighted fat saturation black blood turbo spin echo (TSE) cross-sectional images covering the descending thoracic aorta (figure la) were used for atheroma burden

ment. A Water Suppression (WS) TI weighted (TIVV) Turbo Spin Echo (TSE) multi-slice sequence was used for visceral adipose tissue (VAT) and abdominal subcutaneous adipose tissue (SCAT) measurement.

Methods: We studied 128 patients (48% women) aged 55-85 years in which aortic distensibility was determined during intravenous dobutamine infused to achieve 85% of predicted

*p = 0.06 vs. baseline. Table | (abstract O61) Changes in perfusion, metabolism, and exercise parameters over time Methods: 61 patients with mild-to-moderate symptomatic PAD (mean age 63 +10 years, ankle brachial index 0.69 + 0.15) were studied before and | year after starting one of 3 lipid lowering therapies. Statin-naive patients were randomized to simvastatin 40 mg or simvastatin 40 mg plus ezetimibe 10 mg (n = 31) and patients already on a statin were given open-label ezetimbibe 10 mg (n = 30). Lipid measurements were obtained as part of the VAP test. Patients with interval stenting of the leg studied (n = 4) or bypass surgery (n = |) were excluded from analysis. All 56 remaining patients had calf muscle phosphocreatine recovery time constant (PCr) measured using 3'phosphorus (P) MRS immediately after symptom-limited calf muscle exercise using a MR compatible ergometer on a Siemens Sonata 1.5 T scanner. Exercise time was recorded. ?!P MRS was obtained using a single-pulse, surface coil localized, 512 ms free induction decay acquisition with 20 averages centered on the mid-calf. PCr was then calculated using a monoexponential fit of phosphocreatine concentration versus time, beginning at cessa- tion of exercise. Calf muscle tissue perfusion (TP) was measured in 50 patients using first-pass contrast-enhanced MRI at peak exercise. The remaining 6 patients were excluded due to compromised renal function. The patients pushed a MR- compatible foot pedal ergometer at a steady rate (10-12 bpm) until limiting symptoms or exhaustion while in a Siemens Avanto 1.5 T scanner and gadolinium (0.1 mM/kg) was infused at peak exercise. Work expended during exercise was recorded. Time intensity curves were generated with ARGUS image analysis software from the region of calf muscle with the greatest intensity post contrast and the slope of this curve was defined as TP. To assess microvascular blood flow within the calf muscle, TP

Change in Ascending Aortic Distensibility (10-3 mmbhg-1) From Low Dose to Peak Dose

Table | (abstract O63) Logistic univariate regression analysis testing the association between endomyocardial biopsy proven cardiac amyloidosis and various noninvasive imaging parameters Methods: We retrospectively evaluated 55 consecutive patients with known sarcoidosis who were referred for CMR. Imaging was performed ona 1.5 Tesla MRI scanner with a flexible surface coil. Retrospectively gated cines of the left ventricular

appearance, E/A ratio, E/E’ ratio, stage of diastology, deceleration time (msec) and myocardial performance index [(isovolumic contraction time + isovolumic relaxation time)/ejection time]. DHE-MR images were obtained in standard long and short axis orientations (covering the entire LV), after injection of Gadolinium dimenglumine using an inversion recovery spoiled gradient echo sequence: TE 4 msec, TR 8 msec, flip angle 30°, bandwidth 140 Hz/pixel, 23 k-space lines acquired every other RR-interval, field of view (varied from 228-330 in the x-direction and 260-330 in the y-direction) and matrix size (varied from 140-180 in the x-direction and 256 in the y-direction). CMR was considered positive in the presence of DHE of entire sub- endocardium with extension into the neighboring myocardium.

(LY) 2-, 3-, and 4-chambers, and a short-axis stack were obtained using steady state free precession imaging (TR 2.9 ms, TE |.5 ms, flip angle 60°, temporal resolution 25—40 ms). LGE images of the same views were obtained 10-20 minutes after infusion of Gd- DTPA (0.15-0.2 mmol/kg) using a Tl-weighted inversion recovery GRE pulse sequence (TI based on optimal myocardial nulling, TR 3.9 ms, TE |.7 ms, flip angle = 15—-30°). The cines of the short-axis stack were used to determine LV and right ventricular end-diastolic volume (LVEDV and RVEDY), end- systolic volume (LVESV and RVESV), ejection fraction (LVEF and RVEF), and LV mass. The presence or absence of LGE was determined for each of the 17 segments of the American Heart Association left ventricular model. Continuous variables were reported as mean + standard deviation, groups (based on the presence or absence of LGE) were compared using a t-test, and linear regression analysis was used to investigate the correlation between LGE and LV size and function. A p value < 0.05 was considered statistically significant.

Conclusion: In patients with known sarcoidosis, nearly a quarter had evidence of cardiac involvement, as defined by the presence of LGE, despite the lack of any difference in traditional measures of cardiac structure and function. These preliminary findings suggest that volumetric and functional assessments alone are inadequate for the detection of cardiac sarcoidosis and that tissue characterization using LGE is essential. Further studies are needed to determine the long-term prognostic value of LGE and to further clarify the relationship between the extent of LGE and left ventricular morphology and physiology in patients with cardiac sarcoidosis.

Chi-square for multivariate model = 12.27, p-value = 0.007. Table | (abstract O65) Cox proportional hazard analysis of various clinical and noninvasive imaging predictors of long-term mortality in patients with biopsy proven cardiac amyloidosis

Fused DE-MRI and endocardial maps. The endocardial map is projected on the endocardial surface as determined by MRI. Areas with low voltages are illustrated in red, and areas of normal voltages are in purple. The 3-D distribution of DE on MRI (including both surface and transmural extent) are represented in gray. There is excellent correlation between the distribution of scar as represented by DE on MRI, and the areas of low voltage on the endocardial map.

T|-weighted late gadolinium enhancement images. Left panels: Evidence of subepicardial non-ischemic myocardial fibrosis in athletes | and 2 (arrows), with none in healthy control. Right panels: Quantification of myocardial fibrosis by manually tracing a region of interest in health remote myocardium (blue outline). Automated computer thresholding set at 5 standard deviations above the mean signal intensity of remote myocardium detects fibrosis (red overlay). The detected extent of fibrosis for athlete | is 14.12%, 11.30% for athlete 2 and 0.20% for control. Figure | (abstract O68)

Table | (abstract O69) Clinical and Imaging Characteristic in 6 Patients (CMR) can differentiate myocarditis from acute ischemic myocardial injury and provide a basis for adoption of an appropriate treatment strategy in such patients. imaging tests in genotype negative family members in families identified to have a disease causing mutation, present approxi- mately 50% of the time. Because large, long-term trials to compare these methods are not feasible, modeling of costs and outcomes can provide insight into the potential costs and benefits of various screening strategies.

Left panel: Through-plane velocity-encoding magentic resonance imaging ascending (red circle) and proximal descending aorta (blue circle). Middle panel: Corresponding flow measurement at ascending (red line) and proximal descending aorta (blue line). Right panel: The measurement of arrival time at ascending and proximal descending aorta. The pulse wave velocity was calculated as the aortic path length between these 2 sites divided by the time delay between the arrival of the foot pulse wave at these 2 sites. Introduction: Beyond the morphologic and functional abnormalities of the bicuspid aortic valve (BAV) there is also intrinsic pathology of the aortic wall, manifested by potentially lethal complications such as aortic aneurysm or dissection. Aortic distensibility and compliance are impaired in athero- sclerotic aortic aneurysms and Marfan syndrome. Similar abnormalities of compliance are felt to occur in the setting of BAV, though this has been little studied with velocity-encoded magnetic resonance imaging (VENC-MRI), and further there is no data on the influence of BAV morphology on this abnormality. VENC-MRI is a potent non-invasive technique to determine aortic distensibility via aortic pulse wave velocity (PWV) measurements; in addition these measurements do not depend on knowledge of central arterial pressure or geometrical assumptions that may limit alternative measurement tools.

The pulse wave velocity (PWY) in normal patients, right and left (R-L) fusion, and right and non-coronary (R-NC) bicuspid aortic valve patients. Circal, mean and whiskers. 95% confidence intervals.

Conclusion: We have shown a short term, high fat diet raised plasma FFA concentrations, impaired myocardial energetics without effect on systolic or diastolic function. This suggests that high plasma FFAs may be detrimental for heart in normal subjects and shows a potential mechanism of impairment in heart failure patients.

Figure | (abstract O74) Cardiovascular magnetic resonance imaging for the assessment of cardiac inflammation and injury following prolonged exercise set a ae ar nt es ee ees te Lo, een

antibody (n= 7) or nonspecific IgG (nlgG) (n = 5) with 0.075 mmol Gd/kg, followed by excision of the aorta for frozen cross-sections. The fluorescence image used to indicate whether the probe bound to the plaque was examined under fluorescence microscopy.

Table | (abstract O77) Heart rate and volumes and function Table 2 (abstract O77) Hemodynamics Introduction: Cardiac output is dependent, in part, on the ability of LV to accept preload at low filling pressure. Systematic modulation of preload is thus an important capability for the study of the preload dependence of any given aspect of cardiac performance in health and disease. Previously, LV (un)loading has been studied by control of lower body pressure, used to

Table 3 (abstract O77) Tissue mechanics Conclusion: We have shown that a simple MRI-compatible lower body pressure chamber can significantly unload the LV and that a comprehensive systolic and diastolic function study is feasible during this unloading (45 minute study duration for both atmospheric and -30 mmHg unloading). Our changes in EDV and

Figure | (abstract O78) Volunteer in the lower body pressure chamber.

T2*-weighted in vivo MRI for different allografts after USPIO administra- tion. Areas with UPSIO-labeled macrophage infiltration show signal loss in both LV and RV. Figure | (abstract O79)

Figure 2 (abstract O79) (A) T2*-weighted MRI for an allograft after USPIO administration. (B) Pseudo-coloring of different contrast-to-noise ratios (CNR) for easier visualization. (C, D) Tagged MRI of the same allograft heart at ED (C) and ES (D). The block arrows point to regions with compromised contractile functions. (E) Colored Ecc strain map of the allograft. (F) Ecc stran values of 48 probe-points throughout LV.

(A) Ecc strain values of 48 probe-points throughout LV for an isograft (triangle) and 2 allografts with Grade II (square) and Grade IV (circle) rejection. (B Status of probe-points for 3 allografts with different rejection grades. Compromised probe-points are black, whereas the normal probe-points are lef blank. (C) Degrees of compromised probe-points for different rejection grades.

Results from three patients show fat deposition in areas of late gadolinium-enhancement (LGE).

Figure | (abstract O81) Magnitude images acquired using (a) conventional segmented, (b) real- time GRE-EP| SVE PC-MRI, and phase velocity maps from (c) conventional segmented PC-MRI, and (d) real-time GRE-EPI SVE PC- MRI sequence in a slice cutting the ascending and descending aorta close to the aortic arch.

Table | (abstract O81) Peak velocities and RMSE) of the real-time SVE technique with respect to conventional PC-MR 160 x 120, flip angle = 25°, spatial resolution = 2.18 x 2.18 mm?, GRAPPA acceleration rate = 2, and Venc = 150 cm/s. The TE/TR/temporal resolution of the conventional gradient-echo and real-time GRE-EPI sequences were 3.5/7.0/42.0 ms and 2.9/14.6/58.4 ms, respectively.

Figure | (abstract O82) 2D late diastole cine SSFP BOLD images of a dog with severe LCX stenosis and adenosine infusion obtained at different TR and FA: TR/FA = 3.5 ms, 30° (A); 3.5 ms, 70° (B); 6.0 ms, 30° (C); and 6.0 ms, 70° (D). LCX territories are the shorter arcs subtended by arrows.

SSFP-baseed Myocardial BOLD contrast obtained from MCS and experimental studies with the same parameters (TR = 3.5 ms, 4.7 ms, 6.0 ms; Flip angle = 30°, 50°, 70°). For a given FA and TR, MCS oxygen contrasts were computed assuming myocardial oxygenation in the territory supplied by a stenotic vessl is 20%, while that supplied by a healthy vessel is 80% under pharmacological stress, assuming myocardial blood volume fraction of 9%.

3D representation of LE MR images of acute (A-B) and chronic (C-D) RFA lesions. Overlaid in blue are the ablation points acquired with NavX. It can be seen in the chronic image that an area of enhancement is no longer apparent around the right lower PV.

In-vivo temperature increase (a) with the MR-EP catheter and (b) with the conventional catheter.

Figure 2 (abstract O84) EP signals acquired in the RV (a) with the MR-EP catheter equipped with highly resistive wires and (b) with the conventional catheter.

Top: Roadmap-based real-time 3D-visualization of the catheter position during recording (red dot) on the MR-EP workstation. The yellow dots in the 3D rendering of the heart indicate previous mapping positions. Bottom: In-bore EP recordings at two selected positions showing an atrial signal (left) and a ventricular signal (right).

Blood flow changes and BOLD-SI in canine model under adenosis infusion.

A) Real-time MR-temperature map during HIFU septal ablation (red zone on image), B) T2 MR image of HIFU lesions, C) Gross pathology of septal lesions from (B) using multiple 20s pulses, D) Lesions, each 4 mm x 4 mm created with single 20s HIFU pulses along the LV lateral wall. Background: Invasive catheter-based myocardial ablation has become an important treatment of hypertrophic cardiomyopathy (HCM) and cardiac arrhythmias, but has known complications as well as the inability to actively visualize and control the extent of ablated tissue. High-intensity focused ultrasound (HIFU) can noninvasively create focal ablation lesions and has been

Lesion size (mm2) as a function of pulse duration. Pulses were cycled on and off with a signal generator at | Hz.

Multislice MR images showing the LV on cine (top) and the LAD perfusion territory on corresponding FPP (middle) and early CE (bottom) MRI as hyperenhanced regions.

Figure | (abstract O90) CMR DENSE dyssynchrony measures.

Measuring deformation of the LV and blood velocity during early filling.

(a) The 3-dimensional left ventricular volume is superimposed on the 4-chamber long-axis MRI. Two of the short-axis images are also displayed to demonstrate the spatial arrangement of the images. The left ventricle was divided into 16 wedge-shaped regional volumes as shown. (b) Interal Flow Fraction (IFF) discriminates between patients and healthy controls with 95% accuracy.

Figure 2 (abstract O92) Internal flow is significanly increased thoughout the cardiac cycle in patients with dyssycnhronous heart failure compared to age and sex- matched healthy controls. Note the peaks in internal flow surrounding the periods of isovolumic contraction (time 0) and relaxation (vertical dashed line marking the end of systole).

Net flow volumes for mitral valve (MV) vs. aortic valve (AV), tricuspid valve (TV) vs. MV and TV vs. pulmonary valve (PV) in 23 patients with valv regurgitation. 3D 3-directional velocity-encoded MRI is performed at the basal level of the heart. In offline analysis, retrospective valve tracking for each of the heart valves can be performed, resulting in flow at the mitral valve (MV), tricuspid valve (TV), aortic valve (AV) and pulmonary valve (PV). Introduction: In regurgitant heart valves, surgical decision- making is based on the severity of the regurgitation through the particular valve. Conventional two-dimensional (2D) one-direc- tional velocity-encoded (VE) MRI is routinely used for flow assessment, but this technique has been shown to be inaccurate and correlation between the net flow volumes through the valves is weak, even in the absence of regurgitation. 2D one-directional VE MRI is limited because the position and angulation of the acquisition plane cannot be adapted to the valve motion and the direction of the inflow and regurgitant jet.

Table | (abstract O93) Statistical results

Radial (a-c), circumferential (d-f), and longitudinal (g-i) strains computed throughout the LV from 3D cine DENSE MRI of a normal volunteer. Data are shown at end diastole (a, d, g), mid systole (b, e, h), and end systole (c, f, i). Three of 22 cardiac phases are displayed. Individual dots represent displacement, while color represents strain.

