All‐systolic first‐pass myocardial rest perfusion at a long saturation time using simultaneous multi‐slice imaging and compressed sensing acceleration

Figure 1. Sequence diagram for the proposed scheme. Acquisition of base, mid-, and apical slices is performed simultaneously at end-systole using a long TS and at diastole using a short TS. The time variable Trest and Tshot are reported to facilitate the reading of Numerical Simulation (subsection 2.5.1 of this article). Sequence elements are displayed for a target heart rate of 85 bpm. bpm, beats per minute; Trest, time between saturation pulse and the beginning of the shot; TS, saturation delay time

Figure 2. Schematic representation of the SMS framework. (A) Left: SMS3 with CAIPIRINHA encoding but without GC-LOLA. The application of CAIPIRINHA RF phase increments Δϕ1=2π3 and Δϕ3=−2π3 determines spatial shifts along the PE direction of apical and basal slices of FOV/3 and –FOV/3. When combining CAIPIRINHA encoding with bSSFP imaging, the bSSFP frequency response of apex and base is also shifted by ± a third of the total bandwidth with respect to the mid-slice (orange arrows). Right: schematic through-plane representation of the bSSFP frequency response for bands 1, 2, and 3 at positions z1 z2, and z3 with CAIPIRINHA RF phase increments Δϕ 1, Δϕ 2, and Δϕ 3. The shifts in the frequency response across slices are represented with an orange arrow. (B) GC-LOLA correction with SMS3. After the application of ΔMLOLA and ΔϕLOLA from Equations (1) and (2), all bands are aligned irrespective from their position along the z axis (orange arrow). bSSFP, balanced steady-state free precession; CAIPIRINHA, controlled aliasing in parallel imaging results in higher acceleration; GC-LOLA, gradient-controlled local Larmor adjustment; PE, phase encoding; SMS, simultaneous multi-slice

Figure 3. Impact of TS on M-PD rel. and BM signal ratio evaluated in numerical simulations. (A) and (B) M-PD rel. plotted as a function of TS and defect T1 times at peak myocardium with bolus injections of 0.075 mmol/kg and 0.05 mmol/kg, respectively. The dashed line highlights the TS/T1 defect combination with highest M-PD rel. (C) and (D) BM signal ratio calculated as a function of TS and blood T1/T2 times at simulated peak blood and peak myocardium, respectively. BM, blood-to-myocardium signal ratio; M-PD, myocardium-perfusion defect contrast; M-PD rel., relative M-PD contrast

Figure 4. Data from 1 exemplary patient. Left to right: base, mid-, and apical slices. Top to bottom: LTS and STS data from the SMS3 CS acquisition and the TGRAPPA sequence. In each panel, data at peak blood in the right ventricle (hence no perfusion in the left ventricle), peak blood and peak myo in the left ventricle are shown. Note that in this case the TGRAPPA scan was run before SMS3 CS. CS, compressed sensing; LTS, long saturation time; SMS, simultaneous multi-slice; STS, short saturation time; TGRAPPA, temporal GRAPPA

Figure 5. Data from 1 large patient. Reconstruction of the entire FOV of base (left), mid-, (middle), and apical (right) slices. Top and middle rows report LTS and STS data from the SMS3 CS acquisition, whereas the bottom row shows the TGRAPPA sequence. All data are shown at peak blood signal enhancement in the left ventricle. LTS, long saturation time; SMS, simultaneous multi-slice; STS, short saturation time; TGRAPPA, temporal GRAPPA

Figure 6. Same as in Figure 4 but from another patient. Arrows highlight the location of dark rim artefacts. Note that in this case the SMS3 CS scan was run before TGRAPPA. SMS, simultaneous multi-slice; TGRAPPA, temporal GRAPPA

Figure 7. Percentage signal change in (A) M-PP rel., (B) BM signal ratio at peak blood, and (C) BM signal ratio at peak myo of LTS with respect to STS data in all patients. Values were averaged across base, mid-, and apex for each patient. BM, blood-to-myocardium signal ratio; LTS, long saturation time; M-PP, myocardium-peak pre contrast; M-PP rel., relative M-PP contrast; STS, short saturation time

Figure 8. Subjective assessment of LTS, STS, and TGRAPPA datasets in 10 patients. Percentage of scores in each sequence for (A) perceived SNR. Scores were 0: low SNR resulting in nondiagnostic image quality, 1: major noise level but not limiting diagnosis, 2: minor noise level but not limiting diagnosis, 3: high SNR. (B) Image quality. Scores were 0: severe artefacts/non diagnostic, 1: major artefacts but of diagnostic quality, 2: minor artefacts and of diagnostic quality; 3: excellent/no artefacts. (C) Motion artefacts/blurring and (D) dark rim artefacts. In both (C) and (D), scores were 0: no slices, 1: one out of three, 2: two out of three, and 3: all slices affected. *denotes statistically significant differences between pairs. LTS, long saturation time; STS, short saturation time; TGRAPPA, temporal GRAPPA