Meeting Banner
Abstract #0126

Magnetic Resonance Imaging-Assisted Diagnosis of Major Depressive Disorder Using a Multiparameter Classification Approach Based on Gray Matter Abnormality

Lihua Qiu1, Xiaoqi Huang1, Qizhu Wu1, Shiguang Li1, Su Lui1, Peiyu Huang2, Qiyong Gong1

1Huaxi MR Research Center, Department of Radiology, West China Hospital of Sichuan University, Chengdu, Sichuan, China; 2Institute of Neuroscience, Chongqing Medical University, Chongqing, China

Past studies applied Support Vector Machine (SVM) using structural MRI data had yielded some promising results in distinguish psychiatric disorders. However, only volumetric information had been consideration in those past studies. In the present study, we use multiparameter(seven morphometric parameters including volumetric and geometric features on grey matter) classification approach to distinguish first-episode, drug-nave MDD patients from normal controls. Among all parameters, we found cortical thickness measurement showed the highest accuracy in revealing group differences between control and MDD. Current study provided new approach which might be useful for translational application of MRI in future.

Keywords

abnormality account accuracies accuracy accurately affect affected aimed alignment allows analytical anatomy antidepressant applied assisted automated brain build chance china classes classification classifier clinical cohort combined confirm confounds consideration contribute control controls correlations cortex cortical cross curvature depressed depression depressive depth detect detecting diagnosis differentiate dimensional discriminates disorder disorders distinct distinguish distortion distributed drug early education entire episode equally evaluation explore features female find generalized geometric gong gray handedness healthy help heterogeneity highest hospital individual individually individuals influence institute inter investigating leave listed location longitudinal machine male matched maximal medical medication medications might network observations overall particularly past patient patients pattern people plots populations prediction procedures processing profile promising psychiatric radiology rapidly reconstruction recruited regional republic resolution respectively response revealing scanner sensitivity seven software spatial spatially specificity springer structural structure studies subject subtle suited support surface surfer table takes thirty transformation treatment type useful vector volume volumetric west whether years yielded