Meeting Banner
Abstract #1000

A Multiparametric DTI analysis for Stratification of CSVD Subgroups

Rajikha Raja1, Jeffrey Thompson2, Jill Prestopnik2, Erik Erhardt3, John Adair2, Vince Calhoun1,3, Gary Rosenberg2, and Arvind Caprihan1

1The Mind Research Network, Albuquerque, NM, United States, 2University of New Mexico Health Sciences Center, Albuquerque, NM, United States, 3University of New Mexico, Albuquerque, NM, United States

Vascular cognitive impairment and dementia (VCID) which includes all forms of small vessel diseases is associated with white matter (WM) damages. Diffusion tensor imaging (DTI) has been used widely in several studies in characterizing these WM changes in VCID. The aim of this study is to evaluate the classification accuracy of diffusion measures to distinguish cerebral small vessel disease (CSVD) subjects from healthy controls (HC) and be further able to discriminate CSVD subgroups consisting of subcortical ischemic vascular disease , mixed dementia , leukoaraiosis , and Alzheimer’s disease. The proposed classification framework includes feature extraction followed by multiclass random forest classification with leave one out cross validation. The classification features were defined based on histogram measures calculated on 4 WM regions from four DTI modalities (FA, MD, AD, and RD). Multiple classification tasks were performed owing to the different subject groups. We have shown that DTI combined with FLAIR and cognitive score can give up to 78% classification accuracy for the 5-group and the 4-group classification without the need of the harder to measure cerebrospinal fluid biomarkers.

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

Join Here