Investigating apparent differences between standard DKI and axisymmetric DKI and its consequences for biophysical parameter estimates

Abstract

PurposeIdentify differences between the acquisition-time efficient axisymmetric diffusion kurtosis imaging (DKI) model and standard DKI and their consequences on biophysical parameter estimates using standard DKI parameters as the ground truth. MethodsNoise-free, synthetic diffusion MRI (dMRI) human brain data are generated using standard DKI and fitted with axisymmetric DKI and standard DKI. Then, the five axisymmetric DKI tensor metrics (AxTM), the parallel and perpendicular diffusivity and kurtosis and mean of the kurtosis tensor, attainable with both DKI models are computed. Next, the five biophysical parameters axon water fraction and dispersion, extra axonal parallel and perpendicular diffusivity and intra axonal parallel diffusivity are estimated from the AxTM using the WMTI-Watson model. Finally, the number of substantially differing voxels (SDV), defined as voxels where estimation results of both DKI models differ more than 5%, is calculated for the AxTM and the biophysical parameters. ResultsFor the AxTM, the number of SDV was biggest for the parallel (26%) and perpendicular (51%) kurtosis while the other three AxTM had very few SDV (less than 5%). The biophysical parameters had much more SDV than the AxTM from which they were computed, ranging from 29% to 50%. ConclusionAxisymmetric DKI is a viable alternative to standard DKI in studies focusing on effects based on the parallel and perpendicular diffusion and mean of the kurtosis tensor. However, our findings urge caution when using axisymmetric DKI to investigate effects based on the parallel and perpendicular kurtosis or use it to estimate the biophysical parameters.