INTRODUCTION

Cardiovascular Magnetic Resonance (CMR) imaging is considered the gold standard for assessment of biventricular function and morphology. Clinically, a typical short-axis (SAx) CINE protocol consists of a retrospectively cardiac gated two-dimensional segmented k-space cine balanced steady state free precession (bSSFP) sequence(1). The advantage of bSSFP is that it provides superior blood pool to myocardium contrast with a high signal-to-noise ratio (SNR) facilitating accurate delineations of the left and right ventricular endo- and epicardial borders, wall trabeculation and papillary muscles. In routine clinical practice a stack of 14 2D SAx slices is acquired covering the whole heart from base to apex and covering the full cardiac cycle. The acquisition is segmented over multiple breath-holds to minimize respiratory-motion artefacts. With SENSE 2 parallel imaging, typically 2 slices can be acquired per breath-hold of approximately 12 seconds each, totaling about 7 breath-holds. Performing multiple breath-holds can be problematic for some patients (e.g. heart failure) and motion-related artefacts are therefore not uncommon (2). Higher acceleration factor could be facilitated by higher parallel imaging factors requiring a receiver coil with increased number of receiver coils (3-6). Here, we performed a comparison between a 16 channel local array with SENSE 2 and a recently developed 72-channel receiver array(7) with (compressed‑)SENSE (C-SENSE) factors of 2, 4 and 6, facilitating heart-covering CINE imaging with as low as 2 breath-holds.

METHODS

8 healthy subjects (5 female), aged 31 ± 8 years, body weight 73 ± 10 kg, heart rate 59 ± 11 beats/min were included with writtenly acquired informed consents. CMR examinations were performed with a 3T MR scanner (Ingenia, Philips Healthcare, Best, The Netherlands). Whole-heart SAx cine bSSFP images were acquired using ECG-gating. Firstly, cine SAx acquisitions were acquired with a standard 16-channel coil and SENSE=2 requiring 7 breath-holds. Subsequently, the subjects were repositioned in the scanner with the 72-channel receiver coil and cine SAx acquisitions were repeated with SENSE acceleration factors of 4 (3 breath-holds), 6 (2 breath-holds) and C-SENSE=2 (7 breath-holds), 4 (3 breath-holds), and 6 (2 breath-holds) (Figure 1). The standard 12-element posterior coil array integrated in the patient table was used during all acquisitions. Imaging parameters were: Field-of-View (FOV): 350 x 350, repetition time (TR) ms/echo time (TE) ms, 2.8–3.1/1.40–1.70; flip angle (FA), 45°; slice thickness 8 mm (0.5 mm slice gap); acquired temporal resolution, 37-40 ms. Volumetric quantifications were done using Medis Suite 3.1 (Medis Medical Imaging Systems, Leiden, The Netherlands). Quantitative analysis was performed by a single reviewer, supervised by an experienced cardiologist (15+ years) specialized in cardiac imaging. Quantitative variables included left and right ventricular volumes and LV masses in end-diastole and systole. In this abstract we will focus on left ventricular ejection fraction (LVEF), but similar analyses were performed for a range of functional parameters (including volumes, cardiac output). Bland-Altman analyses and two-sided paired t-tests were performed between 16-channel SENSE=2 acquisitions and the accelerated protocols with the 72-channel coil. A sigmoid-curve was fitted to the signal intensity at the blood-pool – septum interface to quantify image sharpness (width of the sigmoid curve) and blood-to-myocardium contrast (BMC). Visual expert scoring of all slices, graded on a scale of 1 to 5 (1 = nondiagnostic, 2 = suboptimal, 3 = adequate, 4 = good, 5 = excellent) were performed by the cardiologist with focus on endocardial edge delineation (EED) and presence of artefacts. All statistical analyses were performed using R (v. 4.0.3) and Rstudio (v. 1.3.959). P-values of < 0.05 were considered significant.

RESULTS

Figure 2 shows mid-ventricular single slices acquired with both setups (16ch, 72ch) and various (C-)SENSE factors 2, 4 and 6. No differences (Table 1) were found in LVEF compared to the reference 16 channel local array SENSE 2 setup (61.33 ± 4.41%), although an overestimating trend is visible with increasing C-SENSE factors according to the Bland-Altman mean bias (C-SENSE 4: 3.23 [-3.70; 10.17], C-SENSE 6: 3.79 [-4.56; 12.14] (Figure 3). Image sharpness was not significantly different (Figure 4), though a declining BMC trend was visible in C-SENSE. The qualitative scores in all slices of EED where significant decreasing from the reference 16 channel local array SENSE 2 setup except for 72 channel C-SENSE 2. Expert quality scores of artifacts were gradually decreasing from the reference 16 channel local array SENSE 2 setup with increasing acceleration.

DISCUSSION

LVEF determined from accelerated acquisitions with SENSE and C-SENSE factors up to 6 were similar to conventionally acquired reference values. Image quality (EED and artifacts) decreased progressively with increasing acceleration; however none of the images were graded as suboptimal or lower in EED. Taken together, (C-)SENSE 4 and 6 with the new 72 channel local array seems usable from a quantitative perspective, but qualitatively the EED and artifacts are somewhat comprised. Our next step is to investigate whether free-breathing cardiac CINE imaging is possible with currently investigated (C )SENSE factors and the new 72-channel receive array.

CONCLUSION

Quantification of biventricular morphology and function is feasible in just 3 breath holds using SENSE 4 without comprising image quality (EED and artifacts) using a new 72-channel receive array. (C-)SENSE 4 and 6 remain usable at the expense of somewhat compromised image quality (EED and artifacts).

Acknowledgements

No acknowledgement found.