Normalized diastolic parameters (heart failure patient).

Table 2 (abstract O95) Conventional diastolic parameters Table 3 (abstract O95) Novel diastolic parameters

Table | (abstract O95) Heart rate, volumes and function

References A wave filling velocities (cm/s), mitral annular velocity (E’ in cm/s), E/ A ratio, E/E’ ratio and inflow propagation velocity (V, in cm/s). Additional parameters include the intraventricular (IVPG) and atrial (IAPG) pressure gradients (derived from in-plane blood velocities), peak torsion (deg) and rate of untwisting (deg/sec), peak diastolic radial velocity (ventricular average — cm/s), and peak diastolic circumferential strain rate (ventricular average, s'). All tagged images were analyzed using a user-independent morphing approach. All studies were breath held with ECG gating (Siemens Sonata |.5 T scanner).

Table | (abstract O96) Largest ROI mean offset (cm/s) within 50 mm of Isocentre, for aortic slice, MPA slice (HF phase- encoding) and MPA slice (LR phase-encoding). The four rows per slice are from the four sites using each scanner type. The column order is not specified, i.e. scanner types are not identified. (< + 0.3 cm/s stdev total error as described above)

The planes used for veolicity offset acquisitions at all sites. The aortic (AO) plane at 45 degress between transaxial and sagittal used antero- posterior phase-encode. The main pulmonary artery (MPA) plane at 45 degrees between transaxial and coronal was repeated with head-foot and left-right phase-encoding. The grey bloc represents the gelatin phantom.

Short axis vs 3DR (ml). Purpose: Because of its complex morphology, accurate and reliable quantification of right ventricular (RV) volume and function using MRI is challenging. This study had two aims: (1) to

Figure | (abstract O97) heart disease. Indexing of ventricular volumes to body weight instead of body surface area elimates the gender difference in children and should therefore be preferred.

Materials and methods: In order to assess the transferability of the multislice multiecho T2* technique, heart multislice multiecho T2* sequence was installed on |.5 T MRI scanners (GE Healthcare) at six different sites. Five healthy subjects at each site (n = 30) were scanned to verify the homogeneity of normal ranges (T2* lover limit of normal 20 ms). Then, five TM patients were scanned at the reference site and were rescanned locally

Excellent image quality achieved (grade 4 or 5). Results: On average, 2.8 minutes (5% of total scan time) were spent to eliminate off-resonance banding artifacts in 3 T. This time is made up by more aggressive accelerated parallel imaging technique. As a result, average total scan time using 3 T was not different from 1.5 T (54 + 14 vs. 54 + 12 minutes, P = 0.47). No patients failed to complete the study due to SAR limit. There were no complications during any of the I|.5 T or 3 T CMR studies. A significantly higher proportion of perfusion images were graded as being of excellent quality on 3 Twhen compared to 1.5 T (82.4% vs. 41.4%, p < 0.0001) (Figure I). A significantly higher number of perfusion images also had minimal or no artifact on 3 T when compared to 1.5 T (93.7% vs. 72.4%, p = 0.0016). When LGE images were analyzed, a significantly higher proportion of images on 3 T were graded as being excellent (82.6% vs. 46.2%, p < 0.0001) and the proportion of LGE images having minimal or no artifact was also significantly higher on 3 T (83.0% vs. 56.4%, p = 0.0042). The number of Cine SSFP, pulmonary vein MRA, and phase contrast images that were

End-diastolic (ED) frame of the left and right ventricle (mid-ventricular slice in short-axis orientation); left: multi-breath-hold, standard SSFP, right: free breathing prospective self-gating.

Table | (abstract P3) Stroke volume (mean and stdev) of the different acquisitions for all measured vessel Background: 4D-flow has been introduced as a means of acquiring anatomical and three-directional velocity information for all pixels within a 3D volume over different time points. The acquisition time of such data sets is long and respiratory compensation is thus required. However, navigator beams could disturb the steady state and are time consuming, which can be

circumvented by using a self-navigation approach. Here, we present a new technique for the acquisition of 4D-flow data of an isotropic saggital volume using a real time self gating technique. The data can thereafter be reformatted in any clinical view allowing the quantification of flow in different vessels with arbitrary orientations. This is important in congenital heart (CH) patients where scan planning can prolong the overall scan-time. Methods: Self navigation: A 3D Phase Contrast (PC) retro- spective cardiac trigger sequence was used to acquire 4D flow data. The sequence was modified to enable the acquisition of an extra ko profile at certain time intervals. These profiles were used to derive the breathing motion and to respiratory gate the acquisition in real-time. All the modifications were integrated

Figure | (abstract P5) Examples of pre- and post-lsordil images from four different subjects using combinations of two different sequences (GRE and SSFP) and two different Isordil doses (2.5 mg and 5 mg) at 10-15 minutes after lsordil.

Regional distribution of MDE and corresponding Ae,,.

Figure | (abstract P14) Representative circumferential myocardial strain in cardiomyo- pathy patients with LBBB, non-LBBB, and healthy controls.

LBBB patients have more mechanical dyssynchrony than non- LBBB patients.

Figure 2 (abstract P15) Peak oxygen consumption falls with extent of LGE in patients with HCM.

The extent of late gadolinium enhancement assigned to 3 groups.

whereof in 2 patients no delayed enhancement or intramyocar- dial T! signal suggesting fat deposits was present. Figure |. Conclusion: Patients with proven Brugada syndrome have larger right ventricular volumes, impaired RV function and dilated RVOT compared to matched controls. Right ventricular wall motion abnormalities, localized fat deposits and delayed enhancement can be found in some of the Brugada syndrome patients indicating subtle structural heart disease.

Table | (abstract P24) Acquisition parameters for Tlw-SPACE and ce-MRA and bulk head motion during the long 3D acquisition time. Respiratory movement artefact is controversial [4], but Boussel [5] demonstrated its detrimental effect on carotid wall imaging, using real-time transaxial cines. However, during quiet supine respiration, breathing is predominantly diaphragmatic resulting in the greatest carotid movement in the head-foot direction. We used a novel high temporal resolution interleaved approach to study carotid artery movement in all directions, over the typical 3D scan duration, for a true representation of the potential problem for 3D imaging. Siemens, Erlangen) using two body matrix coils for the abdominal/pelvic area and a peripheral coil for thighs and lower legs. The imaging protocol included localizers, coronal acquisition of abdominal aorta and SFA using TIlw-SPACE, followed by ce-MRA covering the abdomen, thighs and the lower legs.

(a) Regions of interest are drawn around the common carotid artery in the transverse plane and (b) in the oblique-sagittal plane allowing for ROI automatic tracking using MATLAB after image acquisition. Methods: Nineteen healthy volunteers were scanned at |.5 T (Siemens Avanto) with phased array carotid coils (Machnet). They had high-resolution bSSFP scans of the right carotid artery in the oblique-sagittal plane centered about the bifurcation and in the transverse plane to measure movement in the Head-Foot, Antero-Posterior and Left-Right directions. Each volunteer was scanned twice: with a vac-lok™ fixation pillow (CIVCO Medical Solutions) followed by a standard Siemens foam head support. Subjects were all asked “to lie still and relax” during scans. For 500 cardiac cycles, ECG-triggered diastolic oblique-sagittal and transverse single-shot bSSFP images were alternately acquired (one image/cardiac cycle) with pixel size 0.6 x 0.6 x 6mm, ETL |, flip angle 70°, TR 395 ms, TE 1.6 ms. Image post-processing using 2D-correlation tracked an ROI with MATLAB® (The MathWorks) (figure 1). For this, image pixels were 2D-interpolated to 0.31 x 0.31 mm.

(a) Demonstrates the “drift” phenomenon in all three perpendicular directions with both types of head support after smoothing. The “drift” appears to increase as the scan progresses but then seems to decrease towards the end. This could be due to the individual falling asleep during the scan but becoming more alert towards the end. Whereas, in (b), there is a gradual “drift” away from baseline until the end of the scan. In this example, the phenomenon may be explained by increased muscle tension at the start; individuals are generally tense and anxious prior to the scan but once the scan has commenced, they start to relax and so muscle tension decreases resulting in the observed “drift”.

Table | (abstract P26) DCMR imaging in small groups of patients with CHD revealed ventricular and vascular dysfunction that is not apparent at rest. There are no reports in a large series of patients on the safety and accuracy of low-dose DCMR imaging. Results: Obese subjects and normal weight controls were well matched for age, height, systolic blood pressure, fasting plasma glucose and total serum cholesterol. concentration compared to controls (p < 0.005). Obesity was associated with a 14% increase in PWV when compared to the normal weight controls (p = 0.021) and furthermore was associated with significantly higher leptin (p < 0.001) and C-reactive protein levels (p < 0.01), factors known to affect aortic mechanical function. After a one year period of weight loss (Average 21 kg, 50% of excess weight loss) there was a significant 14% decrease in aortic pulse wave velocity when measured between the ascending and abdominal aorta (p = 0.03). In contrast, while showing a trend, aortic pulse wave velocity improvements in proximal sections did not reach statistical significance. Both leptin and C-reactive protein levels were also reduced with weight loss. See Table |.

Table | (abstract P29) Post-MI day 28 LV parameters

Day 28 end systolic black blood images of C21 (A) and control (C) mice, with corresponding Day | DHE images shown in B and D.

LA: left atrium; A: aorta. stenosis, which is mostly diagnosed several weeks after intervention. The cardiac magnetic resonance tomography (CMR) allows analysis of heart function and pulmonary veins after ablation therapy.

(A) Fourier reconstruction of free-breathing MRI. (B) Model-based reconstruction of the same free-breathing MRI. (C) ECG-and-respiration- gated cine imaging for comparison. (D) Enlarged view of the heart showing the infiltrated MPIO-labeled cells.

A representative set of end-diastolic 3D SSFP BOLD images (4 con- tiguous slices from apex to base (obtained from a dog under baseline condition (upper row) and with moderate stenosis (lower row) under adenomise stress is shown. Note the signal enhancement observed in the non-LCX supplying territories (arrows) uder stress, while no signal enhancement is observed within the LCX supplying region in the presence of adenosine.

developed for analysis of €.. from non-tagged standard steady- state free precession (SSFP) cine CMR images.

MRI derived surface mesh depicting the LV endocardial (color coded), RV endocardial (in yellow), aorta (in red) and RCA (in red) anatomy. Coloring of the LV endocardial surface is used to indicate the transmural extent of scar in the endocardial layer of the myocardium (green = 0%, red = 100%). The square dots indicate the mapping locations obtained during the VT ablation procedure. Color coding is used to indicate the bipolar electrogram amplitude (white S$ 0 mV, dark blue 2 1.5 mV).

Late enhancement MRI of the same patient showing transmural scar in the septal-apical region with results of EAM superimposed. Bipolar voltages are low (<I.5 mV) in the scar region and normal (>1.5 mV) in the healthy myocardium.

Figure | (abstract P37) Correlation between Ve-CMR and Doppler echo derived values was tested using Spearman's rank correlation coefficients. There was significant correlation between left atrial size and E/A ratio. A: LA size (R = 0.9, p-value < 0.05), B: E/A ratio (R = 0.88, p-value < 0.05), but not C: (R = 0.12, p = 0.58). LA; left atrium, PV, pulmonary vein. Results: The mean age of the patients was 51 years. The average Echo mitral inflow absolute velocities were close to double the Ve-CMR {Echo E-velocity = 85.8 cm/sec vs. Ve-CMR E velocity = 41.4 cm/sec, Echo A-velocity 70 cm/sec vs. Ve-CMR

(A): The MRA of the right coronary artery (RCA) is used to select a cross-sectional plane (indicated by white line). Images of the proximal RCA (white arrow) in end-systole (B) and at diastole (C) in a patient with coronary artery disease (CAD). (D): Distensibility measurements in healthy subjects (N = 19) and patients with CAD (N = 11).

Figure 2 (abstract P39) spiral 3 T MRI technique, which could potentially enhance vessel wall delineation and accelerate scan time.

SEG-GRE magnitude image at peak mean velocity (a) and corresponding PC image (b). ISPLASH magnitude image at peak mean velocity (c) and corresponding PC image.

Mean velocity measured from SEG-GRE and the averaged mean velocity, 4 respiratory cycles, from ISPLASH.

Table | (abstract P40) The peak mean velocity measured from SEG-GRE averaged ISPLASH (4 respiratory cycles) and the full ISPLASH

Figure 3 (abstract P40) Full ‘real-time’ plot of normalised mean velocty with the respiratory (Resp) and the LED square wave used for breathing regulatior (Ext).

MRI-derived lung water (mid-sagittal right lung) for a healthy subject (total average lung water 21.7%) and a heart failure patient (37.2%).

MRI-derived lung water as a function of LVEDP in heart failure

Experimental setup of isolated heart with the aorta cannulated. Left: perfused with STH, right: replaced with PFC emulsion. Coronary arteries shown are filled with the emulsion.

Manual selection of ROls. (diffusion weighted: left to right). Right: free wall |, bottom: free wall 2, and middle: papillary muscle.

representative logarithmic normalized signal attenuation of STH and PFC. Each point corresponds to the mean and SD of selected ROls.

Figure 4 (abstract P42) Logarithmic normalized signal attenuation with changes in perfusate flow. Free wall | (left panel) is parallel to the diffusion encoded direction. On thé other hand, Free wall 2 (middle panel) and Papillary muscle (right panel) is oriented orthogonal to the diffusion encoding direction.

53 year old female patient with diabetes and hypertension, presenting to the emergency room with chest pain. (A) Diffuse subendocardial perfusion defect on adenosine stress MRI which was reversible at rest (B). The nuclear cardiac SPECT exam was normal, and there was no myocardial scar on delayed enhancement MRI. Methods: 27 patients (15 male, 12 female, mean age 56.3 + 13.2 years) underwent a comprehensive stress cardiac adenosine MRI and a nuclear cardiac SPECT exam within 24 h after presenting to the emergency room with chest pain. MRI at 1.5 T (Siemens Avanto, Erlangen, Germany) included steady state free precession (SSFP) cine MR imaging in short and long axis planes, adenosine stress and rest perfusion imaging, as well as delayed enhancement MR imaging. For the stress perfusion MR images, adenosine was injected i.v. at a rate of 140 [1g/kg/min over six minutes. Four minutes into the adenosine injection, satura- tion recovery (SR) SSFP MR images were obtained. Scan parameters were: TR/TE 2.4/1.0 ms, TI 180 ms, 50° flip angle, bandwidth 1000 Hz/pixel, FOV 36 x 27 cm, matrix 192 x 115, acquisition duration 150 ms, slice thickness 8 mm, acceleration

An anterior wall perfusion defect with FFR trace below showing a significant lesion (FFR 0.69) in the proximal left anterior descending artery. An FFR < 0.75 is considered to be haemodynamically significant. Synthelabo, Guildford, United Kingdom) and at rest. Main exclusion criteria were previous MI, previous CABG, renal failure, severe LV dysfunction or significant co-morbidities. All patients underwent clinically indicated cardiac catheterisation within 48 hours of CMR. Invasive measurements of FFR were performed in the catheter laboratory on all lesions of more than 40% severity. FFR was calculated as (Py-P,)/(P,-P,), where P., P, and Pq are simultaneous aortic, right atrial and distal coronary pressures measured during an intravenous infusion of adenosine at 140 Ug/kg/min. A vessel with <40% stenosis angiographically was assumed to be non-flow limiting. The ability of visual analysis of CMR to detect the presence of myocardial ischaemia per coronary territory was determined, blinded to the results of the angiogram and the FFR result.

