Evaluating the accuracy of multicomponent T₂ parameters for luminal water imaging of the prostate with acceleration using inner-volume 3D GRASE

FIGURE S1. Phantom diagram: A diagram of the phantom, used for validating the volume fractions for two T2 components, is shown. The dimensions were H=42 and W=40mm. When an angulated slice is excited, the signal from the imaged slice in voxels at the interface is made of two contributions (from each compartment), which produces a signal that is a mixture of the short-T2 and long-T2 compartments. With an angulated slice, a range of different combinations of the fractions between 0 and 1 (of either the short- or long-T2) are acquired at once along the imaged slice. The slice thickness was 3mm with an in-plane voxel size of 1.9 × 1.9mm².

FIGURE S2. Simulation results varying T2_S and T2_L: NNLS simulation results are shown for various T2_S and T2_L parameter values, with a) T2_L={300, 500, 700, 1000}ms with T2_S fixed to 50ms, b) T2_S={30,50,70,100}ms with T2_L fixed to 500ms, c) four “corner” cases with [T2_S,T2_L] = {[100,1000], [30,300], [30,1000], and [100,300]}ms.

FIGURE S3. T2 distributions: Example T2 distributions for the peripheral zone (PZ) are shown for NNLS, without regularization. a) The distribution for each voxel is shown as a map, for all PZ voxels. b) The distribution for each voxel is plotted, for all voxels in PZ (overlaid). c) The average distribution is shown with d) the resulting luminal water fraction map. These T2 distributions correspond to the images in Figure 5.

FIGURE S4. Subtraction of images between repeated acquisitions: Example images of the individual signal acquisitions are shown for echo 4 (TE=44ms). The difference images are also shown between the first and second acquisitions, and between the first and third acquisitions. Qualitatively, it can be seen that the prostate is minimally affected by motion occurring between repeated acquisitions. Most of the signal difference (just below the prostate) results from motion of the colon due to peristalsis. These images correspond to the maps shown in Figure 5.

FIGURE S5. In vivo parameter estimates: The mean T2_S, T2_L and luminal water fractions, averaged over all 5 subjects, are shown for the peripheral zone (PZ) and the transition zone (TZ). Results for both fitting methods (NNLS and BIEXP-SC) are shown, for all three averages (“All”) and also for the first average (“a1”). Error bars represent standard deviations of the mean over all subjects. Paired groups are significantly different by the Wilcoxon rank-sum test (P<0.01).

FIGURE S6. Bland-Altman plots of LWF in PZ and TZ: Bland-Altman plots comparing the luminal water fraction between two methods (NNLS and BIEXP-SC) are shown for PZ and TZ regions in the prostate. Pixelwise parameter values have been combined for all 5 subjects.

TABLE S1. Standard T2-weighted protocol: The scan parameters are shown for the three standard T2-weighted scans used. The prostates were imaged in three (axial, sagittal and coronal) orientations.