Clinicians interested in cardiac perfusion and MR physicists interested in using MS-CAIPIRINHA.The simultaneous multi-slice Parallel Imaging (PI) technique MS-CAIPIRINHA [1] is an effective method to extend the anatomical coverage in multi-slice MR acquisitions. This is especially attractive for the time-critical investigation of myocardial perfusion, where Cartesian MS-CAIPIRINHA is often combined with in-plane PI acceleration [2]. High in-plane undersampling factors, however, can lead to high g-factor noise amplification, which is impeding the evaluation of myocardial tissue. When using radial readouts in conjunction with MS-CAIPIRINHA [3, 4], the aliasing energy is reduced such that lower g-factors are expected. Furthermore, contrast information is acquired with every readout, which can be used together with model-based techniques for perfusion quantification [5]. In this work, saturation-recovery-prepared and ECG-triggered radial MS-CAIPIRINHA was implemented and tested for perfusion investigations of the human myocardium. A study on g-factors was performed to evaluate the benefit in comparison to the Cartesian equivalent.

A pulse sequence was developed by in-house modifying the TurboFLASH prototype sequence presented in [6]. A dual-band pulse was used to excite two slices simultaneously, while the RF-phase of the first slice was kept unmodulated and the phase of the second slice was toggled between 0° and 180°. Radial readouts were applied, such that the information of the second slice cancels out in the central k-space of a standard gridding reconstruction. The conjugate gradient SENSE (CG-SENSE) algorithm described in [3] was implemented in MATLAB (The MathWorks, Natick, MA) to reconstruct two fully sampled slices. The L-curve of the residual-norm [7] was consulted to define a termination criterion for this iterative reconstruction.

In a first experiment, a phantom was scanned using both Cartesian and radial MS-CAIPIRINHA (MAGNETOM Skyra, Siemens Healthcare, Erlangen, Germany; dual-slice mode, slice distance = 24mm, TE = 1.4ms, TR = 3.1ms, slice thickness = 8mm, in-plane resolution = 2mm x 2mm, 37 phase-encoding steps or radial projections, image matrix = 128 x 128) together with a measurement of noise amplitude and correlation in the phased-array receiver. To allow for a fair comparison, both the radial and the Cartesian scan were reconstructed using the CG-algorithm described above. A fully sampled measurement was performed to determine the coil sensitivities. The method described in [8] was used to obtain g-factor maps for the two techniques.

The radial sequence was applied to perform a saturation-recovery-prepared and ECG-triggered perfusion measurement in the heart of a healthy volunteer (same scanner, dual-slice mode, slice distance between adjacent slices =7.7mm, slice distance between simultaneously acquired slices = 23mm, saturation preparation & ECG triggering, TE = 1.5ms, TR = 2.6ms, slice thickness = 5mm, in-plane resolution = 2.5mm x 2.5mm, 52 radial projections per image pair, image matrix = 160 x 160). A bolus of 6 ml Gadovist (Bayer Schering Pharma, Berlin, Germany) was injected and observed over 60 RR-intervals with 3x2 short-axis slices, acquired within each interval. A pre-scan was performed to determine the coil sensitivities needed for the CG-reconstruction.

Fig. 1 shows the results of the phantom study. The reconstructed images (top-row) indicate that the results obtained by radial CAIPIRINHA feature superior SNR values, especially in the center of the phantom. This is confirmed by the g-factor maps shown in the bottom row of the same figure. Peak values up to 4.5 resulted for the Cartesian acquisition while the maps of the radial scans show a spatially homogenous g-factor not exceeding a value of 3.2. Figure 2 shows the six slices of the perfusion measurement in the healthy volunteer, exemplarily for one timeframe. The images have a high overall quality, especially when considering the high acceleration factors of the acquisition. At most, in one slice (Fig. 2e) few artifacts remained in the area of the lateral wall.The results of our g-factor study confirm that radial CAIPIRINHA is capable of improving the SNR as compared to its Cartesian equivalent. It is noteworthy that the semi-convergence behavior of the CG-algorithm has the consequence that the number of iterations applied directly influences the image noise of the reconstruction. At a certain stage of the algorithm, further iterations mainly cause an increase in the statistical error (g-factor). A general termination criterion as applied in this study was therefore crucial to guarantee a meaningful comparison. The images obtained in the radial MS-CAIPIRINHA acquisition are promising for improving the SNR in future applications of this investigation and furthermore pave the way towards the application of model-based quantitative perfusion measurements [5] with extended coverage.