ROC curves comparing the combined visual interpretation algorithm and individual components for the detection of significant CAD Results: Sixty-five subjects were prospectively recruited. Two individuals did not complete the CMR examination owing to claustrophobia and one withdrew consent, so 62 subjects were included in the final analysis. The prevalence of CAD was 66%. All CMR images were visually interpretable. For the diagnosis of significant CAD, the combination of perfusion and delayed enhancement CMR had a sensitivity, specificity and diagnostic accuracy of 90%, 81%, and 87% respectively, compared with 95%, 62%, and 84% for perfusion alone. The combined algorithm had higher specificity and accuracy owing to the high specificity (95%) of delayed enhancement imaging. The addition of resting wall motion analysis did not improve the diagnostic yield further. Receiver operating characteristic curve analysis demonstrated that the combination of perfusion and delayed enhancement imaging provided better performance than perfusion, delayed enhancement or wall motion assessment alone (figure |). Diagnostic accuracy was high in single-vessel and multivessel disease (84% and 81%, respectively), and independent of CAD location (LAD 82%, LCx 81% and RCA Methods: Subjects scheduled for elective diagnostic angiogra- phy for investigation of exertional chest pain were studied prior to angiography. Patients were studied with first-pass perfusion at 3 Tesla (Trio, Siemens Medical Solutions), at stress (140 mcg/kg/ min intravenous adenosine) and at rest. Three short-axis images were acquired every heartbeat using a saturation recovery fast gradient echo sequence and 0.05 mmol/kg contrast agent (Gadodiamide, Omniscan™, GE Healthcare) bolus injection.

Table | (abstract P50) CMR-derived infarct characteristics for patients with inducible and non-inducible VT

Correlation between VT cycle length and MCDE-derived gray zone.

Figure | (abstract P51) (a) Raw slice image, (b) Slice image with LV contours added and computer segmentation shaded.

Figure 2 (abstract P51) Bland-Altman plot of differences between computer and visual segmenta- tion.

3D rendering of LV and infarct (infarct mass: 14.1 g, iSA: 52.9 cm’, iSA/V: 3.9 cm/ml) superimposed on a basal LV short-axis DE-MR image.

Decrease in LV wall thickening in relation to infarct surface area to volume (iSA/V) ratio and relative infarct mass. shrinkage of hyperenhanced tissue over time, DE-MRI derived infarct surface area (iSA) and the iSA to infarct volume ratio (iSA/ V) may provide more stable indices of the extent of infarction than does scar volume and may better predict LV functional impairment.

Figure 2 (abstract P52) Relationshop of LV volumetric measurements to infarct surface area.

Short axis cine (top), delayed contrast (middle row) and stress perfusion (bottom row) images in a patient with apical hypertrophic cardiomyopathy. Stress perfusion abnormalities seen corresponding to (white arrows) and distinct from (yellow arrows) established fibrosis.

Table | (abstract P58) Figure | (abstract P58)

between LVEDV and RVEDV (R? = 0.83, p < 0.0001) (Fig. IA). The correlation between LA and RA was somewhat weaker, yet significant (R? = 0.56, p < 0.0001) (Fig 1B).

Conclusion: The present study presents normal values for wall thickening derived from short-axis MRI. These normal values can be used as reference for quantitative assessment of wall motion abnormalities in various cardiac disease states. Results: After one year there was no significant change in body mass index, weight, visceral fat mass or total fat mass. There were no significant differences in fasting serum glucose, serum cholesterol or systolic blood pressure. After one year of continued obesity, there was a significant 7% increase in left ventricular mass, and left ventricular mass when indexed to both height and height”. Right ventricular mass was unchanged over the year period. Both left and right ventricular end-diastolic

Measurements of absolute wall thickening (left panel) and fractional wall thickening (right panel) are displayed in a |7-segment model. Wall thickening measurements in the most apical slice is not meaningful in short-axis images and were therefore left out*. Absolute wall thickening is expressed in millimetre (standard deviation), fractional wall thickening in percentage (standard deviation).

*p < 0.05 vs CEMRA, 'p < 0.05 vs SSFP (same CA); no significant differences between corresponding sequences (Gd-DTPA vs Gadofosveset) Table | (abstract P63) Comparison of contrast agents and imaging sequences

Relative abdominal adiposity indices. normal variation in body fat seen in the general population during early adulthood is unknown. Cardiovascular magnetic resonance imaging offers the unique capability for clear depiction and accurate quantification of adipose tissue alongside an effective non invasive and radiation free assessment of both early anc advanced atherosclerotic disease. We investigated the relation between abdominal adiposity and early sub-clinical carotid atherosclerosis. P65

Newborn (body weight 3 kg) with heterotaxia, tetralogy of Fallot with pulmonary atresia and total anomalous pulmonary venous drainage to the portal vein. 3D volume rendered images obtained on a 64 slice MDCT scanner following hand injection of contrast in the umbilical venous catheter. Images show the pulmonary veins (PV, diameter < | mm) draining to a confluens which crosses the diaphragm supplying the hepar veins (HY) in posteriorlateral and inferior projections. Aorta (Ao), right atrium and ventricel (RA and RV), left and right pulmonary artery (LPA and RPA). P68

Figure | (abstract P68) Background: In children the visualization of the course of the coronary arteries is highly desirable for surgical planning and for follow up in CHD and Kawasaki disease. CMRA has been successful applied in adults and recently has been demonstrated in young children and infants. In children with CHD a whole heart CMR approach is of great clinical advantage, since the anatomy of the heart and the great vessels can be evaluated in addition to the coronary arteries. Usually, a 3D volume is obtained either at end-systole or mid-diastole. However, in children with high heart-rates and/or with RR variability it is unclear which phase of the cardiac cycle results in better image quality. Therefore, the simultaneous acquisition of the

Table | (abstract P68) Mean + SDV values for different parameters

Results: The dual cardiac phase whole heart scan was applied successfully in all patients. An example of each coronary segment is shown in figure |. The vessel length, diameter, sharpness and consensus score for each segment during systole and diastole are shown in Table |. Bland Altman plots of the systolic versus diastolic data from each coronary segment are shown in figure 2. No statistically difference was found comparing vessel length, diameter, sharpness for all vessels between systole and diastole. Moreover, no statistically differ- ence was found in image quality. Although, there was no difference for the mean vessel length, diameter, and sharpness, it was found that on a patient level, those parameters and image quality showed differences either favoring systolic or diastolic image acquisition for different coronary segments within the same patient. SEIS YSLONE GLI TERESI SLO Pildse HT COMPUTICLIOTT WIC LI WHOL heart approach allows retrospectively selection of the best image phase for coronary artery visualization without any scan time penalty. Methods: Eight children (age = 6.38 + 4.27, height = 116.57 + 30.77, weight = 23.57 + 14.25, Heart-rate = 85.02 + 8.59) with CHD were scanned under general anesthesia on a 1.5 T MR system (Achieva, Philips Healthcare). The cardiac rest period for end-systole and mid-diastole was determined from a 2D SSFP cine scan with high temporal resolution (TR/TE = 3.1/1.6 ms, flip angle 60°, slice thickness 6 mm, 60 to 80 cardiac phases). A previously developed free-breathing navigator gated 3D SSFP dual cardiac phase sequence [I] was then applied in sagittal orientation for imaging of the whole heart including the coronary arteries and great vessels (TR/TE = 3.4/1.7 ms, flip angle 90°, 60-120 slice, isotropic resolution of I-I.5 mm?, temporal resolution of 60 — 75 ms, Sense of 2 in AP direction). Data was obtained during end-systole and mid-diastole and the acquisition window of the 3D scan was adapted accordingly to the shortest rest period. Images were then reformatted along the major axes of the left and right coronary artery for both cardiac phases. Vessel length, diameter and sharpness of the visualized coronary arteries (RCA, LM, LAD, LCx) were measured using the “SoapBubble” software. Image quality was assessed by two independent observers. Statistical analysis and Bland Altman plots were used to compare the different data sets.

Table | (abstract P71) Numerical measurements of PMs for normal and dilated LV patients

Table | (abstract P72) Mean positioning error with respect to 3 manual contours and inter-observer variability. Dreg is the mean distance to reference contour number i addressed by any publication or commercial product. This is mainly due to the non-homogeneous intensity of the myocardium resulting from contrast agent accumulation in infarcted areas.

Final myocardium contours on 4 different patients. Then the delineation of the myocardial contours is performed by alternating automatic deformation of a geometrical template and computation of a binary map of the enhanced areas. The template is ribbon-shaped with a variable width, its position is updated depending on image gray values. The map is a 2D binary image showing enhanced areas, it is updated by thresholding the image gray values in a region of interest centered on the endocardium, in order to include sub-endocardial scars. The segmentation is performed as follows: (1) automatic delineation of myocardial contours on short-axis LECMR slices [2],

Table | (abstract P73) The levels of agreement, bias and standard deviations from Bland Altman tests for the cardiac motion analysis are shown

Bland Altman tests for the comparison of DENSE and single peak HARP analysis are shown for time to peak circ. shortening (A), relative circ. shortening (B), time to peak rotation (C) and amount of rotation (D).

Table | (abstract P74) Bland-Altman Analysis: k-t BLAST and SENSE cine SSFP

a) Volume-time curves of LV short axis slices at different locations. B) Time-Volume and dV/dt curves of mid-LV cavity short axis slice.

Bland-Altman analysis of IVRT between MRI and echo values.

The LGE sequence is modified so that a “fat restore” pulse precedes the first inversion pulse, and fat is not inverted. This permits the 2°¢ fat- selective inversion pulse to be effective in providing fat-suppression for a sequential acquisition order. Methods: Without a “fat restore” pulse, the fat experiences two 180°s. The I** 180° (the non-selective inversion always used in LGE) results in greatly reduced M,,, at the time of the 29 fat- selective |80°. The result is that fat regrows very quickly after the 2" fat-selective!80° and fat-suppression is not achievable for a sequentially ordered acquisition: fat crosses the null point too

Figure 2 (abstract P75) Phantom comparison of the fat suppression using standard method with centric order, and the sequential order with and without the fat restore pulse. Equivalent fat-suppression is provided by the fat restore technique. Edge enhancement is less using a sequential order. F = fat.

2D LGE comparing matched slices using centric, and sequential orders with and without the fat-restore pulse. The fat-restore pulse is needed for fat-suppression (thick arrows). Subtle edge-effects can be seen in the image with a centric acquisition (arrow heads). An area of enhancement at the tip of a papillary muscle (confirmed in other views) is also visualized in each image (thin arrows).

Table | (abstract P76) RMS positioning errors in mm for the LV endocardial and epicardial contours

First-pass perfusion image before (left) and after (right) KLT filtering. KLT filter provides marked noise suppression without loss of contrast between normally and abnormally perfused regions. Methods: Ten consecutive first-pass perfusion datasets inter- preted positive for ischemia or infarction were processed and analyzed retrospectively. All images were acquired using GRE-EPI with TSENSE acceleration rate 2 on a 1.5 T MR system (MAGNETOM Avanto, Siemens Healthcare, Germany). For each subject, all slices showing clinically interpreted perfusion defect were included in the analysis. Non-rigid body registration [3] was first performed on each series to allow for semi- quantitative analysis of signal enhancement, and to improve the

Figure | (abstract P79) (A) and (B) mid-ventricular T2-weighted image of a 2- and 8-weeks post-MI pig hearts, respectively. The yellow rectangles represent the ROls used for toroidal glyph visualization, which includes NI, I and B regions. Glyph are colored according to the fiber inclination angle as defined by Scollan et al. (1998, Am J Physiol). (C) Myocardial fiber angles in the infarct region at 2 weeks post-MI change from helical to more longitudinal with reorganization at 8 weeks post-MI (less longitudinal) (D).

A three-dimensional view of an 8-weeks post-MI heart, with the myocardial segmentation is shown (A). The other frames show a mid- ventricular slice of a T2-weighted image (B), a color-coded map of TV expressed as a percentage (C) and the color-coded fiber inclination angle map (D). The Infarct and Border regions are clearly distinguishable in the TV map and the laminar fiber architecture is altered in the Infarct region. scanner (Siemens, Erlangen, Germany) using a segmented EPI sequence, 6 gradient directions; b-values = 0 (T2-weighted) and 600 s/mm?; voxel-size = 2 x 2 x 2 mm?; 50 short-axis slices; TR = 5400 ms; TE = 84 ms; 40 averages (EPI factor = 7). Based on 12-weighted images, tissue was classified into Infarct (I), Non-infarct (NI), and Border (B) regions. The toroid-based

Mean TV (A) and fiber angle variance (B) computed from 16 regions of the standard cardiac polar map for normal, 2- and 8-weeks post-MI hearts. Myocardial regions were classified as NI, B or I based on the T2- weighted image. TV and fiber angle variance for all categories is significantly elevated at 2 weeks post-MI and remains elevated only in the infarcted region at 8 weeks post-MI. These findings indicate the progress of the remodelling process post-MI.

Objective: Interventricular dyssynchrony is typically assessed by pulsed-wave echocardiography (PW-Echo) as the delay between onset of aortic and pulmonary flow. Recent multicenter trials demonstrated the value of this interventricular dyssychrony to predict response to cardiac resynchronization therapy (CRT). In the present study, the ability of phase contrast magnetic resonance angiography (PC-MRA) was assessed to quantify interventricular dyssynchrony in comparison with PW-Echo.

Introduction: Early assessment of atherosclerosis (leading cause of death in West and soon world) remains an elusive

clinical goal, which if realized could lead to significant improve- ments in mortality and morbidity.

Scores of the anterior tibial artery on uncorrected images vs. corrected images in 6 subjects. Scores were improved in all legs with appreciable motion, but unchanged in all still. Introduction: Recently several new approaches for non- contrast MRA have been proposed. Many of them are based on

Absolute improvement in image scores for groups with initial score of |, 2, 3 and 4, respectively. All legs (60 segments) were pooled together. (Red figures: the total number of each group). Vessel depiction of segments with initial score of | or 2 (primarily due to motion) were substantially improved after motion correction.

Figure 3 (abstract P84) Image quality and artery depiction was improved after correction when motion was present in both legs (A) or only one leg in which case correction didn’t affect the originally still leg (B). UC uncorrected; CC: corrected.

recovery between baseline and 5 months (from 41.60 + to 45,8 + 12.4, p= .15), with no further change after 5 months (45,8 + 12.4; 45.56 + 10.7 p = .92) (see figure 1). ieee Samia neritic ncaa eeninaincinieniniese. \inliiaiaie’ —_—orerieo 45,8 + 12.4, p= 15), with no further change after 5 months (45,8 + 12.4; 45.56 + 10.7 p = .92) (see figure 1). SWT improved significantly between baseline and 5 months follow up (from 23% to 30%, p = 0.04), no additional improve- ment was seen at 5 years (SWT 33%, p = 0.29). TEI at baseline showed a good correlation with SWT at 5 months and 5 years follow up.

64-year-old female with mean PAP = 51 mmHg demonstrates hypertro- phy of SMT (septomarginal trabecula), white arrow, in a dilated right ventricle. Introduction: In the right ventricle (RV), the septomarginal trabecula (SMT) arises as a muscular band originating from the interventricular septum (IVS) at the lower segment of the crista supraventricularis. It forms a functional unit with the moderator band, which attaches to the lateral free wall of the RV [I, 2]. Strategically situated between the RV inflow and outflow tracts, the whole unit serves to help emptying blood into the pulmonary trunk during systole. Thus, it should be anticipated that the SMT may undergo changes in RV hypertrophy secondary to chronic pulmonary hypertension.

