Towards a Five-Minute Comprehensive Cardiac MR Examination Using Highly Accelerated Parallel Imaging with a 32-Element Coil Array: Feasibility and Initial Comparative Evaluation

Abstract

PURPOSE

To evaluate the feasibility and perform initial comparative evaluations of a 5-min comprehensive whole-heart MRI protocol with four image acquisition types: perfusion (PERF), function (CINE), coronary artery imaging (CAI) and late gadolinium enhancement (LGE).

MATERIALS AND METHODS

This study protocol was HIPAA compliant and Institutional Review Board approved. A 5-min comprehensive whole heart MRI examination protocol (Accelerated) using 6- to 8-fold-accelerated volumetric parallel imaging was incorporated into and compared with a standard 2D clinical routine protocol (Standard). Following informed consent, 20 patients were imaged with both protocols. Datasets were reviewed for image quality using a 5 point Likert scale (0=non-diagnostic, 4=excellent) in blinded fashion by two readers.

RESULTS

Good image quality with full whole heart coverage was achieved using the accelerated protocol, particularly for CAI, although significant degradations in quality, as compared with traditional lengthy examinations, were observed for the other image types. Mean total scan time was significantly lower for Accelerated as compared to Standard protocols (28.99±4.59 min vs. 1.82±0.05 min, p<0.05). Overall image quality for Standard vs. Accelerated protocol was 3.67 ± 0.29 vs. 1.5 ± 0.51 (p<0.005) for PERF, 3.48 ± 0.64 vs. 2.6 ± 0.68 (p<0.005) for CINE, 2.35 ± 1.01 vs. 2.48 ± 0.68 (p=0.75) for CAI and 3.67 ± 0.42 vs. 2.67 ± 0.84 (p<0.005) for LGE. Diagnostic image quality for Standard vs. Accelerated protocols was 20/20 (100%) vs. 10/20 (50%) for PERF, 20/20 (100%) vs. 18/20 (90%) for CINE, 18/20 (90%) vs. 18/20 (90%) for CAI, and 20/20 (100%) vs. 18/20 (90%) for LGE.

CONCLUSION

This study demonstrates technical feasibility and promising image quality of 5-min comprehensive whole heart cardiac examinations, with simplified scan prescription and high spatial and temporal resolution enabled by highly parallel imaging technology. The study also highlights technical hurdles that remain to be addressed. Although image quality remained diagnostic for most scan types, the reduced image quality of PERF, CINE and LGE scans in the accelerated protocol remain a concern.

INTRODUCTION

Cardiovascular magnetic resonance (CMR) is challenging due to the competing demands of spatial/temporal resolution, robustness against physiological motion, signal-to-noise ratio (SNR) and total examination time. The idea of a comprehensive MR examination to provide a multifaceted cardiac evaluation has motivated extensive research (1–3). Four basic scan types can provide comprehensive assessment of ischemic heart disease (4): cardiac function including regional cardiac wall motion (CINE), myocardial viability via late gadolinium enhancement (LGE), myocardial perfusion imaging (PERF), and angiography of the coronary arteries (CAI). A typical CMR examination for comprehensive evaluation of ischemic heart disease usually takes 30 minutes to 1 hour (1–3). Major portions of the total examination time are devoted to anatomic scan planning and coverage of target cardiac anatomy. To this end, multiple breath–holds (BH) and free-breathing with navigator techniques (NAV) are commonly required to examine the entire heart. Substantial user expertise and interaction is required, and resultant image quality tends to be operator dependent. As a result, specialized expertise is required which, combined with scan costs and limited throughput, are hurdles to routine utilization of comprehensive CMR.

Rapid volumetric imaging can provide comprehensive information, and avoid potential slice misregistration between separate acquisitions. However, volumetric scans are currently limited in either spatial/temporal resolution or volumetric coverage. Nevertheless, these techniques are conceptually appealing as strategies to reduce total scan time, minimize operator dependence, enhance patient comfort, improve image quality and increase throughput. Whole heart (WH) CAI was first described by Weber et al. (5), followed by extensive work of Sakuma et al. (6) and others (7–11). The majority of these approaches used free breathing with navigator gating, which requires operator input and extended imaging time. WH approaches have also been reported for other cardiac MR scan types, such as PERF (12–13), CINE (14–16) and LGE (17–18). Due to particularly stringent constraints on temporal resolution, development of 3D/4D WH perfusion sequences has been limited, and multi-slice techniques remain the current clinical standard.

