Segmented diffusion-weighted imaging of the prostate: Application to transperineal in-bore 3T MR image-guided targeted biopsy

Abstract

Objective: This study aims to evaluate the applicability of using single-shot and multi-shot segmented diffusion-weighted imaging (DWI) techniques to support biopsy target localization in a cohort of targeted MRI-guided prostate biopsy patients.

Materials and methods: Single-shot echo-planar diffusion-weighted imaging (SS-DWI) and multi-shot segmented (MS-DWI) were performed intra-procedurally on a 3Tesla system in a total of 35 men, who underwent in-bore prostate biopsy inside the scanner bore. Comparisons between SS-DWI and MS-DWI were performed with (in 16 men) and without (in 19 men) parallel coil acceleration (iPAT) for SS-DWI. Overall image quality and artifacts were scored by a radiologist and scores were compared with the Wilcoxon-Mann-Whitney rank test. Correlation between the presence of air and image quality scores was evaluated with Spearman statistics. To quantify distortion, the anteroposterior prostate dimension was measured in SS and MS b=0 diffusion- and T2-weighted images. Signal-to-noise ratio was estimated in a phantom experiment. Agreement and accuracy of targeting based on retrospective localization of restricted diffusion areas in DWI was evaluated with respect to the targets identified using multi-parametric MRI (mpMRI).

Results: Compared to SS-DWI without iPAT, the average image quality score in MS-DWI improved from 2.0 to 3.3 (p<0.005) and the artifact score improved from 2.3 to 1.4 (p<0.005). When iPAT was used in SS-DWI, the average image quality score in MS-DWI improved from 2.6 to 3.3 (p<0.05) and the artifact score improved from 2.1 to 1.4 (p<0.01). Image quality (ρ=-0.74, p<0.0005) and artifact scores (ρ=0.77, p<0.0005) both showed strong correlation with the presence of air in the rectum for the SS-DWI sequence without iPAT. These correlations remained significant when iPAT was enabled (ρ=-0.52, p<0.05 and ρ=0.64, p<0.01). For the comparison MS-DWI vs SS-DWI without iPAT, median differences between diffusion- and T2-weighted image gland measurements were 1.1(0.03-10.4)mm and 4.4(0.5-22.7)mm, respectively. In the SS-DWI-iPAT cohort, median gland dimension differences were 2.7(0.4-5.9)mm and 4.2(0.7-8.9)mm, respectively. Out of the total of 89 targets identified in mpMRI, 20 had corresponding restricted diffusion areas in SS-DWI and 28 in MS-DWI. No statistically significant difference was observed between the distances for the targets in the target-concordant SS- and MS-DWI restricted diffusion areas (5.5mm in SS-DWI vs 4.5mm in MS-DWI, p>0.05).

Conclusions: MS-DWI applied to prostate imaging leads to a significant reduction of image distortion in comparison with SS-DWI. There is no sufficient evidence however to suggest that intra-procedural DWI can serve as a replacement for tracking of the targets identified in mpMRI for the purposes of targeted MRI-guided prostate biopsy.

Keywords: Diffusion weighted imaging; Image distortion; Image-guided interventions; Image-guided prostate biopsy; Magnetic resonance imaging; Prostate cancer.

Figure 1

Illustrative example demonstrating the differences in quality and artifact severity between SS-DWI and MS-DWI-iPAT in a case with a large rectal air volume. The scores assigned were as follows: SS-DWI IQ=1 (non-diagnostic), IA=3; MS-DWI- iPAT IQ=3, IA=2.

Figure 2

Illustrative example demonstrating the differences in quality and artifact severity between SS-DWI and MS-DWI-iPAT in a case with a moderate rectal air volume. The scores assigned were as follows: SS-DWI-iPAT IQ=2 (non-diagnostic), IA=3; MS-DWI-iPAT IQ=3, IA=2.

Figure 3

Graph showing correlation of the anteroposterior (AP) measurements in the T₂-weighted images with measurements in the SS-DWI (blue crosses) vs MS-DWI-iPAT (red circles) on the left and SS-DWI-iPAT (blue crosses) vs MS-DWI-iPAT (red circles) on the right. Data points are shown only for images with IQ>1. The dotted diagonal indicates perfect match between T₂-weighted image and diffusion-weighted image- based measurements. Dotted lines connecting blue crosses and red circles indicate measurements corresponding to the same patient.

Figure 3

Graph showing correlation of the anteroposterior (AP) measurements in the T₂-weighted images with measurements in the SS-DWI (blue crosses) vs MS-DWI-iPAT (red circles) on the left and SS-DWI-iPAT (blue crosses) vs MS-DWI-iPAT (red circles) on the right. Data points are shown only for images with IQ>1. The dotted diagonal indicates perfect match between T₂-weighted image and diffusion-weighted image- based measurements. Dotted lines connecting blue crosses and red circles indicate measurements corresponding to the same patient.

Figure 4

Bland-Altman plots summarizing the differences between T2-weighted image and diffusion-weighted image estimation of the prostate AP dimension at mid-gland level. Top row: T2WI vs SS-DWI (left), T2WI vs MS-DWI-iPAT (right). Bottom row: T2WI vs SS-DWI-iPAT (left), T2WI vs MS-DWI-iPAT (right). Solid horizontal line corresponds to the mean difference between the two measurement approaches. Dashed lines correspond to +/− 1.96 times standard deviation.

Figure 4

Bland-Altman plots summarizing the differences between T2-weighted image and diffusion-weighted image estimation of the prostate AP dimension at mid-gland level. Top row: T2WI vs SS-DWI (left), T2WI vs MS-DWI-iPAT (right). Bottom row: T2WI vs SS-DWI-iPAT (left), T2WI vs MS-DWI-iPAT (right). Solid horizontal line corresponds to the mean difference between the two measurement approaches. Dashed lines correspond to +/− 1.96 times standard deviation.

Figure 5

Intra-procedural images showing the appearance of the biopsy needle artifact (at tip of white arrow shown in the T2-weighted image), prostate gland (white outline corresponds to the location of the gland on the T2-weighted image) and suspected lesion area (indicated by white arrows on the ADC maps). Compared with SS-DWI, location of the prostate and needle artifact in MS-DWI-iPAT is in better agreement with the T2-weighted image, meanwhile the corresponding ADC map provides superior lesion visualization.

Figure 6

Illustration of the deformations observed in DWI of the prostate phantom comparing SS-DWI-iPAT and MS-DWI techniques. Slices shown are at the “mid-gland” (left column) and “apex” level (right column) of the phantom. Yellow outline corresponds to the threshold segmentation of the bright part of the phantom on the T2W image. Top panel shows the axial cross-sections for the T2 and SS-DWI sequences, bottom panel shows SS-DWI-iPAT and MS-DWI sequences.

Figure 6

Illustration of the deformations observed in DWI of the prostate phantom comparing SS-DWI-iPAT and MS-DWI techniques. Slices shown are at the “mid-gland” (left column) and “apex” level (right column) of the phantom. Yellow outline corresponds to the threshold segmentation of the bright part of the phantom on the T2W image. Top panel shows the axial cross-sections for the T2 and SS-DWI sequences, bottom panel shows SS-DWI-iPAT and MS-DWI sequences.