Alexander Haubold, Bradley S. Peterson, and Ravi Bansal
Brain morphometry in recent decades has increased our understanding of the neural bases of psy-chiatric disorders by localizing anatomical disturbances to speciﬁc nuclei and subnuclei of the brain. Atleast some of these disturbances precede the overt expression of clinical symptoms and possibly areendophenotypes that could be used to diagnose an individual accurately as having a speciﬁc psychiatricdisorder. More accurate diagnoses could signiﬁcantly reduce the emotional and ﬁnancial burden ofdisease by aiding clinicians in implementing appropriate treatments earlier and in tailoring treatment tothe individual needs. Several methods, especially those based on machine learning, have been proposedthat use anatomical brain measures and gold-standard diagnoses of participants to learn decision rulesthat classify a person automatically as having one disorder rather than another. We review the generalprinciples and procedures for machine learning, particularly as applied to diagnostic classiﬁcation, andthen review the procedures that have thus far attempted to diagnose psychiatric illnesses automaticallyusing anatomical measures of the brain. We discuss the strengths and limitations of extant proceduresand note that the sensitivity and speciﬁcity of these procedures in their most successful implementa-tions have approximated 90%. Although these methods have not yet been applied within clinicalsettings, they provide strong evidence that individual patients can be diagnosed accurately using thespatial pattern of disturbances across the brain.
Keywords: Machine learning, brain morphometry,automated diagnosis, cortical thickness, psychiatric disorders, cross validation, support vectormachines.