Partial combinations of accelerated imaging methods have been demonstrated previously (1–3,19–20). The feasibility of a combination of WH rest PERF, CINE, CAI and LGE in a short comprehensive examination utilizing only a few BHs has been established in prior work (21–22). In the current study, highly accelerated 3D imaging techniques for comprehensive cardiac imaging in a total examination time of 5 minutes were developed and assembled into a single streamlined protocol. This protocol (Accelerated) was evaluated and compared to a standard clinical protocol (Standard) to assess its feasibility in healthy subjects and cardiac patients, and to characterize the remaining technical hurdles to be overcome for routine clinical implementation.

METHODS AND MATERIALS

Study Population

This study protocol was HIPAA compliant and approved by our institutional review board; written consent was obtained from all subjects. 20 consecutive patients (14 male, mean age 60.25±12.3 yrs; range 37–81 yrs) scheduled to undergo imaging for evaluation of the etiology (e.g. ischemic vs. non-ischemic) of new onset cardiomyopathy were recruited.

Imaging Protocol

All CMR examinations were performed on a 1.5-T MRI system (Avanto, Siemens Healthcare, Erlangen, Germany) equipped with a 32-element cardiac coil array (In vivo Corp., Gainesville Florida). Figure 1 outlines the tailored WH imaging protocol used for the 5-minute comprehensive cardiac examination, using parallel imaging (23–25) with 6–8 fold acceleration. This protocol includes assessment of cardiac function, viability, and coronary anatomy in the context of new onset cardiomyopathy. Typical imaging parameters are summarized in Table 1. Our initial goal was to perform rapid volumetric imaging using transverse orientation data acquisition with isotropic data resolution to in order to allow for reformats in any orientation, including short-axis views. While this still remains as our ultimate goal, limited through-plane resolution and the need for high resolution short axis planes for interpretation has led us to modify our current protocol to allow for acquisition in the short-axis imaging planes. The Accelerated protocol can be performed alone with an earlier contrast loading dose injection 5–10 minutes before image acquisition, but in our study it was incorporated into the Standard protocol prior to 2D LGE (see below). The Accelerated protocol was organized as follows:

Figure 1.

Basic scheme of the five-minute tailored protocol used for comprehensive, accelerated WH cardiac exams.

Table 1.

Typical imaging parameters used for accelerated whole heart imaging

Type	PERF		CINE		LGE		CAI
	Stand.	Accel.	Stand.	Accel.	Stand.	Accel.	Stand.	Accel.
Sequence	2D SSFP	3D FLASH	2D SSFP	3D SSFP	2D SSFP	3D SSFP	3D SSFP	3D SSFP
PAT	TGRAPPA	TGRAPPA	TGRAPPA	TGRAPPA	GRAPPA	GRAPPA	GRAPPA	GRAPPA
View	SAX	SAX	SAX	SAX	SAX	SAX	TRA	TRA
Matrix	122×192×3	76×128×10	139×192×13	109×176×20	101×192×13	144×144×20	192×192×110	192×192×72
FOV	287×340	340×340	254×280	340×340	255×340	340×340	320×320	340×340
Voxel Size	2.3×1.8×8	4.4×2.6×8	1.8×1.5×8	3.1×1.9×5	2.5×1.8×8	2.4×2.4×6	1.7×1.7×2	1.8×1.8×2
TI/TR/TE	120/220/1.1	130/270/0.9	-/39/1.2	-/45/1.1	~250/3.0/1.5	~250/3.0/1.5	-/2.6/1.3	-/2.4/1.2
Accel. R	2	4 × 2	2	4 × 2	2	3 × 2	2	4 × 2
BW	1370	1392	930	915	1350	500	590	500
FA	70	10	70	70	70	70	70	70
Temp. Resolution	220ms	270ms	41.9ms	44.8ms	152ms	144ms	94ms	115ms
Scan Time	48s	48s	7~14BHs	1BH	Multi BHs	1BH	8~13min	1BH

• Step 1: A gadolinium-based contrast agent was injected in conjunction with multi-slice standard 2D perfusion imaging as part of the Standard protocol, approximately 5–10 minutes before the starting point of the Accelerated protocol acquisition. This provided a loading dose to ensure sufficient enhancement by the time of the LGE scan.

