Machine Learning in Neuroimaging for Diagnosis of Major Depressive Disorder

Abstract

Is machine learning an effective diagnostic tool to identify the presence of major depressive disorder in a patient? Major depressive disorder (MDD) is a common mood disorder, distinguished by prolong emotional distress and dysfunction in normal activities, yet its current diagnosis still depends on subjective interpretation of clinical interviews and self-reported symptoms. Not only are these prevalent methods susceptible to inconsistent diagnoses, the lack of clear boundaries in present diagnostic criteria to evaluate MDD also further magnifies the need for objective diagnostic tools in clinical environment. One of those methods explored by psychologists is the application of machine learning in neuroimaging techniques. Due to its analytical prowess, machine learning has become popular in psychological research to understand and discover biomarkers in illnesses. This literature review aims to evaluate the effectiveness of different machine learning models and neuroimaging in classifying MDD patients before discussing some of the major setbacks faced by researchers and the suggested solutions for overcoming these barriers to clinical application.