Xinyu Hu, Presenter: Nothing to Disclose

Lizhou Chen, Abstract Co-Author: Nothing to Disclose

Yi Liao, Abstract Co-Author: Nothing to Disclose

Qi Liu, Abstract Co-Author: Nothing to Disclose

Fei Li MD, Abstract Co-Author: Nothing to Disclose

Yanchun Yang, Abstract Co-Author: Nothing to Disclose

Qiyong Gong, Abstract Co-Author: Nothing to Disclose

Xiaoqi Huang MD, Abstract Co-Author: Nothing to Disclose

Obsessive-compulsive disorder (OCD) is one of the most common disabling psychiatric disorders. Many magnetic resonance imaging (MRI) studies have already revealed brain structural abnormalities in OCD patients involving both gray matter (GM) and white matter (WM). However, results of those publications were based on average differences between groups, which limited their usages in clinical practice. Multivariate pattern analysis (MVPA) approach is a promising analytical technique which allows the classification of individual observations into distinct groups. Therefore, the aim of this study was to examine whether the application of MVPA to high-dimensional structural MR images would allow accurate discrimination between OCD patients and healthy control subjects (HCS).

High-resolution T1-weighted volumetric 3D MR images were acquired for 33 OCD patients and 33 demographically matched HCS using a 3.0 T MRI system. Structural images were preprocessed with the Diffeomorphic Anatomical Registration using the Exponentiated Lie algebra (DARTEL) toolbox. Differences in GM volume and WM volume between OCD and HCS were examined respectively using two sorts of well-established MVPA techniques, namely, Support Vector Machine (SVM) and Gaussian Process Classifier (GPC). We also drew a receiver operating characteristic (ROC) curve to help evaluate the performance of each classifier.

Results of SVM and GPC classification between OCD patients and HCS utilizing both GM and WM were shown in the figure. Overall, the classification accuracies for both classifiers regarding GM and WM anatomy were all above 75% and the highest classification accuracy (81.82%, P<0.001) was achieved with SVM classifier using WM information.

The current study illustrated that both GM and WM anatomical features might be used to classify OCD patients from HCS. WM volume with SVM approach showed the highest accuracy in current population to reveal group differences, which indicated its potential diagnostic role in helping detecting OCD.

Using multivariate pattern analysis approach, we revealed structural MR images might be used to classify obsessive compulsive disorder from controls and provided supports for its potential role as a diagnostic tool.

Hu, X, Chen, L, Liao, Y, Liu, Q, Li, F, Yang, Y, Gong, Q, Huang, X, Investigating the Predictive Value of Whole-brain Structural Neuroimaging in Obsessive Compulsive Disorder: A Multivariate Pattern Classification Approach.  Radiological Society of North America 2014 Scientific Assembly and Annual Meeting, - ,Chicago IL. http://archive.rsna.org/2014/14017463.html