• Step 2: A rapid low spatial resolution scout scan was acquired along all three main axes, using a 2D balanced steady-state free precession (SSFP) technique.

• Step 3: To acquire first-pass perfusion images, BH WH PERF was applied for short-axis views, using a 3D saturation-recovery T1-weighted gradient echo (FLASH) sequence. Parallel imaging with 2D-TGRAPPA(26), with acceleration factor R=8 (4 (ky) × 2 (kz)), was used to increase imaging speed. An additional 0.075 mmol/kg Gd-DTPA contrast agent bolus injection was administered. Signal intensity was tracked over 60 heartbeats with a 1-RR interval temporal resolution, with the subject holding his or her breath for as long as comfortable, followed by shallow breathing.

• Step 4: Short axis view BH WH CINE imaging was applied using an ECG-triggered 3D SSFP sequence together with 2D-TGRAPPA reconstruction with acceleration factor R= 8 (4 (ky) × 2 (kz)), which resulted in breath-hold durations of ~20 seconds).

• Step 5: BH WH CAI was performed using a T2-weighted and fat saturation prepared 3D SSFP sequence. To increase the acquisition efficiency, a single breath-hold (SBH) sequence was developed to acquire both the coil sensitivity and the accelerated scan data in two separate cardiac phases (early systole and mid diastole, respectively)(27). 2D-GRAPPA reconstruction (28) was used to reconstruct the accelerated data (R=8 (4(ky) × 2(kz))). A slab thickness of 120 mm was employed, encompassing 60 partitions with slice oversampling 20%.

• Step 6: BH WH 3D LGE oriented in the short-axis view, with a SBH approach, was performed 10–15 minutes after the administration of the first contrast agent dose, using an inversion-recovery 3D SSFP sequence with 2D GRAPPA (R=6 (3 (ky) x2 (kz))). A Look-Locker technique was applied prior to the 3D LGE scan, in order to adjust the evolution time (TI) so that normal myocardium was nulled in the 3D LGE san.

All scans were synchronized with the cardiac cycle, using ECG gating and performed in end-expiration. Although the accelerated protocol would ultimately be performed alone following the steps outlined above, for comparison purposes it was concatenated into a Standard routine clinical protocol prior to 2D LGE, as shown in Figure 2. Two contrast injections were used for Standard and Accelerated perfusion acquisitions, each with 0.075mmol/kg gadolinium-DTPA (Berlex Magnevist, Schering AG) at 5ml/s followed by a saline flush (20ml at 5ml/s), leading to a cumulative 0.15 mmol/kg dose of gadolinium contrast to ensure sufficient contrast administration for LGE acquisitions. The interval between the first contrast injection and 3D LGE was set for approximately 10–15 minutes, to allow contrast agent washout after the first injection and to allow for sufficient delay after the second contrast injection to ensure optimal conditions for LGE acquisitions. For comparison with Accelerated 3D CAI, a standard free-breathing and navigator-gated 3D SSFP CAI study was included at the end of the protocol. Note that the LGE scan of the Accelerated protocol (“3D LGE” in figure 2) was followed immediately by the LGE scan of the Standard protocol (“2D LGE” in figure 2). Acquisition time for each component of Standard and Accelerated protocols was recorded. Effective overall table times were computed for both protocols. WH CAI images were reformatted using CoronaViz software (Siemens Corporate Research, Princeton, NJ, USA) to project multi-branch vessels with their surroundings onto a single image.

Figure 2.

Basic overview of the exam, which combines the Standard protocol and the Accelerated protocol. Blocks labeled “Extra” indicate quiescent times in the standard protocol which may be used for alternative scan types, such as phase-contrast (PC).

