Structural magnetic resonance image averaging in schizophrenia.

OBJECTIVE Intersubject averaging of structural magnetic resonance (MR) images has been infrequently used as a means to study group differences in cerebral structure throughout the brain. In the present study, the authors used linear intersubject averaging of structural MR images to evaluate the validity and utility of this technique and to extend previous research, conducted using a different approach to image averaging, in which reduction in thalamic size and abnormalities in perithalamic white matter tracts in the brains of schizophrenic patients were reported by Andreasen et al. METHOD A 1.5-T MR scanner was used to obtain high-resolution, whole brain T1-weighted structural MR images for an age-matched sample of 25 schizophrenic patients and 25 normal control subjects. A "bounding box" procedure was used to create a single "averaged" brain for the schizophrenic group and for the control group. Differences in signal intensity between the two average brains were examined on a pixel-wise basis through use of one-tailed effect size maps. RESULTS Effect size maps revealed widespread patchy signal intensity differences between the two groups in both cortical and periventricular areas, including major white matter tracts. The signal intensity differences were consistent with cortical thinning/sulcal widening and ventricular enlargement. No differences were found within thalamus or in immediately surrounding white matter. Effect size maps for differences (schizophrenic minus normal subjects) had only small values. CONCLUSIONS These results are consistent with diffuse structural brain abnormalities of both gray and white matter in schizophrenic populations such as the one in this study.