Evaluation of image quality of diffusion weighted readout segmentation of long variable echo-trains MR pulse sequence for lumbosacral nerve imaging at 3T

Limited magnetic resonance (MR) pulse sequences facilitate lumbosacral nerve imaging with acceptable image quality. This study aimed to evaluate the impact of parameter modification for Diffusion Weighted Image (DWI) using Readout Segmentation of Long Variable Echo-trains (RESOLVE) sequence with opportunities for improving the visibility of lumbosacral nerves and image quality.

The modified DWI-RESOLVE images facilitated visualization of the normal lumbosacral nerves with acceptable image quality, which support the clinical applicability of this sequence.

Following ethical approval and acquisition of informed consent, imaging of an MR phantom and twenty healthy volunteers (n=20) was prospectively performed with 3T MRI scanner. Acquired sequences included standard two-dimensional (2D) turbo spin echo sequences and readout-segmented echo-planar imaging (EPI) DWI-RESOLVE using three different b-values b-50, b-500 and b-800 s/mm2. Signal-to-noise ratio (SNR), apparent diffusion coefficient (ADC) and nerve size were measured. Two musculoskeletal radiologists evaluated anatomical structure visualisation and image quality. Quantitative and qualitative findings for healthy volunteers were investigated for differences using Wilcoxon signed-rank and Friedman tests, respectively. Inter and intra-observer agreement was determined with κ statistics.

Phantom images revealed higher SNR for images with low b-values with 206.1 (±10.9), 125.1 (±45.2) and 59.2 (±17.8) for DWI-RESOLVE images acquired at b50, b500 and b800, respectively. Comparable results were found for SNR, ADC and nerve size across normal right and left sided for healthy volunteer images. The SNR findings for b-50 images were higher than b-500 and b-800 images for healthy volunteer images. The qualitative findings ranked images acquired using b-50 and b-500 images significantly higher than corresponding b-800 images (P<0.05). Inter and intra-observer agreements for evaluation across all b-values ranged from 0.59 to 0.81 and 0.83 to 0.92, respectively. Conclusions: The modified DWI-RESOLVE images facilitated visualization of the normal lumbosacral nerves with acceptable image quality, which support the clinical applicability of this sequence.