Image evaluation and statistical analysis

Accelerated volumetric datasets were reformatted to match the image planes of the Standard clinical examination. Images obtained from both exams were anonymized and randomized for blinded review by a cardiologist and a radiologist (with 8- and 5-years’ cardiac imaging experience, respectively) for assessment of image quality, diagnostic quality and image artifacts. Images were reviewed on a 3D workstation capable of multi-planar reformation, maximum intensity projection and volume rendering (syngo MultiModality Workplace; Siemens Healthcare, Erlangen, Germany). Image quality was graded on a 5 point scale: 0=non-diagnostic, 1=poor, 2=fair, 3=good, 4=excellent. Scores of 2 and higher were considered to have diagnostic image quality. Artifact severity was graded on a 5 point scale: 0=non-diagnostic, 1=severely limiting, 2=mildly limiting, 3=not limiting, 4=no artifact.

All results are summarized as mean ± standard deviation (SD). The results of Standard and Accelerated images were compared using the Wilcoxon matched-pairs signed ranked test for statistical analysis. SPSS (v.13, Chicago IL) was used for all tests. Statistical significance was defined as p < 0.05.

RESULTS

All subjects successfully completed the full protocol. For all 20 patients, mean total imaging scan time, including both standard and accelerated protocols, was 62.5 ±12.5 minutes. Image quality, diagnostic quality, and artifacts scores are summarized in Table 2. The average scan time of each imaging type for accelerated and standard protocols are summarized and compared in Figure 3. The scan time for 2D and 3D PERF is the same, since the same number of heart beats was used for both methods, but 10 partitions were acquired without gaps in 3D PERF, while only 3 slices with gaps were able to be acquired in 2D PERF. Mean total scan time (in minutes) was significantly lower for Accelerated as compared to Standard protocols (1.82±0.05 vs. 28.99±4.59, p<0.05). Three out of 20 cases (15%) demonstrated evidence of myocardial fibrosis in both 2D LGE and 3D LGE images.

Table 2.

Image quality, diagnostic quality and artifact scores

Imaging Type	Imaging Quality		Diagnostic Quality		Overall Artifact
	Stand.	Accel.	Stand.	Accel.	Stand.	Accel.
CINE	3.48±0.64*	2.6±0.68*	20/20 (100%)	18/20 (90%)	0.60±0.7*	2.07±0.82*
CAI	2.35±1.01	2.48±0.68	18/20 (90%)	18/20 (90%)	1.7±1.11	1.77±0.75
LGE	3.67±0.42*	2.67±0.8*	20/20 (100%)	18/20 (90%)	0.72±0.43*	1.09±1.01*
PERF	3.67±0.29	2.15±0.51	20/20 (100%)	10/20 (50%)	0.43±0.36	2.9±0.47

Notes: 1. Image quality was graded on a 5 point scale: 0=non-diagnostic, 1=poor, 2=fair, 3=good, 4=excellent,

2: Score ≥2 for diagnostic image quality;

3: Artifact severity was graded on a 5 point scale: 0=non-diagnostic, 1=severely limiting, 2=mildly limiting, 3=not limiting, 4=no artifact.

^*p value <0.05 and indicates statistically significant

Figure 3.

Scan time comparison: Standard vs. Accelerated protocols. There is no difference for PERF since this is dependent on transit of contrast and the 2D and 3D PERF were acquired over the same number of heartbeats. Please note that only SAX scan time was calculated for 2D LGE and compared to 3D LGE.

Perfusion (PERF)

Perfusion scans have the most stringent spatiotemporal constraints of the scan types in the comprehensive examination, and they constituted the greatest challenge for our Accelerated protocol. For Standard and Accelerated PERF acquisitions, overall image quality scores were 3.67 ± 0.29 and 1.5 ± 0.51, respectively, with p<0.05. Artifact scores were 0.43 ± 0.36 and 2.9 ± 0.47, p<0.05 respectively. The proportion of cases with diagnostic image quality was 20/20 (100%) for Standard and 10/20 (50%) for Accelerated PERF acquisitions, respectively. The scan times for Standard and Accelerated protocols were 49.4±5.4s and 48.5±4.6s, respectively (p=0.12), both corresponding to 2% of the total scan time, with the Accelerated protocol yielding greater anatomic coverage at a slightly smaller in-plane spatial resolution than the Standard protocol (Table 1). A representative example of 2D and 3D PERF is shown in Figure 4. Sixty distinct time points were obtained during an initial breath-hold followed by gentle breathing. The passage of the contrast agent bolus may clearly be discerned in the comprehensive dataset. For first-pass perfusion imaging, our study used an acquisition window of ~270 ms, in-plane resolution of 4.4×2.6 mm² and slice thickness of 8 mm, matching suggested requirements (29–30).

