Estimate the effects of imaging gradients and gradient miscalibration on the b-matrix using an isotropic diffusion phantom

Motivation: To achieve "harmonized" multi-site diffusion MRI (dMRI) results, it is important to account for differences in diffusion sensitization across scanners caused by imaging gradients and gradient miscalibration. Goal(s): To use a dMRI dataset on an isotropic phantom to measure the effects of imaging gradients and gradient miscalibration and correct both effects on another dMRI dataset. Approach: Use a mathematical model of b-matrix to fit contributions of imaging gradients and gradient miscalibration to a multi-shell dMRI dataset. Results: Our correction reduced the $χ²$ error in fitting test data, relative error in diffusivity, artifactually high fractional anisotropy by a factor of 21, 2, 6, respectively. Impact: Our method can be used to account for the effects of imaging gradients and gradient miscalibration on different scanners and sites and should help harmonizing multi-site diffusion MRI data.