Meeting Banner
Abstract #2592

Evaluating the accuracy of diffusion models at multiple b-values with cross-validation

Ariel Rokem 1 , Kimberly L Chan 1 , Jason D Yeatman 1 , Franco Pestilli 1 , Aviv Mezer 2 , and Brian A Wandell 2

1 Stanford University, Stanford, CA, United States, 2 Stanford University, Stanford, California, United States

Models of diffusion MRI (DWI) are used for inferences about the properties of the tissue and fiber orientations. Though stability of DWI model parameters is often evaluated, there are no extensive studies of model prediction accuracy. We evaluated different models using cross-validation in a test-retest data set and data from the Human Connectome Project. In most of the white matter and multiple b-values, we find that the classic diffusion tensor model predicts the data more accurately than test-retest reliability. However, modeling the signal as a combination of contributions from distinct white matter fascicles provides more accurate model predictions.

This abstract and the presentation materials are available to members only; a login is required.

Join Here