Independent Component Analysis of DTI Reveals Multivariate Microstructural Correlations of Human Brain White Matter

Yi-Ou Li¹, Fan-Pei Yang¹, Christopher Nguyen², Shelly Cooper¹, Sara LaHue¹, Sandya Venugopal¹, Pratik Mukherjee¹

We perform a whole-brain multivariate analysis of microstructural white matter correlations at the voxel scale using independent component analysis (ICA). The resulting independent components show distinct spatial localization patterns indicating high fractional anisotropy co-variation across subjects. These component maps localize the strongest microstructural correlations to well-known anatomic white matter pathways, or homologous pairs of pathways, as well as segmenting certain large tracts into functionally distinct subdomains that have different patterns of connectivity. Thus, ICA of DTI maps might provide an interesting alternative to tractography for studying white matter microstructure in the normal human brain and in neurological and psychiatric disorders.