The use of functional networks to identify patients with schizophrenia and assess models of underlying deficits.

摘要

Schizophrenia is characterized by a number of cognitive deficits and neural abnormalities resulting in the production of a variety of symptoms. However, the question of which neural networks and which cognitive deficits best represent the fundamental processes in schizophrenia remains unresolved. The default mode network, temporal lobe network, and working memory network were extracted from two functional neuroimaging tasks using Independent Components Analysis. These networks were compared in order to determine the ability of the networks to differentiate patients from controls. Furthermore, the discrimination ability of different networks were assessed in order to evaluate which most likely represent the fundamental deficits underlying the symptoms of schizophrenia.;All four networks -- the temporal lobe network and default mode network from an auditory oddball task and the working memory network and default mode network from a serial item response paradigm -- were able to discriminate patients with schizophrenia from controls with a moderate to strong level of accuracy. The discrimination ability of the individual networks ranged from 73.7% accuracy (default mode network in the serial item response paradigm) to 88.3% accuracy for the temporal lobe network. The temporal lobe network was significantly more accurate than the other three individual networks and no combination of networks produced a better discrimination result. Overall, patients with schizophrenia activate similar functional networks to controls, but show reliable patterns of connectivity within those networks that differentiate patients and controls.