Figure 4.

A) Three short-axis slices acquired with a standard 2D PERF sequence, and B) Three of ten acquired slices/partitions from corresponding 3D WH PERF images, depicting the left ventricle at apical, mid and basal levels in a 39-year old Female. Four phases of the contrast bolus are depicted (left to right): pre-contrast, RV enhancement, LV enhancement and myocardium enhancement. The entire LV can be covered by ten partition slices without gaps in 3D WH PERF, while only 3 slices with gaps could be acquired in 2D PERF.

Cardiac function (CINE)

For Standard (2D CINE) and Accelerated (3D CINE) protocols, overall image quality scores were 3.48 ± 0.64 and 2.6 ± 0.68, p<0.05, and artifact scores were 0.6±0.70 and 2.07±0.82, respectively (p<0.05). The number of cases with diagnostic image quality was 20/20 (100%) for Standard and 18/20 (90%) for Accelerated CINE acquisitions. The scan times for Standard and Accelerated protocols were 8.8±1.9 minutes and 22.7±3.1 seconds (p<0.05), corresponding to 27% and 1% of the total scan time, respectively. A representative example of 2D and 3D CINE is shown in Figure 5. For wall motion evaluation, 3D CINE had a temporal resolution of <45 ms, ~18 phases per cardiac cycle, in-plane spatial resolution of 3.1 x1.9 mm², and slice thickness of 5mm, meeting requirements recommended in the literature (15).

Figure 5.

Representative standard and accelerated cine imaging in a 38-year female patient with no wall motion abnormalities. A) 2D CINE (6 out of 13 slices) covering the WH volume from apex to base (7 BHs, GRAPPA factor 2) with slice thickness 8mm and 1.6mm gap between adjacent slices; B) 3D CINE (12 out of 20 slices) covering the WH volume from apex to base (TGRAPPA with acceleration factor 4×2=8, 20-second single BH, 19 phases, and temporal resolution <50 ms), without gaps between adjacent slices.

Coronary artery imaging (CAI)

For Standard (3D NAV CAI) and Accelerated (3D SBH CAI) protocols, overall image quality scores were 2.35 ± 1.01 and 2.48 ± 0.68 (p=0.64) and artifact scores were 1.7 ± 1.11 and 1.77±0.75 (p=0.75), respectively. The number of cases with diagnostic image quality was 18/20 (90%) for Standard and 18/20 (90%) for Accelerated acquisitions. 3D NAV CAI failed in 2/20 patients, due to an irregular breathing pattern or drift of the diaphragm position during acquisition, while the accelerated CAI was successful in all cases. Representative examples of 3D NAV and 3D SBH CAI are shown in Figure 6. In our studies, the navigator efficiency was around 30%~35% in 3D NAV CAI, the mean scan times for standard and accelerated data acquisitions were 11.6±2.89 minutes and 25.6±2.8 seconds and corresponded to 36% and 1% of the total scan time, respectively.

Figure 6.

Representative CAI of the LCX and LAD obtained from two different subjects, in columns A and B, respectively. The top panel was obtained in one single BH using the accelerated protocol, and the bottom panel was obtained in 8 minutes using the standard protocol.

Late Gadolinium Enhancement Imaging (LGE)

For Standard (2D LGE) and Accelerated (3D LGE) protocols, overall image quality scores were 3.67 ± 0.42 and 2.67 ± 0.84 (p<0.05) and artifact scores were 0.72 ± 0.43 and 1.09 ± 1.01 (p<0.05), respectively. The number of cases with diagnostic image quality was 20/20 (100%) for Standard and 18/20 (90%) for Accelerated. A representative example of 2D and 3D LGE is shown in Figure 7. The mean scan times (SAX only) for Standard and Accelerated protocols were 2.58±2.11 minutes and 11.8±3.6s (p<0.05), respectively.

