White matter abnormalities of microstructure and physiological noise in schizophrenia

Abstract

White matter abnormalities in schizophrenia have been revealed by many imaging techniques and analysis methods. One of the findings by diffusion tensor imaging is a decrease in fractional anisotropy (FA), which is an indicator of white matter integrity. On the other hand, elevation of metabolic rate in white matter was observed from positron emission tomography (PET) studies. In this report, we aim to compare the two structural and functional effects on the same subjects. Our comparison is based on the hypothesis that signal fluctuation in white matter is associated with white matter functional activity. We examined the variance of the signal in resting state fMRI and found significant differences between individuals with schizophrenia and non-psychiatric controls specifically in white matter tissue. Controls showed higher temporal signal-to-noise ratios clustered in regions including temporal, frontal, and parietal lobes, cerebellum, corpus callosum, superior longitudinal fasciculus, and other major white matter tracts. These regions with higher temporal signal-to-noise ratio agree well with those showing higher metabolic activity reported by studies using PET. The results suggest that individuals with schizophrenia tend to have higher functional activity in white matter in certain brain regions relative to healthy controls. Despite some overlaps, the distinct regions for physiological noise are different from those for FA derived from diffusion tensor imaging, and therefore provide a unique angle to explore potential mechanisms to white matter abnormality.

Keywords: Physiological noise; Resting state fMRI; Schizophrenia; White matter.

Fig. 1

TSNR map of the resting state fMRI time series for a represented control subject. (a) oblique axial slices without smoothing; (b) with smoothing. Smoothing can reduce thermal noise but has little effect on the physiological noise. Four slices outlined in (A) and (B) are displayed in (C) and (D). Note that the white matter becomes prominent after smoothing

Fig. 2

Comparison of motion information across SZ and NC subjects included in the study. (a) Total number of volumes selected for TSNR analysis for SZ and NC; (b) translational (blue) and rotational (red) movements for both SZ and NC. The translational and rotational movements were computed according to Eq. 2

Fig. 3

Statistical results from two-sample t-test of the TSNR maps between SZ and NC groups. Data are displayed in coronal planes 8 mm apart. The numbers on top indicate the location of slices in each hemisphere in MNI space. The statistical threshold is p<0.001 and cluster size> 100 voxels

Fig. 4

Results of TBSS analysis of FA values. Regions of significant FA reduction for SZ are shown in red-yellow with p value<0.05. The mean FA skeleton is displayed as green lines. The results are overlaid on the MNI template from Z=−30 mm to Z=47 mm. We applied 5000 permutations in the permutation-based nonparametric statistics. The p value was FWE corrected

Fig. 5

A comparison of SZ and NC group data for two sets of analyses. Data are displayed on the MNI template from Z=−30mm to Z=47 mm. Voxels that are significantly different between SZ and NC from the TSNR analysis (red-yellow, p<0.001 uncorrected to p<0.05 FEW corrected) overlaid on the results of the TBSS analysis (blue-light blue, p<0.05 to p<0.01)