Figure 3 (abstract P91) Methods: Imaging was performed in two centers; one using 3 T and the other using |.5 T MR systems. 33 catheter proven PAH patients (mean age = 61.4 + 12.1 years and mPAP = 45.9 + 12.4 mmHg) were enrolled in the study [Idiopathic pulmonary arterial hypertension (IPAH) = 21 and Scleroderma (SSc) = 12 patients]. Similarly, 9 healthy volunteers (mean age = 45.56 + 8.6 years) were included in the study for comparison. Short axis cine images were acquired using fast gradient echo technique. End diastolic frames were analyzed using MASS 6.2.1 software (Medis, the Netherlands). Starting from the basal slices, the SMT was identified in patients and controls as the most anterior trabeculation arising from the IVS below the outflow tract level. Two independent observers manually contoured and traced the SMT from its origin towards the apex where the moderator

Figure 2 (abstract P91) Purpose: To assess SMT mass in pulmonary arterial hyperten- sion (PAH) and its relationship with RV function and pulmonary hemodynamics using cardiac MRI.

Myocardial HEP metabolism in patients with DM2. PCr/ATP ratio = ratic of phosphocreatine over adenosine triphosphate. *P < 0.05. p < 0.05), reduced whole body insulin sensitivity (0.45 + 0.48 vs 0.74 + 0.44 (mg/kg -min)/(pmol/L), p< 0.05), and reduced MMRglu (0.21 + 0.13 vs 0.34 + 0.14 mmol/mL/min, p < 0.05), as compared with patients without FL, while MBF was not different. The ratio of phosphocreatine over adenosine tripho- sphate, a marker of myocardial HEP metabolism, was reduced in patients with FL (1.90 + 0.35 versus 2.27 + 0.29; p < 0.05), also after adjustment for BMI, and correlated to MMRglu (r = 0.43, p < 0.05). LV systolic and diastolic function were not statistically significantly different. Figure |

Methods: Ten patients with severe, but compensated AS underwent 4 serial 3D cardiac MRIs (CMR) with |.5 T EXCITE (GE Milwaukee, WI) out to 4 years (40 time-points: baseline, 6 months, | year and up to 4 years). 3D LV volumetrics were measured. LV diastolic function was assessed by a phase velocity mapping slice placed at the tips of the mitral valve leaflets, acquired in a through-plane manner with temporal resolution of 25 +5 ms. Interrogation of resultant time-velocity curves was performed to resolve: |) E:A ratio as mean and peak absolute and relative velocities 2) deceleration time. Figure |.

This pilot study specifically assessed patients identified on echocardiography to have had anthracycline-induced cardiac damage. A minority of these patients were found to have normal CMR findings. Most patients had poorly contractile LV myocar- dium, with low LV mass. Fibrosis or scarring, sought using conventional late gadolinium imaging techniques, was not evident in any patient. Further work is needed to find discriminative CMR indices to describe the long-term effects of myocardial toxicity and to predict prognosis in this patient group. See Figures | and 2.

Table | (abstract P96) Aortic root dimensions

Introduction: Magnetic Resonance Imaging (MRI) can be used to measure common carotid artery maximum wall thickness Mean differences and 95% Cl for diameter aortic ring, sinotubular junction and aortic sinus are displayed in table | (*p < 0.05). the continuity equation. Diameter of aortic ring, sinus and sinotubular junction were measured using steady state free precession CMR, 2D echocardiography and invasive aortography. Results: Mean differences and 95% Cl in AVA were 0.02 cm? (0.04, 0.08) (p = NS) for catheterization versus Doppler echocardiography, -0.01 cm? (-0.08, 0.06) for catheterization versus 3D echocardiography (p = NS) and 0.01 cm? (-0.07, 0.08) for catheterization versus CMR (p = NS). Mean differences and 95% Cl for diameter aortic ring, sinotubular junction and aortic sinus are displayed in table | (py < 0.05). Journal of Cardiovascular Magnetic Resonance 2009, | (Suppl 1):P97

Probability of freedom of re-intervention at 10 years after the Ross Procedure. Data presented as Kaplan Meier plots demonstrating the probability of freedom from re-intervention for the autograft (Panel A) and the homograft (Panel B). The total number of subjects relative to the years after the Ross procedure is shown. Conclusion: Biomarkers of oxidative stress and inflammation are better predictors of MRI quantified carotid artery disease than conventional risk factors. Whether progression of disease will also be predicted by these markers needs further study. A multi-modality approach was employed, with subjects performimg cardiopulmonary exercise testing, 2D-echocardio- graphy and cardiac MR imaging (including late enhancement).

SE = subendocardium, FT = full thickness. Table | (abstract P98) re-intervention rates of less than 20% are in agreement with earlier published work.

Table | (abstract P102) Differences in net flows between corrected and uncorrected flow mapping

(a) Bull's eye plot showing patient-specific coronary supply territories. The dotted lines represent the |7-segment model. (b) Patient-specific coronary supply territories as an overlay on a bull's eye plot of a late enhancement scan.

Table | (abstract P104) Mean EDV, ESV and EF values + SD and ranges as determined by reader | and 2 with the SoS and GPM approach, respectively \2. = reader 2, EDV = end-diastolic volume, ESV = end-systolic volume, EF = ejection fraction, SoS = summation of slices, GPM = guide-poin' nodeling.

Bland-Altman-Plot for assessment of ejection fraction. (a) Reader | (RI): summation of slices (SoS) versus guide-point modeling (GPM) method. (b) Reader 2 (R2): SoS versus GPM approach. (c) Interobserver comparison of the SoS method based on a stack of short axis views. (d) Multi-orientation highly accelerated sequences analyzed with the GPM approach: RI versus R2.

Pulse sequence diagrams of rectangular pulse and adiabatic-rectangular pulse train. Methods: Pulse sequence: Figure | shows pulse sequence diagrams of a conventional rectangular pulse and an adiabatic- rectangular pulse train. This pulse train consists of a rectangular 130° pulse with 0.8 ms duration, a rectangular 85° pulse with 0.5 ms duration, and an adiabatic half-passage pulse with |.5 ms

Representative normalized saturation images in six different views of the heart: PDW (left), rectangular pulse (middle), and adiabatic-rectangular pulse (right).

(a) Decomposed water and (b) Fat fraction images (c) Fat fraction from six segments (contoured on the water image). Fat deposition was observed on the inferior lateral wall (segment 3).

The fat fraction comparisons shown on the MRI is greater than those of MRS, likey due to noise bias contributing to the fat fraction calculation.

Purpose: To develop and assess the accuracy of a novel prototype-based segmentation method where few cases, typically five, are needed to introduce the required a priori information. prototype-based segmentation method where tew Cases, typically five, are needed to introduce the required a priori information. Methods: Ten healthy volunteers underwent contrast enhanced MR imaging of the aorta at 1.5 T (Philips). Imaging employed a steady state free precession sequence, resolution 1.6 x 1.6 X | mm. The main idea behind the novel prototype- based segmentation algorithm is to use spatial a priori information to restrict the segmentation rather than to govern what to include. A priori information was extracted from five of the cases (training set) and stored into a prototype. Manual delineation of the aorta was performed in all ten cases and four anatomic landmarks were defined in each set of images. A standard level set segmentation was applied to the training set. The level set segmentation was compared to the manual delineation for each of the cases in the training set. The training set images were aligned to each other by the use of the four landmarks. The spatial information on where to restrict the segmentation, called a correction map, was calculated as the mean of the difference between the level set segmentations and the manual delineations for the training set. The correction map and intensity information from the training set were stored into a prototype. The remaining five cases were used as test set and the segmentation was done by the use of the prototype and a level set method. In the test set the segmentation error was calculated as both a volumetric error and a mean absolute distance between the manual delineation and the prototype- based segmentation.

Introduction: Recent studies of right ventricle (RV) structure and function have recognized the critical role of the RV in maintaining cardiac function. Previous studies on the RV have been limited by sample size and inadequate inclusion of participants from different races. The Multi-Ethnic Study of Atherosclerosis (MESA) is a large ongoing multi-center study that aims to investigate the subclinical progression of cardiovascular disease in an ethnically diverse clinically asymptomatic population. Conclusion: Age, gender and BMI were related to RV mass and volumes. Other than blood pressure, traditional cardiovascular risk factors (diabetes mellitus, smoking, alcohol consumption, LDL cholesterol) had little relationship to RV mass and volume in the MESA cohort.

Correlation between epicardial and visceral fat at baseline.

Table | (abstract P115) Diagnostic clarity score

Table 2 (abstract PI15) Proportion of valve components with a high-quality diagnostic imaging (Diagnostic clarity scores | or 2)

Methods: We retrospectively reviewed 7] consecutive cardiac magnetic resonance (CMR) studies for BPA size and PCMR data. We also reviewed |3 consecutive pts who underwent both CMR and catheterization.

Table | (abstract P!17) Comparison of LV-METRIC to manual tracing automated algorithms often fail or require extensive user interface, possibly due to the fact that they employ assumptions regarding cavity shape or regional propagation. We developed an automated segmentation algorithm (LV-METRIC) that involves no geometric assumptions and instead quantifies LV EF and volume based on local per-pixel signal intensity while accounting for partial voxel effects. The purpose of the current study was three- fold; (1) to evaluate LV-METRIC performance vs. MT among a broad unselected patient population; (2) to compare processing time by LV-METRIC to MT, and (3) to compare LV-METRIC and MT to an independent standard of LV flow quantification. than MT measurements, respectively. Full voxel computation yielded less drastic differences, with diastolic volumes smaller by 4 mL and systolic volumes larger by 1.4 mL. All EF and volumetric differences between LV-METRIC and MT were significant (p < 0.001). Correlations for cavity volumes, global stroke volume, and ejection fraction were high (all R? > .95). Both LV- METRIC and MT similarly agreed with phase contrast (all differences p < 0.05); Although volumetric stroke volume by MT was, on average, similar to PC (A5.9 mL), there was substantial variance between the two techniques (SD + 13.4 mL). Similar differences were evident when comparing PC and LV-METRIC with (6.2 + 14.2 ml) or without (—3.9 + 14.7 ml) partial voxel interpolation.

to NL (Figure I). In both groups motion was lower in superior than in mid or inferior regions (p < 0.001 for all).

Breath-hold cine steady-state free precession imaging in a four- chamber view visualizing the mitral regurgitation jet (flow void) due to flail a posterior leaflet.

Figure | (abstract P122) Transesophageal echocardiogram visualizing a flail posterior leaflet of the mitral valve.

Distribution of eGFR in patients referred for contrast-enhanced CMR scans.

Awareness of moderate to severe renal impairment (eGFR 30-60 ml/min/ 1.73 m’) prior to blood results at time of CMR scan. All patients with eGFR of <30 ml/min/I.73 m? were identified either by the referring physician or by the safety questionnaire prior to the scan. The situation was very different for those with eGFR between 60 and 90. Of the 79 patients with eGFR in this range, only a minority were identified by the referring physician or were aware that they had renal impairment (15.2% and 20.3% respectively — Figure 2). Conclusion: This retrospective study highlights the significant numbers of patients with at least moderate renal impairment who are referred for CMR scans with an indication for GBCA. Our results suggest that most patients with moderate to severe renal impairment (eGFR 30-60 ml/min/I.73 m*) will not be identified without blood testing. In view of the fact that NSF has been reported in these patients (albeit with a concurrent illness), this has important implications and supports the current practice of measuring renal function on all patients. Decisions about giving gadolinium-based contrast agents need to be made based on clinical need for the information which will be gained together with the informed consent of the patient concerned. mM . £.. ..- eee

Enhanced MR coronary angiography — normal coronaries. DE in myocarditis.

Pulse sequence diagram. Dotted lines in the gradient lobes represent the original gradient values without the demodulation

Representative cardiac phases for planning results using the proposed technique. Columns represent time frames while rows show different planning slices.

A) Fat image; B) blood pool image; C) fusion of fat and blood pool with fat highlighted in red. Figure | (abstract P128)

Four axial slices (inferior to superior) of the LA are shown in two subjects, with fat and blood pool images combined, and fat highlighted as brighter signal. Fat was present below and adjacent to the right inferior PV (labeled #1 in Figure 2, the posteromedial left atrial GP), below the left inferior PV (#2 the posterolateral left atrial GP), adjacent to the right superior PV near the superior vena cava (#3, the superior right atrial GP), above the left superior PV (#4, the superior left atrial GP), along the anterior wall (#5), and above the right inferior PV (#6, the posterior right atrial GP).

Results for quantification of the kinetic energy over time for the tracked volume from Figure |. Note the slight increase in kinetic energy in the volume during early diastole and the sizeable increase in kinetic energy during early systole. Results: The proposed Volume Tracking method is feasible. Volumes of arbitrary shapes and sizes can be interactively visualized and tracked over the cardiac cycle. Intuitive visualiza-

An example of Volume Tracking visualization in a healthy volunteer. Left: Early diastole, a white spherical volume of blood positioned near the mitral annulus. Middle: Late diastole, the same blood volume has now deformed as it has entered the left ventricle. Right: Early systole, additional deformation of the volume as it is ejected into the aorta.

Table | (abstract PI31) Relationship between cardiac allograft vasculopathy and parameters for left ventricular function

Figure | (abstract P132) Percent change (mean + SD) from baseline in coronary artery area, peak diastolic coronary flow velocity and flow during isometric handgrip stress for arteries with mild vs. severe coronary disease. (*p = 0.007, + p = 0.02 mild vs. severe CAD)

Example images (from left to right) acquired in Bangkok (TE = 2.1 ms), London (bright and black preparation, TEs = 2.2 ms), and ex vivo scan of a mid-ventricular slice (TE = 2.5 ms), respectively.

Examples of T2* curve fitting using both truncation and offset models. Left: In vivo scan data; Right: Ex vivo heart data. Methods: A 23-year-old Thai female thalassemia majorpatient with longstanding cardiac and liver iron overload was studied. The patient underwent MRI scan using the bright blood T2* sequence [2] (twice for reproducibility) in Bangkok on a 1.5 T clinical scanner (GE Signa), again in London 9 days later on a 1.5 T one (Siemens Sonata), using the same sequence. The patient was also scanned using the black blood T2* sequence [4]. The patient died two weeks later. With full ethical approval, the donated heart was formalin fixed before being sliced axially. A

Introduction: A large percentage of thalassemia major (TM) patients shows a significant heterogeneity of segmental distribu-

Table | (abstract P135) Phantom correction of phase-contrast measurements

Figure | (abstract P136) Colour coded plasma flow images over 8 slices (1 = distal, 8 = central) at day 7 after left side SFA excision. Images show significantly higher plasma flow on the ischemic side before angiogenetic therapy. Purpose: To implement and evaluate the use of perfusion MR in monitoring angiogenetic therapies and their effects based on a rabbit model of iatrogenic induced chronic lower limb ischemia. Methods: MR perfusion imaging was performed in 8 rabbits with chronic lower limb ischemia after unilateral SFA excision at 3 Tesla (Magnetom Verio, Siemens Healthcare) using a 32- element coil. In 4 animals MRI was performed at day 7 and in 4 animals at day 7 and 35 (after angiogenesis therapy). Multi-slice coverage was provided using a 2D-TurboFLASH technique at 1.5 s temporal resolution with repeated measurements over |0 min after injection of 0.1 mmol/kg gadobutrol (Gadovist, Bayer Schering Pharma). One slice was placed through the aorta for = oe

The myocardial infarction size was significantly higher in patients who died (21.6% vs. 14.4% p = 0.01). The relationship between infarct size, end-sistolic volume index, end-diastolic volume index, ejection fraction of left ventricle and systolic volume index, were statically significative (p < 0.001). The mayor adverse cardiac events (MACE) were presented in 90% of the group of death patients (p = 0.001). Figure |.

The images obtained by axial acquisition (A) and oblique sagittal acquisition (B). Purpose: The purpose of this study was to evaluate |) the image quality, 2) the coronary length that can be evaluated, and 3) the scan time, between image acquisition by AX and SA.

