TY - JOUR T1 - Evaluating the diagnostic utility of applying a machine learning algorithm to diffusion tensor MRI measures in individuals with major depressive disorder JF - bioRxiv DO - 10.1101/061119 SP - 061119 AU - David M Schnyer AU - Peter C. Clasen AU - Christopher Gonzalez AU - Christopher G. Beevers Y1 - 2016/01/01 UR - http://biorxiv.org/content/early/2016/06/29/061119.abstract N2 - The use of MRI as a diagnostic tool for mental disorders has been a consistent goal of neuroimaging research. Despite this, the vast majority of prior work is descriptive rather than predictive. The current study examines the utility of applying support vector machine (SVM) learning to MRI measures of brain white matter in order to classify individuals with major depressive disorder (MDD). In a precisely matched group of individuals with MDD (n = 25) and healthy controls (n = 25), SVM learning accurately (70%) classified patients and controls across an unselected brain map of white matter fractional anisotropy values (FA). Using a feature selection approach, where maximal discriminative voxels were selected, classification accuracy increased to over 90%. Moreover, when removing voxels identified in univariate analyses as significantly different between MDD and healthy controls, classifier accuracy was not changed; supporting the idea that group differences revealed through descriptive methods do not necessarily provide highly accurate classification. The results provide evidence that predictive methods of machine learning can be applied to neuroimaging data in order to classify the presence versus absence of MDD and that important predictive information is distributed across brain networks rather than being highly localized. ER -