Figure 7.

A) Representative short-axis, long-axis and 4-chamber views of LGE imaging performed in a patient, using 2D segmented IR-prepared gradient echo imaging covering a single slice per BH. B) 9 out of 20 short axis views obtained in the same subject, using a six-fold accelerated single BH 3D segmented IR-prepared gradient echo acquisition with apex-to-base coverage.

DISCUSSION

Our initial experience suggests that comprehensive CMR examination of 3D rest PERF, CINE, LGE and CAI in 5 minutes is feasible in a clinical population using highly accelerated parallel imaging, although the reduced image quality of PERF, CINE and LGE from the Accelerated protocol remains a primary concern. Even though image quality was lower with the accelerated technique, there were still a high proportion of diagnostic images for CINE and LGE sufficient for clinical purposes. The approach of volumetric imaging of the whole heart is a potentially valuable tool for clinical imaging and would significantly alter the way in which cardiac MR exams could be performed in the future. Significant hurdles are, however, still present in some of the exams, particularly in whole heart perfusion imaging, as acceleration factors have to be improved further for results equivalent to standard imaging of individual slices.

3D volumetric imaging using highly accelerated parallel imaging allows whole heart anatomic coverage within a breath hold, which dramatically reduces demands on scanner operators and patients. One added advantage of 3D volumetric imaging is that it enables acquisition of all data at the same time point of the contrast agent kinetics in LGE imaging, which helps ensure uniform suppression of normal myocardium for all images. This comprehensive volumetric imaging protocol is desirable from clinical and economic perspectives, not only to reduce total scan time but also to minimize operator dependence, improve patient experience, and increase throughput.

Each breath hold is restricted to around 20 seconds, which was tolerated by our patient population. The total examination time for the Accelerated protocol, including suitable rest periods and injection time, was approximately 5 minutes, comparing favorably with a routine standard protocol with total examination time 62.5 ±12.5 minutes. 8-fold accelerations approach the maximum practical capability of the 32-channel coil array used in this study. 6-fold acceleration was used for LGE, due to the lower baseline SNR of LGE imaging. Accelerations as high as a factor of 12 have been reported using techniques such as k-t-BLAST and k-t-SENSE (31–32), which use additional training data to unfold more highly under-sampled datasets in k-t space. Only TGRAPPA and TSENSE are commercially available dynamic parallel imaging methods. Therefore, TGRAPPA was chosen for dynamic imaging in this study, which increased acquisition efficiency, as no additional reference scans/lines were required. Furthermore, the 32-element cardiac array offered increased SNR and lower g-factor, which reduced the noise amplification inherent to parallel imaging and enabled higher acceleration factors (33).

A separate calibration scan can increase the acceleration efficiency and fit all the required reference k-space lines into a short cardiac period in GRAPPA or SENSE; however, it requires an additional breath-hold. The extra time required for separate calibration scans, and motion occurring between the time of the calibration and the accelerated scan, can result in artifacts. The calibration scan is a low resolution scan which is less sensitive to cardiac motion than respiratory motion. In this study, for non-dynamic imaging, such as 3D CAI and 3D LGE, both the coil sensitivity and the accelerated scan data were acquired in two separate cardiac phases (early systole and mid diastole, respectively) within the same breath-hold to increase acquisition efficiency (27). This approach also decreases vulnerability to misregistration artifacts resulting from subject and coil array motion between the time of a separate RF calibration scan and that of accelerated data acquisition. Use of an acceleration factor of 8 (4 × 2) in CAI acquisition allowed nearly isotropic spatial resolution to be obtained in a breath-hold duration of 20~25 seconds, which, in turn, enabled simple axial scanning without scout scanning or careful view tailoring. Views in traditional cardiac imaging planes, or in the planes of the left and right coronary arteries, could then be obtained by retrospective reformatting and/or segmentation(34).

