Reduced-distortion diffusion weighted imaging for head and neck radiotherapy.

Background And Purpose: Quantitative Diffusion Weighted Imaging (DWI) has potential value in guiding head and neck (HN) cancer radiotherapy. However, clinical translation has been hindered by severe distortions in standard single-shot Echo-Planar-Imaging (ssEPI) and prolonged scan time and low SNR in Turbo-Spin-Echo (ssTSE) sequences. In this study, we evaluate "multi-shot" (ms) msEPI and msTSE acquisitions in the context of HN radiotherapy.

Materials And Methods: ssEPI, ssTSE, msEPI with 2 and 3 shots (2sEPI, 3sEPI), and msTSE DWI were acquired in a phantom, healthy volunteers (N=10), and patients with HN cancer (N=5) on a 3-Tesla wide-bore MRI in radiotherapy simulation RF coil setup, with matched spatial resolution (2x2x5mm) and b = 0, 200, 800 s/mm.Geometric distortions measured with deformable vector field (DVF) and contour analysis, apparent diffusion coefficient (ADC) values, and signal-to-noise-ratio efficiency (SNR) were quantified for all scans.

Results: All techniques significantly (P<1x10) reduced distortions compared with ssEPI (DVF = 3.1 ± 1.3 mm). Distortions were marginally lower for msTSE (DVF = 1.5 ± 0.6 mm) than ssTSE (1.8 ± 0.9 mm), but were slightly higher with 2sEPI and 3sEPI (2.6 ± 1.0 mm, 2.2 ± 1.0 mm). SNR reduced with decreasing distortion with ssEPI=21.9 ± 7.9, 2sEPI=15.1 ± 5.0, 3sEPI=12.1 ± 4.5, ssTSE=6.0 ± 1.6, and msTSE=5.7 ± 1.9 for b = 0 images. Phantom ADC values were consistent across all protocols (errors ≤ 0.03x10mm/s), but ADC values were ∼ 4 % lower with msEPI and ∼ 12 % lower with ssTSE/msTSE compared with ssEPI.

Conclusions: msEPI and TSE acquisitions exhibited improved geometric distortion at the cost of SNR and scan time. While msTSE exhibited the least distortion, 3sEPI may offer an appealing middle-ground with improved geometric fidelity but superior efficiency and ADC quantification.