Neuroimaging: Diagnostic Boundaries and Biomarkers

TLDR: Though a large number of studies have reported differences between patients with psychotic disorders and healthy controls in structural and functional neuroimaging measures, only few results are robust and consistent, and machine-learning algorithms might represent a good opportunity for the progress in this field.

Abstract: Different medical disciplines have adopted biomarkers in order to establish a diagnosis and predict clinical and functional outcome of a disease. In psychiatry, the search for biomarkers could lead to substantial improvement in mental disorder diagnosis and care. Different neuroimaging techniques have contributed to improve our understanding of brain structure and functioning in patients with psychotic disorders. However, though a large number of studies have reported differences between patients with psychotic disorders and healthy controls in structural and functional neuroimaging measures, only few results are robust and consistent. In addition, so far, even robust and consistent findings have differences at group level, which so far did not translate into applications at the individual level. The heterogeneity of psychotic disorders, the use of medications, and the infrequent replication of findings in independent patient cohorts from different centers have limited the identification of reliable neuroimaging biomarkers for diagnosis and outcome prediction of psychotic disorders. Machine-learning algorithms might represent a good opportunity for the progress in this field.