There are several limitations of this study, which serve to concretely characterize future requirements for routine clinical implementation of 5-minute comprehensive cardiac MR examinations. First, stress perfusion imaging has not been included in this initial screening protocol, potentially limiting identification of reversible ischemia in ischemic cardiomyopathy. Stress perfusion can simply replace the rest perfusion without increasing the total scan time (35) or it can be added at various time points in the protocol if both rest and stress perfusion studies are needed. Consequently, total examination time will be increased by at least several minutes to allow for the additional stress agent administration, contrast injection, and scan time. Second, our ultimate goal of simplification via isotropic image acquisition and straightforward axial acquisition, as opposed to targeted double-oblique volumes, was not completely achieved, due to the maximum parallel imaging acceleration factor that we have found practical so far (4 in ky and 2 in kz, for a total of 8). For example, the through-plane spatial resolution for perfusion scans (10 mm in this study) needs to be improved in order to allow effectively reformatting from axial acquisitions to other typical cardiac planes of interest. Finally, and perhaps most importantly, our initial evaluations show significantly reduced image quality in 3D PERF, CINE and LGE for the Accelerated as compared to the Standard protocol. In addition to the SNR reduction associated with highly parallel imaging, technical issues and patient factors might contribute to image quality degradation. These factors include increased sensitivity of 3D imaging to imperfect breath-holding or to arrhythmias compared to traditional short 2D breath-hold imaging acquisitions, each covering a single slice only. One additional factor which might contribute to the low SNR PERF results is that the gradient echo sequence used provides lower SNR than the SSFP sequence used in the standard protocol. Contrast dose is another factor for future optimized high acceleration imaging. TI time needs be optimized to achieve higher SNR and increase the contrast between normal and abnormal myocardium. Use of a higher field system (e.g., at 3.0T) could yield better results. The achievement of higher levels of acceleration without significant additional SNR loss or artifact generation will clearly be essential to address many of the limitations of the current study, and to improve clinical efficacy and robustness.

The total duration of the perfusion imaging run is determined by the time required for transit of the contrast agent bolus through the heart, and thus equal total imaging times were recorded for Standard and Accelerated PERF sequences. However, the use of the Accelerated sequence (with an associated effective shortening of the “time per slice”) permits greater volumetric coverage.

A potential concern might be that residual contrast from the initial injection during subject preparation may perturb first-pass perfusion results in the target protocol. Given the comparatively long time between first-protocol injections and second protocol perfusion scanning (~15 min), we expect such effects to be small, and the logistical benefits of this concatenated scanning strategy are likely to outweigh any small and correctable residual perturbations. Stress-rest studies of myocardial reserve are routinely performed with multiple injections, so this would not be a unique difficulty, and various strategies for baseline correction do exist (36).

Patients with cardiomyopathy tend to have reduced bulk cardiac motion, which may have contributed to the improved coronary images observed in these patients as compared to patients with normal left ventricular function. However, patients with cardiomyopathy also tend to have a higher incidence of other imaging issues, such as arrhythmias, faster heart rates, or difficulty with breath-holding, which could offset this theoretical advantage.

Fortunately, further enhancements in speed and spatial/temporal resolution are now becoming available. New approaches integrating spatiotemporal undersampling techniques with parallel imaging, such as k-t-SENSE (31), and compressed sensing approaches, such as k-t SPARSE (37–38), FOCUSS (39), and k-t SPARSE-SENSE (40), with increased image quality at high levels of acceleration would clearly be beneficial. These may enable true isotropic volumetric imaging and engender development of a 5-minute comprehensive clinical cardiac examination. Currently, we are exploring complementary acceleration techniques, such as compressed sensing (37–39), which has been shown to combine synergistically with parallel imaging, e.g., for rapid cardiac perfusion imaging (40). Applications of compressed sensing are particularly promising for volumetric and dynamic studies and additional acceleration factors in the vicinity of 2 have been found to be feasible (41), with potential further improvements based on sampling trajectory and sparsifying transform.

In conclusion, this study demonstrates the feasibility of a 5-minute whole heart comprehensive cardiac MR examination, including PERF, CINE, CAI and LGE, each accomplished in a single BH. This protocol simplifies scan prescription, substantially reduces the total scan time, and allows full anatomic coverage with high spatial/temporal resolution. Although further work to improve image quality is required, our study indicates an important direction for simplification and improved efficiency of clinical cardiac MRI.