Synopsis

Keywords: Prostate, Shims, Diffusion, Spectroscopy

B0-field inhomogeneities can influence (spectroscopic) image quality of the prostate, causing geometric distortion in diffusion weighted EPI images as well as line broadening in MRSI voxels. We added a 16-element external local shim array around the pelvis to improve B0-field homogeneity within the prostate and evaluated its effects compared to standard shimming in 9 healthy volunteers. Resulting B0-maps showed a comparable or reduced frequency variation across subjects, with a significant decrease of MRSI citrate line widths. DWI did not show a robust effect over the group of subjects.

Introduction

Homogeneity of the B0-field within the prostate is essential for high quality MR (spectroscopic) imaging, but bowel gas and motion can influence the B0-field and therefore image quality, mostly at the posterior part of the prostate¹. With EPI-based DWI and MRSI sensitive to B0 inhomogeneities, off-resonance effects can cause geometric distortions in DWI, and spectral line broadening and frequency shifts in MRSI affecting water suppression and causing lipid contamination of spectra within the prostate^1,2. One way to partly prevent possible susceptibility-induced artifacts of this air is by patient preparation using antispasmodic agents and a rectum catheter³. An alternative method as proposed in the brain is to improve the B0-shim quality locally using advanced shimming methods such as a local shim array⁴. We present initial results of the first add-on local shim array for the prostate and analyse its effect on B0-maps, EPI-based DWI and MRSI.

Method

A 16-channel external array of local abdominal shim coils (named “Dia”, MR Shim GmbH, Reutlingen, Germany) was used in a 3T MR-system (Siemens Magnetom Prisma), consisting of 8 shim coils housed in two anterior modules, positioned onto a 18-channel body receive RF coil (Siemens), and 8 shim coils in the posterior modules (Figure 1), positioned on the spine-array bed.
Two shimming methods were compared: 1) The scanner’s prostate shimming routine with additional manual refining using the interactive adjustments of the scanner until a magnitude FWHM of around 30Hz of the shim volume around the prostate was achieved. 2) Dia shimming, with the “Dia”-system, in which a reference field map was acquired using a dual-echo gradient-echo sequence, and shim values were calculated non-iteratively using the provided software tool with a constrained convex optimization algorithm.
Nine healthy volunteers (mean age 40y, written informed consent) were scanned without any bowel preparation. After anatomical T2w scans in three orientations (transversal: TR/TE 5660/92ms, FOV 256x240mm, resolution 0.3×0.3×3.0mm), field maps were acquired with both shim methods with a dual gradient echo sequence and compared by the standard deviation of the off-resonance frequencies in the prostate region.
For DWI analysis, EPI images with b-value of 50s/mm² were acquired with both shim methods (TR/TE 3200/63ms, FOV 256x240mm, resolution 2.0x2.0x3.0mm, 3 averages). The two different b50-maps were registered to the transversal T2w scan as undistorted anatomical reference, and a whole gland prostate outlining was performed per slice for all three series by a single reader. Both b50-maps were compared to the anatomical images by a Dice similarity coefficient (DSC) and Hausdorff distance (HD) measure^5,6 and tested for significant differences by Wilcoxon Signed-Ranks tests.
Finally, 3D MRSI data were acquired with a GOIA semi-LASER pulse sequence with MEGA lipid and water suppression⁷ (TR/TE 750/88ms, FOV 77x56x56mm³, matrix 8x8x11, Hamming-filtered and interpolated to 16x16x16). Citrate (Cit) linewidth maps were generated for multiple slices of all subjects for both shim methods, and a paired samples t-test was performed in 236 selected independent voxels from the full cohort of acquired datasets.

Results

Examples of 2 representative slices of B0-maps show an improved homogeneity of the B0-field for the Dia shim (Figure 2). Overall the off-resonance frequencies of the Dia shim showed a comparable or reduced frequency variation (Table 1).
EPI b50-images showed some geometric distortions (Figure 3). The corresponding quantitative measures in Table 1, showed no significant overall effect for DWI with or without the Dia shim. On subject-level, the mean DSC was significantly decreased for the Dia shim in two volunteers, indicating a better standard shim (mean DSC 0.81±0.03 and 0.71±0.07 with Dia shim vs 0.85±0.04 and 0.86±0.03 with standard shim). In one volunteer the Dia shim performed better, as shown by a significantly improved HD with the Dia shim (16.3±5.2mm vs 27.1±5.7mm).
Figure 4 presents the effect of both shimming methods on MRSI, with representative examples showing a quantitative (Cit linewidth maps) and qualitative (spectra) comparison. Analysis of Cit linewidths within the prostate over the full cohort, showed 66% of all standard shim voxels reached linewidths below 15Hz, while this result was met in 82% of the voxels with Dia shim. The mean Cit linewidth over all selected voxels showed a significant improvement (p<0.01) for the Dia shim (12.8±2.5Hz) compared to the scanner shim (14.3±4.1Hz). The mean Cit linewidths per subject are presented in Table 1.

Conclusion and Discussion

This study shows the potential of an add-on local shim array to improve the shim quality for prostate imaging in healthy volunteers. For DWI, it was shown that for some slices, geometric distortion was decreased resulting in a lower DSC and HD, although in other subjects, where the scanner shim performed well without distortions, the Dia shim did not result in any improvements. For MRSI, the Dia shimming approach succeeded in overall significantly narrower Cit linewidths than the conventional scanner shim, leading to an improved spectral quality, possibly improving the accuracy of spectral quantification.
In patients with generally larger prostates and bowel-motion-suppressing medication the effect of additional shimming remains to be seen in following work. In this older population metallic implants can be present with severe influence on local field homogeneity that could strongly benefit from additional shimming flexibility.

Acknowledgements

No acknowledgement found.