Purpose: Diffusion weighted imaging (DWI) is sensitive to changes in tumor microstructure and can be used to non-invasively monitor radiotherapy treatment response. Despite being clinically available in MR-scanners, DWI can be further optimized in MR-guided radiotherapy systems. Maximum b-value is an important sequence parameter that controls the amount of diffusion weighting; high diffusion weighting leads to noise and less accurate apparent diffusion coefficient (ADC) estimates. MR-Linac users follow the recommended maximum b-value of 500 s/mm², however, higher maximum b-values are routinely used for DWI in sites such as the brain. The goal of this work was to optimize DWI in an MR-Linac using a standardized phantom.
Methods: DWI sequences for brain, prostate, and abdomen were optimized with respect to maximum b-value and number of signal averages (NEX). DWI scans of the NIST-QIBA phantom (High Precision Devices, Boulder, USA) were conducted in the Elekta Unity MR-Linac’s (Elekta AB, Stockholm, Sweden) Philips Marlin 1.5T MR scanner. ADCs were estimated by fitting with offline reconstruction (MATLAB). Maximum b-value and NEX were varied and differences between measured and manufacturer-specified ADC were reported. The impact of varying NEX on scan duration was also quantified.
Results: For the brain sequence, maximum b-value achieving accurate ADC was 3,000 s/mm² (±4.0% of standard across all diffusion vials). For all three sites, fitting with a subset of b-values below 400 s/mm² reduced ADC accuracy (±17.6%, 22.3%, 24.1% for brain, prostate and abdomen, respectively). Reducing NEX from clinical protocol to the scanner’s minimum resulted in underestimation of ADC by 14.3%(±1.7%), 18.4%(±5.1%), and 17.5%(±4.3%) for the brain, prostate, and abdomen sequences, respectively but reduced average scan duration from 5.8 to 1.1 minutes.
Conclusion: Site-specific DWI sequences were optimized for ADC accuracy with respect to maximum b-value and scan time using a standardized phantom. Future work involves repeating this methodology in vivo.
Funding Support, Disclosures, and Conflict of Interest: FedDev ON, Canada