* = p < 0.05 Pre vs Post Weight Loss # = p > 0.05 Surgical Vs Dietary Weight Loss. Table | (abstract P139)

Table | (abstract P140) Echo performance in relation to LV geometry study. CMR and echo were performed within a seven day interval (A0.8 + .2 days) and interpreted blinded to results of the other modality. Echo included both non-contrast echo (NC-echo) and contrast echo (C-echo) imaging in all patients. CMR (1.5 T) included cine-CMR (SSFP) and DE-CMR (inversion recovery; standard Tl = 250-350 msec/long Tl = 600 msec). DE-CMR was the reference standard for LVT. Left ventricular (LV) function and geometry were quantified by cine-CMR (planimetry) and echo (linear) methods. Infarct size was measured on DE-CMR. Echoes were graded for image quality on a uniform scale to account for endocardial definition, artifacts, and number of segments imaged. Results: 130 patients were studied (62 + 13 yo, 98% CAD, NYHA class 2 + .8). LVT prevalence was higher by DE-CMR than C-echo (21% vs. 14% p < 0.05). Patients with LVT had larger infarcts (OR |.5 per 10% LV p = 0.01), lower ejection fraction (OR 1.9 per 10% EF p = 0.01) and more aneurysms (OR 3.3 p < 0.05). Echo performance did not vary by clinical parameters or image quality, with similar reader-assigned quality scores between NC-echoes that did or did not detect LVT (7.0 + 1.9 vs. 6.8 + 1.7, p = .8). Ejection fraction was lower among patients that derived improved LVT assessment by C-echo compared to those in whom NC-echo alone accurately assessed LVT (32.1 + 10.7% vs. 39.9.0 + 13.4%, p < 0.05). Similarly, improved LVT assessment by DE-CMR was associated with lower ejection fraction and greater cavity dilation as measured by either echo or cine-CMR (Table |). In multivariate analysis, severity of EF impairment (OR 1.8, p < 0.05) was an independent marker for added utility of DE-CMR after controlling for LV volume. Background: Obesity is associated with diastolic dysfunction at rest. However, it is unknown whether this is exacerbated during stress, i.e. whether the “relaxation reserve” of the heart is impaired in obesity. We aimed to determine the effect of simulated exercise on diastolic function in obese subjects without co-morbidity, and compared this to normal weight subjects.

The effect of dobutamine infusion on peak filling rates in obese and normal weight subjects. tl Results: At rest, obesity was associated with a 20% reduction in LV peak filling rate (3.6 + 0.8 vs 4.6 + 1.0 ml/s, p = 0.01). As would be expected from the action of dobutamine, in normal weight subjects, with an increase in heart rate (HR) of 60% during stress (from 63 + 9 to 99 + 15 bpm) there was a 66% increase in maximal PFR; furthermore, HR and diastolic function showed a close linear correlation (R? 0.86, p < 0.001). In contrast, in obese subjects, a 62% increase in HR (from 65 + 9 to

Figure 2 (abstract P142) Data table -VM+BA linear regression AAD vs ASBP and APP.

Table | (abstract P143) Univariate and multivariate regression analysis testing the association between LV-aortic root angle

Quantification of LVARA using the 5 chamber view.

The temperature dependence of T2* measurements in the phantom.

Methods: Male Wistar rats (weight = 200-240 g, n = 21) were anesthetized with 2.1% isoflurane added to | I/min of pure Op. Heart rate, breath rate, temperature, blood oxygen saturation and arterial blood pressure were recorded. In 7 rats, myocardial perfusion was assessed on a Bruker Biospec 4.7 T horizontal MRI system using an ECG- and respiration-gated IR gradient-echo technique (resolution = 234 x 468 um?, TE = 1.52 ms, slice thickness 3 mm, acquisition time 25 min at 350 bpm) at rest and during adenosine infusion (140 [Ug/kg/min). In the I4 other animals, under the same physiologic conditions, a mixture containing 200,000 fluorescent microspheres (Yellow, I5 + 0.1 tum; Triton, San Diego, CA, USA) was injected into the left ventricle at rest in 7 animals and during adenosine infusion in 7 other animals. The animals were killed by instantaneaous injection of pentobarbital. Hearts were harvested and samples were processed for fluorescence spectroscopy. A two-tailed unpaired Student's t-test was used to compare groups, All values are given as mean + SD.

Navigator signal (a) and water suppressed single voxel PRESS spectrum (b) with the selected volume superimposed on short axis (c), 2 chamber (d) and 4 chamber (e) 1H MR images from cine MRI.

2) CN have a FeCo core with a graphite shell, with Cy5.5 also attached for fluorescence imaging (dose — 32.14 UgFe/mouse, 8 nmol Cy5.5/mouse). Results: In vivo MRI showed the accumulation of HFn-Fe and CN in the ligated left carotid arteries (Figure 1A), with T2* signal loss causing significant reduction in luminal area (HFn-Fe; 24 H 48 + 8.9%, 48 H 33.9 + 11.3%, CN; 24 H: 26.1% + 13.0%; 48 H: 28.0% + 5.7%, Figure 2). There was no evidence of T2* signal loss in the non-ligated right carotid arteries (HFn-Fe; 24 H: 1.6 + 2.8%, 48 H: -5.6 + 8.8%, CN; 24 H: 4.6% + 9.6%, 48 H: - 5.5% + 1.2%). No significant signal was seen in sham-operated mice (HFn-Fe; 24 H: 2.7% and 48 H: -0.2%, Figure |B and 2). For confirmation, CN fluorescence imaging showed high signal intensity in the left carotid arteries (Figure 3) with no accumulationseen in either non-ligated right carotid arteries or sham-operated left carotid arteries. HFn-Fe uptake by macro- phages was confirmed on histology (Figure 4). 6 nmol Cy>s.3/mouse). MRI at 24 and 48 hours was performed on 7 T MRI scanner (Varian, Inc. Walnut Creek, CA) using a gradient echo sequence (TR/TE = 50/4.2, slice thickness = 0.5 mm, FOV =3 cm, matrix = 256 x 256, FA = 50, acquisition time =9 min 55 sec). Accumulation of nanoparticles on MRI was measured as the size of susceptibility artifacts (% reduction of lumen area). After final in vivo MRI, both in situ and ex vivo fluorescence imaging was performed using Maestro™ in vivo imaging system (CRi, Woburn, MA). Perl's iron and immunofluorescence staining was also performed to confirm co-localization of nanoparticles with macrophages.

Conclusion: Both iron oxide protein cage nanoparticles and graphite/FeCo core-shell nanocrystals allow noninvasive MRI of macrophage accumulation in mouse atherosclerosis. These powerful, multi-functional nanoparticles may allow improved noninvasive characterization of vascular inflammation and atherosclerosis.

Sliding partial MIP images of 3 T whole heart coronary MRA acquired with a patient-specific narrow acquisition window (50 ms) in the cardiac cycle.

Table | (abstract P155) Averaged image quality score of 3 Twhole heart coronary MRA compared with |.5 MRA.

Panel A depicts a patient in whom the size and extent of perfusion defect and delayed contrast enhancement (DE) are matched. In Panel B, no perfusion defect is seen despite a large area of DE. In Panel C, the perfusion defect significantly underestimates the size of DE.

Values are mean + SD; Mitral E DT = Mitral inflow early diastolic deceleration time; TVI E' = Early diastolic tissue velocity; E/E’ inflow to early diastolic tissue velocity. ratio of early mitrz Table | (abstract P157) Selected parameters and infarct sizes by diastolic function classification

MRI infarct size (% of LV) grouped by diastolic function. Figure | (abstract P157)

Drawing of ROls in the left ventricular septum from the same subject. Left: Tl mapping of the myocardium; Right: Black blood T2 image at TE of 4.8 ms.

Correlation curve between T! and T2* measurements. The vertical broken line represents the cut-off myocardial T2* value to distinguish abnormal and normal patients. All data of T2* 2 20 ms is fitted with a linear trend line (R = 0.84).

An example of mid-wall LGE present in the mid-inferior and lateral walls (arrows) in a patient with a strong clinical suspicion of acute myocarditis (A). The area confined represents the detected LGE (B).

Left anterior descending coronary artery blood flow (ml/min).

Table | (abstract P163) Comparison of MR and echocardio- graphy for thrombus positive studies with thrombus negative studies. *Two of the negative MRs were suboptimal studies due to motion and prothetic valve artifacts.

a. On PSIR Turbo FLASH there is a thrombus (dashed arrow) within the apical region of a mildly dilated left ventricle. This is located adjacent to a transmural infarct (arrow) b. No thrombus is identified on echocardio- graphy.

Figure | (abstract P164) Figure 3 (abstract P164)

Table | (abstract P167) Basal slice mean peak tangential strain values and standard deviation for three confirmed ARVC cases, six suspected cases, and four normal volunteers for the LV and RV segments of abnormal strain and motion, which occur largely in the region of the RV insertion points and the LV inferior and anterior septum. Abnormal RV strain patterns were evident in 2 out of the 3 confirmed, and 3 out of the 6 suspected cases. The other confirmed case had a dilated RV but the strain pattern appeared normal. Table | presents regional peak tangential strains for the basal slice averaged for each group. A confidence measure for Ett was calculated based on the signal-to-noise ratio (SNR) of the magnitude images. Only Ett strain values with confidence greater than 0.5 (normalized to max confidence in entire heart) were used to compute regional peak Ett strain. More than half of the data points in the diaphragmatic RV had to be discarded because of low SNR. Both confirmed and suspected ARVC cases have low peak strain values in the inferior and anterior septum and the RV free wall. Peak Ett is significantly less in ARVC and suspected ARVC cases compared to normals in the anterior septum (p < 0.005 and p < 0.05, respectively, student t-test). over two breath holds per slice. Imaging parameters include: FOV = 400 mn, slice thickness = 7.0 mm, TR = 24 ms, FOV phase = 62.5%, ETL = 9, segments = 18, cardiac phases = |0- 16, and displacement encoding frequency = 0.1 cycles/mm. Epicardial and endocardial contours were manually drawn for both the LV and RV on all cardiac phases. Phase unwrapping and tissue tracking were performed [5], and motion trajectories were estimated using temporal fitting with 5“ order Polynomial functions. Lagrangian strain taken tangential to the midwall (Ett) was computed from both the left ventricle (LV) and RV motion trajectories [3].

Cine DENSE magnitude (left) and tangential strain maps (right) for a normal volunteer (a), and one confirmed ARVC case (b). Results: Example basal slice cine DENSE magnitude images and corresponding end systolic tangential strain maps are shown for a normal volunteer (Figure |a) and a confirmed ARVC case (Figure |b). The underlying colour represents Ett and the purple vectors represent displacement. The arrows in Figure |b indicate regions

MRI system (Siemens, Erlangen, Germany). Phase contrast MR images were prescribed orthogonal to the long axis of the main pulmonary artery (MPA), right pulmonary artery (RPA) and left pulmonary artery (LPA), using a flash 2D gradient echo sequence with a predefined maximum velocity of 100 cm/second (TR = 32 milliseconds, TE = 3.8 milliseconds, Flip angle = 20°, FOV = 34 x 34cm, matrix = 256 x 256, slice thickness = 8 mm). Short axis fast gradient echo cine images were acquired for ventricular function quantification. Using dedicated flow software (MEDIS, Netherlands), magnitude images were semi-automatically con- toured. Flow hemodynamics including: flow volume (ml/min), maximum velocity in cm/sec (Vmax), systolic and diastolic cross- sectional areas (mm7) were computed for each vessel. Pulmonary arterial distensibility (PAD) was derived as (maximum systolic area — minimum diastolic area)/maximum systolic area) and expressed as % change. Mann-Whitney test was used for inter- group comparisons. Spearman's rho correlations were per- formed.

the corresponding CC and IC regions were derived using Simpson's formula from sequential 2 mm slice thickness images. IMT-vol and MRI-vol were compared in 61 evaluable carotid segments and % change in volume was compared in subjects with 6 month follow-up.

Flow profile.

Left subclavian steal. Complex flow analysis in patient with bypass graft for interrupted aortic arch.

Complex flow analysis in patient with bypass graft for interrupted aortic arch. Results: Presented are a series of curves that demonstrate differential flow patterns in the 4 enrolled subjects. Figure | presents flow in the ascending aorta (AsAo), main pulmonary artery (MPA), and branch pulmonary arteries (RPA and LPA) in a normal control. The AsAo and MPA flows are equal, and the sum of the branch PA flows equal the MPA flow, demonstrating internal consistency. Figure 2 presents the same flows in a patient with tetralogy of Fallot and LPA stenosis. Evident is diminished flow in the LPA. Again, internal consistency is demonstrated with equal AsAo and MPA flows, and the sum of the branch PA flows equalling the MPA flow. Figure 3 presents flow in a patient with

Left-ventricular ejection fraction LVEF changes. The rats treated with ESCM showed much higher the increase of LVEF change than the control rats both at | month (p = 0.0003) and 2 months (p = 0.000048), while the infarction size at the day | post-op was no significant difference between two groups (20.02 + 3.11 v.s 19.09 + 4.72, p = 0.65).

Table | (abstract P176) Mean differences in left atrial volumes and ejection fraction during dobutamine CMRI was defined as (LAV max-LAV min)/LAV max. All volumes were indexed to body surface area. One-way Analysis of variance (ANOVA) was used to detect the differences in LA volumes and LAEF across stages of DCMR. Linear regression analysis was used to find the independent predictors of LAEF and LA volumes.

End-systolic E2 maps of (top row) control and (bottom row, patient comparing tagging and DENSE.

Plots showing linear correlation (left) and Bland-Altman (right) on E2 measurements (n = 4334). Cardiac imaging: Twelve healthy human subjects (7 males; 5 females; mean age = 34.5 + I 1.0 years), with no history of heart disease and no risk factors for coronary artery disease, and thirteen

Mean MSI (upper plot) and THI (lower plot) computed from LS (squares) and LD (circles) images as a function of Tres for TR = 3.5 ms (left) and 6.0 ms (right). Note that both mean MSI and THI increase with increasing Tres:

Short axis SSFP images (TR = 6.0 ms) of med ventricle with different Tres A,E:18 ms; B,F: 42 ms; C,G: 84 ms; and D,H: 204 ms (upper row from LS and lower row from LD). Note that as Tres increases the signal inhomogeneity increases and endocardial delineation becomes more difficult as signal from the blood pool blends in with the myocardial signal, likely due to motion.

Table | (abstract p180) LV ejection fraction, volumes and analysis time (Observer A). Data are mean + SD *p < 0.05.

*p < 0.05 for difference. Table 2 (abstract p180) Interobserver variaion between stard LVF and 4DVR analysis.

Bland-Altman plots shows good agreement of both methods for calculating left ventricular ejection fraction (LVEF) for both observers. Mean + 1.96 * standard deviation is indicated for observer A.

Functional confirmation of the MRI RG and MRI and BLI. A In vitro of hESC-RD with SPIO-conjugated anti-c-myc and HA antibody showed strong T2 weighted dephaing signal (c-d). Non-transduced mESCs showed no T2 weighted dephasing signal (a-b). MR image was taken with GRE sequence using the following parameters: 100 ms TR, 30 ms TE, FA 45, FOV 12, NEXI, 192 x 192. B. Long axis view of mouse heart after injection of 0.5 x 10° magnetically pre-labeled hESC onto the lateral left ventricular wall. C. Validation of viability of hESC using biolumines- cence imaging (BLI). D. MEMRI of transplanted mESC-RG on bilateral mouse hindlimbs (indicated by white arrows). E. Viability of the injected mESC was confirmed with positive luciferase activity using BLI.

