Single-shot pseudo-centric EPI for magnetization-prepared imaging

FIGURE S1 Investigation on the effect of phase-correction schemes on single-shot centric-reordered EPI (1sh-CenEPI). The proposed 1sh-CenEPI images for both phantom and human brain with and without phase correction are shown. A,D, Before phase correction. B,E, With conventional three-echo correction. C,F, With whole-echo correction. Comparing the conventional EPI phase correction using three echoes (middle) and whole-echo correction (right), the image distortion and signal dropout due to the phase errors (indicated by white arrows) were clearly reduced after whole-echo phase correction

FIGURE S2 Spin-echo EPI (SE-EPI) images with the proposed pseudo-centric ordering acquired with various N values. The proposed spin-echo 1sh-CenEPI with various Ns (1 ~ 12) also showed no distinct geometric distortions, which is comparable to distortion-free gradient-echo images for both phantom and human brain

FIGURE S3 Preliminary perfusion-weighted imaging (PWI) results of 2D multislice pseudo-continuous arterial spin labeling (pCASL). The preliminary PWI results from multislice pCASL data (10 slices) were acquired with both the conventional linearly ordered 1sh-EPI (1sh-LinEPI) (A) and the proposed 1sh-CenEPI (N = 4) (B). All images are shown in the same window. The signal intensities of PWI were much higher in the 1sh-CenEPI than the conventional 1sh-LinEPI

FIGURE S4 Effect of N values on 1sh-CenEPI signal due to decay and off-resonance (B₀ inhomogeneity). A, Magnitude of point spread function (PSF) of delta function with decay under various N values. B, Effect of phase-encoding ordering on Shepp–Logan phantom image due to decay. C, The evolution of signal intensity along the phase-encode (PE) direction due to decay. D, Magnitude of PSF under on/off-resonance condition. In PSF simulation, the echo spacing, , and temporal delay due to big PE jumps were 0.1 ms, 30 ms (ie, of gray matter at 3 T), and 0.04 ms, respectively, similar to the experimental condition. The FWHM of 1sh-CenEPI was wider than that of 1sh-LinEPI and increased with smaller N values: 0.0191 (linear), 0.0319 (N = 1), 0.0262 (N = 4), 0.0250 (N = 8), 0.0235 (N = 24), and 0.0215 (N = 48) (A) and Shepp–Logan phantom images showed consistent results (B). C, Although the signal decays mono-exponentially in 1sh-LinEPI, there were signal discontinuities in 1sh-CenEPI. D, The off-resonance induced only a unidirectional pixel shift along the PE direction for 1sh-LinEPI, whereas the peak was divided bi-directionally for 1sh-CenEPI and the shift was larger in smaller N values. The frequency offset for off-resonance condition was 40 Hz. It should be noted that the off-resonance effects shown in (D) can be resolved by the modified B₀ map–based correction method that is proposed in this study

TABLE S1 Comparison of minimum TE and readout duration between linear and proposed pseudo-centric EPI. Note: Data-acquisition parameters were as follows: matrix size = 96 × 96, receiver bandwidth = 1580 Hz/px. Minimum TE was significantly reduced with centric reordering compared with linear ordering. The readout duration was elongated a bit with centric ordering compared with the linear, but the difference was within 6% in general condition (N > 2) and 20% for dramatic condition (N = 1)

TABLE S2 Comparison of averaged cerebral blood flow (CBF) values measured from three different regions of interest (N = 5). Note: The CBF values from whole-brain, gray-matter (GM), and white-matter (WM) regions of interest were calculated, respectively. Averaged values are given as mean ± SD across subjects. The GM-to-WM contrast ratio is calculated by (mean GM CBF)/(mean WM CBF)