Status of on-scanner gradient nonlinearity bias correction for extension to diffusion kurtosis imaging in multi-vendor setting

Motivation: Clinical oncology imaging trials utilize non-Gaussian diffusion kurtosis imaging (DKI) metrics to improve diagnostic accuracy and therapy response monitoring. Goal(s): Our ongoing multi-institution project is testing the extension of vendor-provided on-scanner implementations of correction for gradient non-linearity (GNL) bias in Gaussian apparent diffusion coefficient (ADC) for DKI applications. Approach: Previously developed formalism for correction of multiplicative GNL bias in DKI parameters is applied using b-value maps or via corrected diffusion weighted images (DWI). Results: This work evaluates feasibility of using on-scanner generated multi-b DWI outputs for DKI GNL correction across three MRI platforms in a quantitative phantom with non-Gaussian diffusion materials. Impact: Feasibility of diffusion kurtosis parameter correction for gradient nonlinearity bias using vendor-provided on-scanner outputs is tested across three MR vendor systems in a quantitative non-Gaussian diffusion phantom. Correction integration with other DWI enhancements is found necessary for the best performance.