Figure | (abstract P182) Horizontal axis: phases of the cardiac cycle (22.4 ms per sample). Vertical axis: area of the lumen normalised by the proto-diastolic area. Plain line: ascending aorta, dotted line: descending aorta. The propagation delay of the pulse wave in the aortic arch is visible and the larger distensibility of the aortic wall at the level of the descending aorta is visible as well.

subjects. The technique, which has been previously validated, was performed with parallel imaging after the standard clinical MR evaluation. Results: 4D Flow evaluation of the ascending thoracic aorta revealed markedly abnormal systolic helical flow in 6 of 8 patients with bicuspid aortic valve. Five of these patients demonstrated an eccentric right anterior systolic jet and right-handed helical flow; at least two of these patients had fusion of the right and left coronary cusps, and three had aneurysms of the ascending aorta along with aortic stenosis and/or regurgitation. The other patient

had a left posterior jet with left-handed helical flow, dilation of the aortic root and fusion of the right and noncoronary cusps. The two BAV patients with normal systolic flow in the thoracic aorta had central flow jets. Normal and abnormal systolic flow patterns are included as Figure |: la demonstrates normal flow with streamlines mapped onto a sagittal oblique plane viewed from right and left orientations respectively, Ib shows an eccentric jet resulting in marked right-handed helical flow in a patient with BAV and aortic coarctation but without aneurysm, Ic shows left-handed helical flow in a BAV patient with fusion of the right and noncoronary cusps. Evaluation of aortic flow jets in BAV patients and healthy subjects is included as Figures 2 and 3: P185

An example of right inferior PV frequency shift in a healthy subject through the cardiac cycle. PV images (left columns) and the corresponding off-resonance map (middle and right columns) at the same imaging plane are shown. The off-resonance at proximal PV and atrial blood are measured. TD = ECG trigger delay. Introduction: Non-contrast enhanced pulmonary vein (PV) MRA is a potential alternative to contrast enhanced methods for pre-procedural planning and post-procedural evaluation of pulmonary vein isolation (PVI) in treatment of atrial fibrillation (AF). In 2007, 12% of the AF patients referred to pulmonary

Variaions of PV off-resonance frequencies between subjects, PV branches and cardiac phase. RI = right inferior; LI = left inferior; RS = right superior; LS = left superior.

Figure 3 (abstract P185) Example images from a healthy subject using on-resonance (top) and 75 Hz off-resonance (bottom) SSFP sequences. Slices A, B, C and D are part of a 3D data set. PV branches are highlighted in off-resonance SSFP (arrows) whereas they are suppressed in on-resonance SSFP (arrow heads). Methods: PV images were acquired on hea thy volunteers in this study. We used a dual echo sequence based on gradient recalled echo (GRE) to measure the field map left atrium in cine mode. The off-resonance o measured by defining a region of interest (RO of the PV and the a PV branch was ) in the proximal PV of the generated field maps and calculating the mean off- resonance frequency in the ROI. A 3D ECG gated free-breathing SSFP sequence with a navigator echo, which has been used in coronary imaging, was adapted to PV imaging. Balanced SSFP has a well-known signal profile modulated by frequency shift. By applying a linearly increasing radiofrequency (RF) excitation

short Tynax) Would be indicative of preserved microvascular integrity and predictive of small infarct size and vice versa. Assessment of myocardial scar, determined by contrast- enhanced magnetic resonance imaging (MRI), 2 to 4 days after infarction, deemed as the standard reference for estimation of infarct size. Briefly, 10 minutes after 0.2 mmol/kg body weight gadolinium contrast injection, 2 volume stacks of inversion- recovery gradient-echo images covering the whole left ventricle were generated. Infarct size was calculated as the amount of hyperenhanced myocardium related to the total LV-mass. Infarct transmurality was assessed visually based on a 5-grade scale (i.e. | = 0-25%, 2 = 25-50%, 3 = 50-75%, 4/5 = 75-100% transmur- ality without/with microvascular obstruction). in the infarct related coronary territory, in order to estimate the time course of blush intensity rise. G,,,, was defined as the peak grey level intensity and T,,ax as the time to peak intensity rise. By this approach, we anticipated that an adequate and prompt filling of myocardial capillaries with contrast agent (high G,,ax Within a

A comparison between (a) IR-GRE delayed enhancement image, (b) conventional SSFP end-systolic wall motion image, (c) conventional SSFP end-diastolic wall motion image with (d) MCDE infarct-enhanced image, (e) MCDE end-systolic image, and (f) MCDE end-diastolic image. Note that (d-f) are three of the 20 MCDE images acquired during a single breath-hold.

= —_ (3) A model-based approach was utilized to register the two datasets. Models were fit to contours delineating the left ventricular contours in the CTA and DEMR. The Iterative Closest Point method was applied to align them (please see Figure | and Figure 2 for segmentation and registration results). Data acquisition: A patient underwent DEMR (Sonata, SIE- MENS) using single shot TrueFISP inversion recovery technique approximately 20 minutes after intravenous 0.2 mmol/kg Gd- DTPA injection. The technical data were; FOV: 285-380 mm2,

Figure | (abstract P193) 1) To segment the vessels we employ a novel semi-automatic technique. Firstly, high radial symmetry positions (HSPs) in the image, which include the vascular structures’ centerlines and the obe shaped structures’ centerpoints, are detected by vote-based Hough-like approach. Secondly, shape tensors for each of these HSPs are calculated via Principal Component Analysis. As the tensors expose local shape properties, HSPs that do not have vessel ike structures are filtered out. Then, based on these final HSPs and their shape tensors, a graph is formulated where each HSP defines a unique graph node, and node to node connection strengths are calculated based on the shape tensor similarity metric. Next, the user is asked to identify 2 seed positions, for artery and vein respectively (the only manual part of the algorithm). Finally, the graph is partitioned into artery and vein regions based on these seed points by max-flow/minimal cut algorithm.

LDD, contractile reserve during low dose dobutamine; TEI, transmural extent of infarction; EDWT, end diastolic wall thickness; SWTur, Segmental wal thickening of the unenhanced rim; PPV, positive predictive value; NPV, negative predictive value; 95% confidence interval is presented between brackets Table | (abstract P195) Diagnostic performance of each viability index for the prediction of improvement in SWT

Table | (abstract P197) Univariated and multivariate regression analysis testing the association between maximal LVOT gradient

Quantification of the LVARA in the 5 chamber view

Unexpected areas of LGE: a&b, patchy LGE in a non-coronary artery distribution pattern (arrows), consistent with possible myocarditis — a, 4-chamber view showing sptal and apical LGE and b, short-axis view showing inferior LGE; c, subendocardial LGE in a circumflex coronary artery distribution (arrowheads), consistent with silent AMI in a patient with APS.

(a) 2D PSIR images of the left ventricle of a 55-year-old male patient with biopsy proven sarcoidosis showing multiple scattered hyper enhanced lesions (red arrows) within the left ventricular myocardium in short axis, 2 and 3 chamber views. (b) 3D PSIR reformatted images from the same patient demonstrate more numerous lesions in a subepicardial and mid myocardial location (red arrows) compared to 2D PSIR.

Table | (abstract P202) Description of seven patients with NICM in which CMR revealed a specific diagnosis and change management course

Table | (abstract P206) Proportion of lipid-rich necrotic core across normalized wall index

Volume rendered images in (A) 2 month-old with aortic coarctation (arrow) and (B) patient with bidirectional Glenn for DILV.

Volume rendered image in 18 month-old with pulmonary venolobar syndrome consisting of hypoplastic right pulmonary artery (RPA), partial anomalous pulmonary venous return (PAPVR) to superior vena cava (SVC) and inferior vena cava (IVC), and anomalous systemic pulmonary artery from the abdominal aorta to the right lower lobe.

(A) Flow velocity profiles in patient shown in Fig. la. The highest velocity is immediately distal to coarcation (arrow). (B) Flow-time curves were measured at four different locations in patient shown in Fig. 2 using advanced visualization software. With PC VIPR, measurements can be made in any arbirtrary location after the images have been acquired. Results: PC VIPR data sets were successfully acquired in all patients. MR angiograms were created using the average flow PC VIPR dataset (Figure |) to visualize cardiovascular structures. All anatomical structures visualized on CEMRA images were identified on PCVIPR images. This includes a 2 year-old with pulmonary venolobar syndrome with an anomalous pulmonary venous return to the SVC and IVC, in addition to an anomalous

Table | (abstract P216) Optimal imaging parameters for cardiac imaging with T2w-SPACE as acute myocarditis and acute myocardial infarction [2, 3]. 3D acquisitions can provide high resolution images of the whole heart, allowing arbitrary views of the organ, and reducing partial volume effect. Conventional 3D TSE improves SNR but not sampling efficiency or spatial resolution. We propose a novel TSE technique for T2w 3D imaging of the whole heart that provides higher spatial resolution and reduces scan time per slice compared to 2D DB-TSE.

Three cardiac views of a T2w-SPACE acquisition.

Partially suppressed blood may mimic myocardium in 2D DB-TSE. (a) VLA and (b) SAX views from T2w-SPACE, (c) VLA and (d) SAX views from 2D DB-TSE.

Figure | (abstract P217) Increased artifact severity with TSENSE (left) vs. TGRAPPA (right) under deep breathing.

Short axis views obtained with UFLARE Left) without susceptibility weighting and middle) with strong T, *-weighting). Right) corresponding T2 *-map calculated from multi t UFLARE data. The color represents T2 * values ranging from T, * = | ms (black) to T, * 2 85 ms (dark red). P219

Methods and results: |00 cardiac MRI SSFP cine studies were evaluated using a model-based software (Cardiac Image Modeller (CIM)) and a semiautomatic contouring system (MASS). Para- meters of cardiac function included left ventricular (LV) end diastolic mass (EDM), end diastolic volume (EDV), end systolic volume (ESV), and ejection fraction (EF). Peak ejection rate (PER) and peak filling rate (PFR) were also included. Reproducibility

data for intra-observer variability was performed for 20 consecutive cases.

Example of whole-heart and coronary-artery segmentation (surface rendering with different colour per component).

(a) Histogram of segmentation quality before and after correction (frequency = nr. of cases per quality category). (b) Average (blue bars), minimum and maximum (lines) correction times, for each of the categories of judged correction effort.

Figure | (abstract P221) Introduction: Displacement ENcoding with Stimulated Echoes (DENSE) is a method to acquire displacement maps of the myocardial motion. This can in turn be used to estimate strain which is an important measure of myocardial function. Previously, DENSE imaging has primarily been performed using multiple breath holds per slice, severely limiting its incorporation into the tight schedule of clinical routine. Strain maps are further desired throughout the left ventricle, not only in a single slice. A targeted approach, guided by results from perfusion and late

P222 Imaging parameters were: field of view 350 mm, slice thickness 8 mm, matrix 128 x 115, SENSE factor 2, TFE-factor 3, EPI- factor 7, TR 8.9 ms, TE 4.2 ms. Displacement encoding strength

enhancement studies is tempting, but the presence of contrast agent shortens the TI relaxation, which hampers stimulated echo acquisitions such as DENSE. We therefore propose a multi slice acquisition for strain measurements in a single breath hold. We compare this with conventional single slice DENSE acquired in separate breath holds.

An image with various levels of noise: left no noise; middle and right additive zero mean noise of 40 and 100 standard deviation, respectively.

Accuarcy of the proposed method, as measured by the average number of errors, for different noise levels and for a number of different TRes, when the whole image is used.

Accuracy of the proposed method with an ROI that encom- passes the heart, as measured by the average number of errors, for different noise levels and for a number of different TRes. Notice that the performance is similar to Figure 2, illustrating that there is no gain in selecting an ROI, thus demonstrating that the proposed method can be used in the whole image, without needing an ROI selection.

Accuracy of the seeded region growing method (SRGM), as measured by the average number of errors, for the different noise levels and for a number of different TRes. from three canines that were sedated and mechanically ventilated. ECG-gated and breath-held SSFP acquisitions were prescribed over the mid-ventricle following scout scans at various temporal resolutions (TRes:7—153 ms). Scan parameters: voxel size = 1.2 x 1.2 x 6 mm?; flip angle = 60°; TR/TE = 3.5/ 1.8 ms.

Accuracy of the proposed method in finding the center of the LV tested for different noise levels, and TRes. The accuracy is measured as the Euclidean distance from the ground truth provided by an observer. Notice that the y-axis is scaled at 10-? increments. Image processing: Each cine set was denoted as I(x, y, t), where x, y are pixel locations (NxM), and t denotes |,..., F phases. Each image was smoothed by convolution with 4 Gaussian filter (width = 10 and sigma = 1). The normalized cross correlation is computed between all images, giving an FxF matrix C(i, j), (i, j corresponds to image pairs). The i, | corresponding to the minimum of C are the ES, ED images, since they are uncorrelated in the heart region (high motion). Using the same principle, the LV chamber can be localized by finding the temporal variation of I(x, y, t), as matrix V (NxM). V was then denoised using a Gaussian and binarized (Otsu's method). The centroid of the largest connected component was chosen as the LV center. The proposed identification method was compared against LV blood volume segmentation methods. A seeded- region-growing method (SRGM) was chosen for segmentation instead of level-set-active-contours since it proved more robust to noise and the limited spatial resolution of the cine MRI images. Data analysis: Cine images with varying TRes were tested anc additive-white-Gaussian-noise was introduced to demonstrate robustness. The seed location for SRGM was manually set inside the ventricle for each image. Average number of frames away from the true ES, ED images (manually chosen) was used as the performance metric. For each noise level, 40 trials were ey ee Pn ay fe

Methods: Steady-state free precession dynamic gradient-echo images (Siemens MAGNETOM Sonata |.5 T scanner) were obtained in 45 patients with a wide range of LV size and function in short-axis views from LV base to apex as well as 3 long-axis planes rotated around the LV axis. Images were analyzed using the standard MOD technique and, independently, using the new volumetric approach implemented in prototype analysis software (TomTec Imaging Systems). This analysis is based on semi- automated reconstruction of endocardial surface from a combination of short- and long-axis CMR images, followed by direct quantification of LV volume confined within the endocar- dial surface. Both analysis techniques were used to obtain LV end- systolic and end-diastolic volumes (ESV, EDV) and ejection

MR T2* image showing cardiac short axis slice divided into regions of interest for T2* analysis.

Graph of myocardial iron concentration (mg Fe/g dry weight) vs. T2-star (ms). R* = 0.949.

Table | (abstract P225) Protocol parameters

Graph of myocardial iron concentration (mg Fe/g dry weight) vs R7* (Hz). R? = 0.825 P225

Table 3 (abstract P225) Correlation (r) in measurement: between pairs of the single breath-hold techniques

Table 2 (abstract P225) Correlation (r) of LV function para- meters between the standard and each of the single breath-hold techniques

Table 4 (abstract P225) Paired T-tests between the standard and each of the single breath-hold techniques (NS = no significant difference)

Deformable templates modeling LA (left) and SA (right): centerline (dashed) is defined by a few nodes, the template width (grey ribbon) may vary from node to node.

LV segmentation example for one patient: (from left to right, top to bottom): LA 4 chambers and 2 chambers; consecutive slices of SA volume from valve to apex.

Figure | (abstract P227) Motion corrected BLADE image reconstruction diagram (simplified).

Example of a dark-blood prepared TSE BLADE image showing conventional (left) and motion corrected (right) reconstruction for a case in which there are no respiratory image artifacts.

a) Average PWV for all subjects for both PCMR and tonometry (P > 0.05). b) Inter-scan coefficient of variation for each modality (P < 0.05).

P230 CTA (upper image) and FBI (lower image) of same patient after AA-ESG. The diastolic-triggered FBI image clearly depicts the lumen of the endovascular stent graft and its iliac sleeves in bright blood. The abdominal veins and large renal cysts are also well visualized.

distensability. Distensibility was calculated as the relative change in area divided by the central pulse pressure. Pulse wave velocity was measured by an ECG-gated, free breathing, spoiled gradient echo phase-encoded acquisition (TR = | RRinter- val, TE = 2.8 ms, 1.3 mm x 1.3 mm, slice thickness = 5 mm, temporal resolution = 10 ms) at the same locations as the aortic cine images. Three variations of the transit time method were used for calculation of pulse wave velocity (PWV). Arrival of the pulse wave was defined as a) the foot of the curve of mean blood velocity of the blood during the cardiac cycle b) the foot of the curve of total blood flow during the cardiac cycle c) maximum blood acceleration rate (maximum differential of the mean velocity) during the cardiac cycle (Figure 1).

