Diagnostic Classification with Neural Correlates of Verbal Fluency Distinguishes Schizophrenia from Bipolar Disorder and Healthy Individuals

Sergi G. Costafreda¹, Cynthia H.Y. Fu, Marco Picchioni, Timothea Toulopoulou, Fergus Kane, Sri Kalindindi, Colm McDonald², Janaina Mourao-Miranda, Eugenia Kravariti, Muriel Walshe, Nicolette Marshall, Diana Prata, Michael J. Brammer, Robin M. Murray, Philip K. McGuire

Using pattern recognition methods, we sought to examine whether the spatial pattern of brain activation during language production as measured with fMRI could be used for the identification of schizophrenia patients (n=32) among bipolar patients (n=32) and heathy controls (n=40). In a three-way classification, 71% of 104 subjects were allocated to their correct diagnosis. The accuracy for the diagnosis of schizophrenia was even higher (Sensitivity=84%, Specificity=90%). Accurate differential diagnosis of schizophrenia patients was achieved with computerized analysis of individual brain activation during verbal fluency. Pattern classification of fMRI measurements may provide a step towards developing neurobiological diagnostic tools for schizophrenia