Structural MRI-Based Predictions in Patients with Treatment-Refractory Depression (TRD)

• DOI: 10.1371/journal.pone.0132958
• Publication Date: 2015-07-17T18:17:13Z
• Authors: Blair A. Johnston; J. Douglas Steele; Serenella Tolomeo; David Christmas; Keith Matthews
• Keywords: Treatment-Resistant Depression; Depression; Refractory (planetary science); Grey matter

The application of machine learning techniques to psychiatric neuroimaging offers the possibility identify robust, reliable and objective disease biomarkers both within between contemporary syndromal diagnoses that could guide routine clinical practice. use quantitative methods is consequently important, particularly with a view making predictions relevant individual patients, rather than at group-level. Here, we describe treatment-refractory depression (TRD) diagnosis using structural T1-weighted brain scans obtained from twenty adult participants TRD 21 never depressed controls. We report 85% accuracy subject diagnostic prediction. Using an automated feature selection method, major regions supporting this significant classification were in caudate, insula, habenula periventricular grey matter. It was not, however, possible predict degree 'treatment resistance' least as quantified by Massachusetts General Hospital (MGH-S) staging method; but insula again identified region interest. Structural imaging data alone can be used status, not MGH-S staging, high patients TRD.