PWV over the whole length of the aorta (r= 0.615, P<0.001) (Figure 2b) but also related to flow (r = 0.484 P< 0.05) and velocity (r = 0.372, P < 0.05) based MRI PWV measurements. Aortic pulse wave velocity calculated by all three MR methods yielded an inverse correlation with descending aortic distensi- bility (r = —0.521, P < 0.05) (Figure 2d) but PWVS and Al did not correlate with aortic distensibility

Figure 2 (abstract P234) were TIMI Ill. The MO quantification was determined by perfusion MRI and LGE techniques.

Table | (abstract P235) myocardial infarction. The purpose of this study was to compare 0.1 mmol/kg gadobenate dimeglumine (Gd-BOPTA, MultiHance, Bracco Imaging SpA, Milan, Italy) with 0.2 mmol/kg gadopentetate dimeglumine (Gd-DTPA, Magnevist, Bayer-Schering Pharma AG, Berlin, Germany) in cardiac MRI. njection, but thereafter undergoes a dual route of elimination with approximately 96% of the injected dose excreted renally and the remainder taken up by functioning hepatocytes and excreted n the bile. It has a weak reversible binding to albumin, which results in an increase of relaxation rate, compared to zadopentetate dimeglumine.

Purpose: The two-fold higher TI relaxivity of gadobenate dimeglumine compared with gadopentetate dimeglumine and can be used for imaging delayed enhancement in the assessment of

2D CVMRI images of geometry in A) 65 YO female with a small thick LV and B) 68 YO male with a larger, thinner LV, both with similar mean gradient (52 + 4 mmHg), BSA (2.1 + 1), and LVMI corrected for EDV (1.33 + 0.10) and with similar LVMI.

Table | (abstract P247) Types of aortic coarctation repair and the main complications repair: resection and end-to-end anastomosis (n = 94, 83%), subclavian flap repair (n = 62, 22%), aortic stenting (n = 43, 15%), interposition graft repair (n = 31, 11%: 22 dacron, 9 gelseal), balloon dilatation (n = 24, 9%), dacron patch repair (n = 19, 7%), and bypass graft (n = 8, 3%). Structural complica- tions were most marked in patch repair (15/19 aneurysmal dilatation within the patch, 2/19 suture line false aneurysms, 2/19 residual stenosis), and subclavian flap repair (38/62 aneurysmal dilatation within the flap, 2/62 suture line false aneurysms, and 2/ 62 residual stenosis). There were no direct structural complica- tions from resection and end-to-end anastomosis, but 57/94 had residual stenosis. Balloon dilatation had 17/24 residual stenosis and mild aneurysmal dilatation in 4/24. Aortic stenting showed less residual stenosis (6/43), with minimal structural complica- tions (no dissection, and only 1/43 displacement of the stent, 1/43 aneurysm). Reasonably good long term results were observed in graft repairs: interposition grafts (3/31 suture line false aneurysms, 4/31 residual stenosis at arch and graft anastomosis) and bypass grafts (none with residual stenosis, and 2/8 suture line false aneurysm). Main complications of the types of repair are in Table I.

2) Update of k-space should smoothly vary over time (as in KT- BLAST/SLAM) avoiding sudden transitions between k-space regions to result in smoother transition between frames. Introduction: Previously, we developed a sparsely distribute k-space-time sampling approach termed BRISK, Block Regional Interpolation Scheme for K-space [I]. This approach allowed a nominal acceleration factor of 4 with good quality and low artifact. Others have developed alternative k-space-time sampling schemes, such as KT-BLAST and SLAM [2, 3]. We note that even at high acceleration factors, KT-BLAST/SLAM allowed a smooth transition from frame to frame, while BRISK experienced ringing artifacts. From these considerations we isolated key features that contribute to a successful k-space-time sparse sampling scheme: |) Update of k-space should be rapid near the center and lower near the periphery (as in BRISK) to capture highly dynamic features.

From these design principles we developed a new sparse k-space- time sampling scheme, MACH, Multiple Acceleration Cycle Hierarchy. MACH incorporates a gradually changing rate over time, starting with the highest rate near the center of k-space and becoming progressively sparser towards the periphery, Figure | shows the k-space-time sampling scheme. In MACH, the progressively decreasing sampling rates are not confined to integer steps, thereby providing greater opportunity for a

SNR comparison between images from sliding CG-HYPR and conven- tional protocol.

Figure | (abstract P252) One typical example of the 6 volunteer studies. A) Comparison of the images from conventional method and sliding CG-HYPR. Blood signal change is observed in sliding CG-HYPR perfusion images, and comparable to the traditional images. B) Comparison of the left ventricle and myocardium signal changes vs. time.

Conclusion: Use of a disposable single element endo coil allows the assessment of LV function in the mouse with reasonable temporal, spatial resolution and imaging characteristics without the purchase of a dedicated small animal system.

Table | (abstract P256) Clinical and aortic wall characteristics of the study population Data are mean (SD) or n (percentage).

tno significant differences between corresponding sequences (Gd-DTPA vs Gadofosveset). Image quality: | = non diagnostic, 2 = diagnostic = good, 4 = excellent. Table | (abstract P261) Values are expressed as mean + standard deviation

Examples of resulting LV contours (red) and the golden standarc (yellow). We have used 28 black-blood multi-echo CMR scans from 10 patients. Each scan consisted of 4-8 echos at TE |.3—-17 ms and the isotropic image resolution was 0.94-1.45 mm. We are grateful to Children's Hospital Boston and Bioiatriki, Athens for providing us with clinical image data. Golden standard contours were defined manually on all images.

T2* from detected contours versus T2* from manual contours

Table | (abstract P266) Cardiac performance Methods: We investigated 28 patients with anterior AMI in whom primary percutaneous coronary intervention was per- formed successfully (mean age : 63.4 +- 8.4 year old, female 14.3%). From nuclear imaging, ejection fraction (LVEF), left ventricular volumes, and summed defect score (SDS) 2 weeks and 3 months post-MI were obtained by using rest gated

Change of strain over time at one day, one week, six weeks, three months and one year, for injured and remote myocardium, respectively.

Table | (abstract P272) Atrial Area (mean +/— SD) measured by CMR and TTE

37 year old female with 3 year evolution of SLE and myocarditis, with precordial pain and dispnea, left ventricle is dilated, low ejection fraction and fibrosis with patchy pattern, other sings of damage were mitral and aortic regurgitation, biatrial enlargement and pericardial effusion.

Table 2 (abstract P272) Fractional Area Change (mean +/— SD) measured by CMR and TTE

End-systolic adenosine images without (AD) and with LCX stenosis (SS) from the two controlled canine studies are shown in panels (A) and (B) respectively. The top row shows the AD and SS images without any windowing. The bottom row shows the same images but windowed the same t« increase the regional myocardial contrast corresponding to oxygenation changes. Red arrows on the windowed SS images illustrate the limits of the affected myocardial segments corresponding to the LCX artery.

Probability density plots for the two canine studies (A) and (B) are shown. For each case the empirical densities of the adenosine and stenosis myocardiz signals are plotted in red and green bars, respectively. Overlaid are the maximum likelihood estimated Student-T densities for each signal. Finally, th threshold is shown in blue at 143.3 for study A, 189.7 for study B, respectively. Note that m — s is the threshold computed from the Student- distribution fitted to the adenosine signal as described in the text. All intensities below this threshold are color-coded in hues of red, while intesitie above this threshold are color-coded in hues of yellow.

Color-coded end-systolic adenosine images without (AD) and with LCX stenosis (SS) from the two controlled canine studies are shown in panels (A) and (B), respectively. Only the myocardium was color coded. The segmentation masks were found using the algorithm in Tsaftaris, et al. ICIP 2008. The corresponding optimized colormaps are also shown. The yellow hues are for those values that are larger than the threshold as shown in Figure 2 and described in text. This method allows the observer to rapidly assess the presence of disease, as well as potentially identify the section of the coronary tree that is stenotic on the basis of BOLD MRI. and inter-observer variability. Figure 2 illustrates that student-T distribution closely approximates the intensity distribution of the myocardium. The motivation for choosing the m-s as the midpoint for the colormap is to highlight intensities below this value, which are expected to have low probability of occurrence in a normal healthy case. It is expected that regions supplied by stenotic arteries will be hypointense, and will thus shift the intensity distribution towards the left. Hence, in cases where stenosis is present, the occurrences of intensities lower than m-s are greater, and therefore the amount of red-colored pixels are greater. Note that the contiguous red-colored region corre-

Results: At rest, no significant difference in bias and standard deviation were observed with the identification process between the two shapes of the arterial input function (narrow and wide). Under-sampled curves introduce an increase of standard

Reference Figure | (abstract P278) Radial velocity graph at left ventricular base. | — isovolumetric contraction, 2 — rapid ejection, 3 — reduced ejection, 4 — isovolumetric relaxation, 5 — rapid filling, 6 — diastasis, 7 — atrial contraction, ED — end diastole, ES — end systole, a — peak systolic radial velocity, b — upward directed notch in early diastole, c — peak diastolic radial velocity, d — additional thrust of outward radial movement during rapid ventricular filling.

Example CMR image from a patient with an implanted dual chamber pacemaker. Although artifacts from both the pulse generator (top center) and the RV pacing lead are visible (arrow), image quality is sufficient for accurate analysis of myocardial function.

Figure | (abstract P282) CT-MRA dataset with excellent image quality without any venous enhancement or artefacts. Clear depiction of several atherosclerotic changes like occlusion of the superficial femoral artery on both sides. GFR < 30 ml/min) were not included. All exams were performed on a 3.0 T MR System (Magnetom Verio). Maximal contrast agent (CA) volume was 31.5 ml (1.5 ml testbolus, 15 ml/MRA technique). For both techniques the same monophasic CA injection protocol was used, 15 ml of standard 0.5 molar CA were injected at a flow rate of | ml/s directly followed 25 ml of saline also at | ml/s. Spatial resolution of the ct-MRA datasets is technically limited to a reconstructed voxel size of 1.0 x 1.0 x 1.3 mm? for the entire FoV. Acquired voxel sizes differ throughout the FoV due to a technique called “variable resolution” helping to increase resolution in the most distal part of the FoV. Step-by-step MRA reached a spatial resolution

Intra-operator reproducibility of bright blood (left) and black blood (right) acquisitions for all patients.

Intra-operator reproducibility of bright blood (left) and black blood (right) acquisitions for patients with T2* values <20 ms

Methods: Eighty healthy volunteers with informed consent (age: 59.5 + 13.9) were screened to exclude hypertension, elevated total cholesterol and cardiovascular disease. Using the ‘candy cane’ view of aorta, an axial plane through the ascending and descending aorta at the pulmonary artery level was prescribed and a through-plane velocity encoded PC cine imaging was acquired with VENC of 150 cm/s, TR/TE/FA = 98 ms/2.9 ms/15° and voxel spatial resolu- tion 1.3 x 2 x 6mm? ona 1.5 T MRI scanner. The distance traveled by the aortic pulse wave, AD, was determined as the distance along the central line between ascending and descending aorta in the ‘candy cane’ image. FLOW (Medis, Lunden, the Netherlands) was used to measure flow in both ascending and descending aorta

regions. For At assessment, a Matlab program was developed and flow upslope was calculated for each time point by taking flow values from neighboring time points using least squares means. The maximal upslope was selected as the largest upslope among all time series. We calculated PWV = AD/At and aortic compliance as AC = I/(p*PWV"), where blood density p = 1057 kg/m. Linear regression was used to determine the relationships between AC and age.

Simulation of magnetization signal intensity regrowth curves of blood (TI = 300 ms), diffuse fibrosis (T] = 350 ms) and normal myocardium (TI = 380 ms), after a 180° pulse. Zoomed view shows the difference in null times between blood, diffusely fibrosed myocardium and normal myocardium (ATI, and ATI,). Sb; is the blood signal intensity when normal myocardium is nulled, and Sb» is the blood signal intensity when diffusely fibrosed myocardium in nulled.

SNR and CNR of the blood and myocardium. SNRbd = SNR of the blood. SNRAv = Average SNR of the myocardium including the septum, anterior wall, lateral wall, and the inferior wall. SNRAv-sep is the average of SNR of the myocardium excluding the septum. SNR sep = SNR of the septal myocardium. CNRAv = Average CNR of the blood to myocardium including all 4 segments. CNRAv-sep = CNR of the blood to mhocar- dium excluding the septum. CNR sep = CNR of blood to septal myocardium.

Measured optimal inversion time (Tl). DCM patients have lower optimal Tl for myocardium and a smaller difference between the blood and myocardium Tl (ATI) lower.

Figure | (abstract P287) BOLD SI changes for suspected CAD patients and healthy volunteers. Methods: |! healthy volunteers (mean age 29 + 4 years) and || patients with suspected CAD (mean age 59 + 8 years) who all had abnormal perfusion results on SPECT were recruited for scanning. Using a clinical 1.5 T MRI system (MAGNETOM Avanto, Siemens Healthcare, Erlangen, Germany), SSFP BOLD- CMR was performed on a mid left ventricular short axis slice at baseline and during adenosine infusion (140 micro-g/kg). Typical scan parameters were: Field-of view 193 x 280 mm; matrix size 106 x 192; slice thickness 10 mm; Tp/T_ 5.8/2.9 ms; flip angle 90°; typical breath-hold duration 14 s. Images were analyzed using clinically validated software (cmr*”, Circle Cardiovascular Imaging Inc., Calgary, Canada). After automatic definition of the

*p < 0.05 compared to whole myocardium. Table | (abstract P287) Adenosine induced changes in BOLD SI and HR subendocardial 50% and the subepicardial 50% of the wall thickness, the BOLD signal intensity (SI) for each was analyzed and the relative change from baseline during adenosine infusion was calculated. Late enhancement was performed to ensure that patients did not have an infarct in the selected slice for analysis.

Figure | (abstract P288) FSD module diagram. Materials and methods: The FSD module was modified by using bipolar gradient rather than unipolar gradient before and after the center 180-RF pulse to address the artifactual issue resulted from imperfect frequency response (Figure 1). Nine healthy subjects (5 M 4 F) were imaged at 1.5 T (Avanto, Siemens) using a peripheral phased-array coil and spine coils. Phase-contrast flow scan was first performed right above the popliteal trifurcation to derive the arterial flow peak time T. bFFSP scans were then conducted: (1) in 2 subjects, non-FSD- bSSFP (bright-blood, BB) scans and FSD-bSSFP (dark-blood, DB) scans (FSD gradient strength G = 10 mT/m and duration d = 1.2 ms, applied on x-axis) with 5 ECG-trigger delays (0, 1/2 T, T, 3/2 T, 2 T), and arterial blood SNR for each leg was measured from

Typical blood velocity curve in one cardiac cycle.

MIP images after subtraction. AS the FSD gradient increased, more arteries were shown, but venous signal might have interfered with artery visualization.

Arterial blood SNR measured from 4 legs in BB (a) and DB scans.

edwt and eswt, end-diastolic and systolic wall thickness; CMR, cardiac magnetic resonance; LVM, left ventricular mass; SL, Sokolow-Lyon. *p < 0.05 and **p < 0.01. Table 2 (abstract P289) Correlation coefficients (r) of ECG and CMR variables in females

edwt and eswt, the sum of end-diastolic and end-systolic wall thickness; CMR, cardiac magnetic resonance; LVM/I, left ventricular mass and mass index; SL, Sokolow-Lyon. *p < 0.05, **p < 0.01 and ***p < 0.001. Table | (abstract P289) Correlation coefficients (r) between ECG and CMR parameters in males P290

The whole image was generated from a single mouse click (yellow arrow). Based on the local center line (in blue), a slab volume rendering (left), a cross section (top right), and a curvilinear reformat (bottom right) images were generated.

