Insular functional alterations in emotional processing of schizophrenia patients revealed by Multivariate Pattern Analysis fMRI

Abstract

Emotion perception is impaired in schizophrenia patients (SP) and related to reduced social skills performance. There is a remarkable variability across subjects for functional neuroimaging alterations related to this phenomenon. In contrast to the univariate approaches of fMRI, Multivariate Pattern Analysis (MVPA) maintains the within-subject voxel-level variability. The purpose of this study was to assess emotion processing in SP, in previously identified ROIs -i.e. amygdala, hippocampus, insula, and thalamus-, while retaining the functional heterogeneity that may exist between subjects. We evaluated 23 SP and 23 healthy controls (HC). Happy, sad, and neutral faces were presented. A single trial fMRI model was applied. Patterns of activation within each ROI were classified at the subject level. Within each group, stimuli classification scores were tested against random label classification scores. In ROIs with significant results, a whole ROI classification was performed, to test whether en bloc stimuli discrimination was present. A between-group analysis was conducted also. For the classification of stimuli above chance, in the HC results were significant in the left insula in all of the stimuli dichotomies, but were non-significant in SP for happy vs. sad. In whole ROI classification, SP had significant results in bilateral insular cortex for happy vs. neutral. The left amygdala showed diminished stimuli classification scores in SP for sad vs. neutral. In conclusion, MVPA seems useful to study emotional processing in schizophrenia. In SP, either en bloc or no stimuli discrimination was seen in the insula, and reduced stimuli discrimination was seen in the left amygdala.