Purpose: Diagnostic radiology techniques continue to be utilized for improving patient outcome in radiotherapy, including diffusion weighted imaging (DWI) for tumor staging and monitoring. MR-guided radiotherapy (MRgRT) has become a highly desirable treatment modality and provides the capability for daily DWI. This study assesses apparent diffusion coefficients (ADC) from EPI and TSE sequences with a NIST traceable phantom on 0.35T MR-linac.
Methods: DWI acquisitions of a NIST traceable diffusion phantom with 13 vials of aqueous polyvinylpyrrolidone (PVP), concentrations from 0% to 50%, were performed on a 0.35T MR-linac. DWI-EPI acquisitions were performed with b-values of 0, 200, 500, and 800 s/mm², diffusion prepared (DP)-TSE acquisitions utilized b-values of 0, 300, 500, and 800 s/mm². Acquisitions were performed with pre-scan normalization on and off, along with noise fields. Additionally, signal averages (NSA) of 1, 4, 16, and 32 were acquired. ADC values were calculated using nonlinear least-squares fitting of all b-values for original and denoised images. Two denoising methods were used including quadratic background subtraction and noise field subtraction.
Results: EPI acquisitions with noise field subtraction had the lowest ADC error of -7.3% and 4.4% from reference for 0% and 40% PVP, NSA=1. ADC values from initial EPI-DWI acquisitions had an ADC error of -14.2% and -9.8% from reference for 0% and 40% PVP, NSA=1. Initial TSE images had an error of -6.9% and -57.6% from reference for 0% and 40% PVP with NSA=1. ADC values were significantly greater than reference value for 50% PVP on all scans and postprocessing methods but is outside the physiological range in humans.
Conclusion: EPI and TSE had similar ADC accuracy but were not equivalent to reference values. However, TSE acquisitions had reduced geometric distortion and greater consistency across NSA values. Further correction is required for accurate ADC calculation on low field MRgRT systems.