Figure | (abstract T1) ratios between fat, normal and enhanced myocardium were measured. One independent observer visually evaluated the diagnostic quality of all images.

2D CVMRI images of geometry in A) 65 YO femal with a small thick LV and B) 68 YO male with a larger, thinner LV, both with similar mean gradient (53 + 4 mmHg), BSA (2.1 + 1), and LVMI corrected for EDV (1.33 + 0.10) and with similar LVMI.

Bland-Altman plot of difference in ejection fraction (EF) and scatter plot using 2-slice versus single slice acquisition.

Table | (abstract T3) Coefficient of variation between scan techniques for single-slice and two-slice per breath-hold acquistion.

PARACEST and '°F MR imaging of a clot (Arrow) treated with fibrin- targeted nanoparticles.

Schematic representation of fibrin-targeted perfluorocarbon nanoparti- cles for MR molecular imaging.

References (127)

Rehwald WG, et al: J Cardiovasc Magn Reson 2007, 9:101-2.
Salerno M, et al: Proc 15 th ISMRM 2007.
Rehwald WG, et al: Proc 15 th 2007.
Koktzoglou I, et al: J Magn Reson Imaging 2006, 23:699-705. References
Kim WY, et al: Circulation 2002, 106(3):296-9.
Edelman RR, et al: Radiology 1991, 181(3):655-60.
Borrello JA, et al: Magn Reson Med 1990, 14(1):56-67.
Xiang QS: J Magn Reson Imaging 1996, 6(5):775-82.
Kellman P, et al: Magn Reson Med 2002, 47(2):372-383.
Meyer CH, et al: Magn Reson Med 1990, 15(2):287-304.
Botnar RM, et al: Circulation 2000, 102(102)21:2582-7. References
Becker, et al: Am J Cardiol 2007, 100:1671-6.
Bleeker, et al: Circulation 2006, 113:969-76.
Fornwalt, et al: J Am Soc Echocardiogr 2007, 20:1330-7.
O26 Papillary muscle involvement in acute and chronic myocardial infarction: an MRI study using multi-contrast delayed enhancement pulse sequence Yuesong Yang, Kim Connelly, Jay Detsky, Gideon Paul, Graham A Wright and Alexander J Dick Sunnybrook Health Sciences Centre, Toronto, ON, Canada References
DePasquale NP: Annu Rev Med 1971, 22:327.
Han Y, et al: JACC Img 2008, 1:294.
Bogun F: JACC 2008, 51:1794.
Detsky JS, et al: MRM 2007, 58:365.
Cerqueira MD: Circulation 2002, 105:539.
Axel L: Circulation 2004, 109:3145. References
Popovic ZB, et al: American journal of physiology 2006, 290: H1454-1459.
Firstenberg MS, et al: Journal of the American College of Cardiology 2000, 36:1664-1669.
Levine BD, et al: Circulation 1991, 84:1016-1023. References
Brun P, et al: Left ventricular flow propagation during early filling is related to wall relaxation: a color M-mode Doppler analysis. J Am Coll Cardiol 1992, 20:420-432.
Garcia MJ, et al: Color M-mode Doppler flow propaga- tion velocity is a preload insensitive index of left ventricular relaxation: animal and human validation. J Am Coll Cardiol 2000, 35:201-208.
Moller JE, et al: Circulation 2006, 114:438-444.
Paulus WJ, et al: Eur Heart J 2007.
Kasner M, et al: Circulation 2007, 116:637-647. References
Yu CM, Fung JWH and Zhang Q, et al: J Cardiovasc Electrophysiol 2005, 16:1117-1124.
Nelson GS, Curry CW and Wyman BT, et al: Circulation 2000, 101:2703-2709.
Yu CM, Chau E and Sanderson JE, et al: Circulation 2002, 105:438-445.
Haber I, Metaxas DN and Axel L: Med Image Anal 2000, 4:335-355. References
Magn Reson Med 2006, 55:242-249.
IEEE International Symposium on Bioimaging 2007, 988-991.
Proc Natl Acad Sci 2006, 103:1852-1857.
Brinegar, et al: Engineering in Medicine and Biology Conference 2008. References
Downie S, et al: JMRI 2007, 26:80-85.
Downie S, et al: AJP -Heart Circ Physiol 2007, 294:2112- 2120.
Miller J, et al: J Physiol 2005, 563:925-943. References
Kolipaka A, Charzimavroudis GP, White RD, Lieber ML and Setser RM: Relationship between the extent of non- viable myocardium and regional left ventricular function in chronic ischemic heart disease. J Cardiovasc Magn Reson 2005, 7:573-579.
Chan J, Khafagi F, Young AA, Cowan BR, Thompson C and Marwick TH: Impact of coronary revascularization and transmural extent of scar on regional left ventricular remodelling. Eur Heart J 2008, 29:1608-1617.
Roes SD, Kelle S, Kaandorp TA, Kokocinski T, Poldermans D, Lamb HJ, Boersma E, van der Wall EE, Fleck E, de Roos A, Nagel E and Bax JJ: Comparison of myocardial infarct size assessed with contrast-enhanced magnetic reso- nance imaging and left ventricular function and volumes to predict mortality in patients with with healed myocardial infarction. Am J Cardiol 2007, 100:930-936.
P53 Contribution of the base of the heart to MR measures of left ventricular mass, volumes and scar in patients with myocardial infarction Carissa G Fonseca, Tariq Balawi, James Sayre and Paul Finn UCLA, Los Angeles, CA, USA Journal of Cardiovascular Magnetic Resonance 2009, 11(Suppl 1):P53 References
Szczepaniak LS, et al: Circ Res 2007, 101:759-767.
Meer van der RW, et al: Diabetes 2007, 56:2849-2853.
Liu CY, et al: MRM 2007, 58:354-364. References
Grobner T: Gadolinium: a specific trigger for the development of nephrogenic fibrosing dermopathy and nephrogenic systemic fibrosis? Nephrol Dial Trans- plant 2006, 21(4):1104-1108.
Sadowski EA, Bennett LK and Chan MR, et al: Nephrogenic systemic fibrosis: risk factors and incidence estima- tion. Radiology 2007, 243(1):148-157. References
Gu TL, et al: AJNR 2005, 26(4):743-749.
Johnson KM, et al: PROC ISMRM 2008, 733.
Lum D, et al: Radiology 2007, 245(3):751-760.
Moftakhar R, et al: AJNR 2007, 28(9):1710-1714.
P127 Fast free-breathing planning in cardiac MR imaging Tamer A Basha, Monda L Shehata, Robert G Weiss and Nael F Osman Johns Hopkins University, Baltimore, MD, USA Journal of Cardiovascular Magnetic Resonance 2009, 11(Suppl 1):P127 References
Scherlag BJ, et al: J Electrocardiol 2006, 39(4 Suppl):S180-3.
Pokushalov J: Current Opin Cardiology 2008, 23:55-59.
Armour JA, et al: Anatomical Record 1997, 247:289-298.
P129 Volume tracking -a novel method for visualization and quantification of intracardiac blood flow from 3D time resolved phase contrast MRI Johannes Töger, Martin Ugander, Håkan Arheden and Einar Heiberg Cardiac MR Group, Dept. of Clinical Physiology, Lund, Sweden Journal of Cardiovascular Magnetic Resonance 2009, 11(Suppl 1):P129 References
Anderson1 LJ, et al: Eur Heart J 2001, 22:2171-2179.
Westwood M, et al: J Magn Reson Imag 2003, 18:33-39.
He T, et al: Magn Reson Med 2008 in press.
He T, et al: J Magn Reson Imaging 2007, 25:1205-1209.
Carpenter JP, et al: Abstract in SCMR 2009.
Ghugre NR, et al: Magn Reson Med 2006, 56:681-686.
He T, Smith GC, Carpenter J-P, Mohiaddin RH, Pennell DJ and Firmin DN: A phantom study of temperature- dependent MRI T2* measurement. J Cardiovasc Magn Reson 2009, 11(Suppl 1):P147.
Anderson1 LJ, et al: Eur Heart J 2001, 22:2171-2179.
Westwood M, et al: J Magn Reson Imag 2003, 18:33-39.
He T, et al: Magn Reson Med 2008, 60:350-356.
He T, et al: Magn Reson Med 2008 in press.
Ghugre NR, et al: Magn Reson Med 2006, 56:681-686. References
Aletras AH, Tilak GS and Natanzon A, et al: Retrospective determination of the area at r risk for reperfused acute myocardial infarction with T2-weighted car- diac magnetic resonance imaging: histopathological and displacement encoding with stimulated echoes (DENSE) functional validations. Circulation 2006, 113:1865-70.
Friedrich MG, Abdel-Aty H, Taylor A, Schulz-Menger J, Messroghli D and Dietz R: The Salvaged Area at Risk in Reperfused Acute Myocardial Infarction as Visualized by Cardiovascular Magnetic Resonance. J Am Coll Cardiol 2008, 51:1581-7.
Thiele H, Kappl MJ, Conradi S, Niebauer J, Hambrecht R and Schuler G: Reproducibility of chronic and acute infarct size measurement by delayed enhancement-mag- netic resonance imaging. J Am Coll Cardiol 2006, 47:1641-5. References
Haber, et al: Am J Physiol Heart Circ Physiol 2005, H1826- H1833.
Kim, et al: Radiology 2004, 230:862-871.
Spottiswoode, et al: Proc 14 th ISMRM, USA 2006.
McKenna, et al: Br Heart J 1994, 71:215-218.
Spottiswoode, et al: IEEE Trans Med Imag 2007, 26(1):15-30.
Zhong, et al: Proc 15 th ISMRM 2007. Reference
Moyssakis I, et al: Rheumatology (Oxford) 2005, 44(2):251-4.
P171 Underestimation of carotid plaque by ultrasound IMT and potential error in measurement of change in plaque burden: simultaneous comparison with
T MRI Mark Bowers 1 , Leanne Harmann 1 , Robert Prost 1 , Megan Bright 1 , Anil Doppalapudi 1 , Tayyab Mohyuddin 1 , John LaDisa 2 and Raymond A Migrino 1 1 Medical College of Wisconsin, Milwaukee, WI, USA
Marquette University, Milwaukee, WI, USA Journal of Cardiovascular Magnetic Resonance 2009, 11(Suppl 1):P171 References
Kim D and Kellman P: NMR Biomed 2007, 20:591-601.
Young AA, et al: Radiology 1993, 188:101-108.
Moore CC, et al: Radiology 1994, 190:765-769.
Lima JA, et al: JACC 1993, 21:1741-1751.
Yeon SB, et al: JACC 2001, 38:555-561. References
Gu TL, et al: AJNR 2005, 26(4):743-749.
Johnson KM, et al: Proc ISMRM 2008, 733.
Liu J, et al: IEEE TMI 2006, 25(2):148-157. P213 Dependence of arterial input function on position in the left ventricle and time in the cardiac cycle Nicholas S DuRocher, Elodie Breton, Sohae Chung, Daniel Kim and Leon Axel NYU Langone Medical Center -Center for Biomedical Imaging, New York, NY, USA Journal of Cardiovascular Magnetic Resonance 2009, 11(Suppl 1):P213 References
Monnig, et al: Europace 2005, 7:540-545.
Kottamp H, et al: JCEP 2005, 16(2):146-150.
Okada T, et al: JACC 2007, 49:1436-1442. P215 Investigation of T2-weighted and delayed enhancement magnetic resonance imaging in reperfused acute myocardial infarction in a porcine model Seong Hoon Choi 1 , Joon-Won Kang 1 , References
Breuer FA, et al: Dynamic autocalibrated parallel imaging using temporal GRAPPA (TGRAPPA). Magn Reson Med 2005, 53:981-985.
Griswold MA, et al: Field-of-view limitations in parallel imaging. Magn Reson Med 2004, 52:1118-1126.
Kellman P, et al: Adaptive sensitivity encoding incor- porating temporal filtering (TSENSE). Magn Reson Med 2001, 45:846-852. References
He T, et al: Magn Reson Med 2008, 60:350.
Friedrich MG, et al: Circulation 2003, 108:2219-2223.
Niendorf T: Magn Reson Med 1999, 41:1189-1198.
Messroghli DR, et al: Assessment of myocardial T2* from pixel-by-pixel maps on a clinical MR system. J Cardiovasc Magn Reson 2008, 10(Suppl 1):A240, SCMR Proceedings 2008.
P219 Evaluation of model-based contouring versus 2D segmentation for cardiac mass and volumes Andrew S Chi 1 , Cuilian Miao 1 , Chia-Ying Liu 1 , Patricia A Cleary 2 , Joao AC Lima 1 and David A Bluemke 1 1 Johns Hopkins, Baltimore, MD, USA
Yong Kim W, et al: New Engl J Med 2001, 345:1863-1869.
Ecabert O, et al: Proc SPIE 2007, 6512: 65120G.
Peters J, et al: MICCAI 2007, 402-410, Part II, LNCS 4792.
Sonnemans J: J Cardiovas Magn Reson 2008, 10(Suppl 1): Proceedings SCMR 2008..
P221 Multi slice DENSE in a single breath hold Andreas Sigfridsson 1 , Henrik Haraldsson 2 , Tino Ebbers 2 , Shinichi Takase 1 and Hajime Sakuma 1 1 Mie University, Tsu, Mie, Japan
Stuber, et al: MAGMA 1999, 9:85-91.
Lin, et al: MRM 2008, 60:8-13.
Gilson, et al: MRM 2004, 51:744-752. References
Tanner MA, He T and Westwood MA, et al: Haematologica 2006, 91:11388-1391.
Ghugre NR, Enriquez CM and Gonzalez I, et al: Magn Reson Med 2006, 56:681-686.
He T, Gatehouse PD and Kirk P, et al: Magn Reson Med 2008, 60:350-356.
Collins W and Taylor WH: Ann Clin Biochem 1987, 24:483- 387.
He T, Carpenter JP and Jongirasiri S, et al: Can we measure iron overload in the heart using in vivo T2* MRI?. Abstract SCMR 2008. References
Viallon M, et al: Proceedings ISMRM 2008, 16: 1008.
Pipe J, et al: Magn Reson Med 1999, 42(5):963-969.
Ledesma MJ, et al: J Magn Reson Imaging 2007, 26:184-190.
Chefd'hotel C, et al: Proc IEEE ISBI 2002. References
Farrar DJ, et al: Circulation 1991, 83(5):1754-63.
Fielden SW, et al: J Magn Reson Imaging 2008, 27(6):1382-7.
Bland JM and Altman DG: Lancet 1986, 1:307-10. References
Doyle M, Walsh EG, Blackwell GG and Pohost GM: Block Regional Interpolation Scheme for K-space (BRISK): A Rapid Cardiac Imaging Technique. Magn Reson Med 1995, 33:163-170.
Kozerke S, Tsao J, Razavi R and Boesiger P: Accelerating cardiac cine 3D imaging using k-t BLAST. Magn Reson Med 2004, 52(1):19-26.
Rehwald WG, Kim RJ, Simonetti OP, Laub G and Judd RM: Theory of high-speed MR imaging of the human heart with the selective line acquisition mode. Radiology 2001, 220(2):540-547.
Petersen SE and Jung BA, et al: Myocardial Tissue Phase Mapping with Cine Phase-Contrast MR Imaging: Regional Wall Motion Analysis in Healthy Volun- teers. Radiology 2006, 238(3):816-826.
P279 Cartesian acquisition with PR-like sampling: applications to 3D contrast-enhanced MRA Stephen J Riederer and Clifton R Haider Mayo Clinic, Rochester, MN, USA Journal of Cardiovascular Magnetic Resonance 2009, 11(Suppl 1):P279

Visualizing regional myocardial oxygenation changes with statistically optimal colormaps

Sign up for access to the world's latest research

Abstract

Related papers

Related topics

References (127)