Joint analysis of fiber shape and cortical folding patterns

Quantitative descriptions of white matter (WM) fiber shape and cortical folding patterns are important for neuroscience research. This paper presents a novel computational method for WM fiber shape pattern analysis, that is, WM fibers are clustered into five primitive shape patterns: closed `U', `M', curved line, open `U' and straight line, based on the automatic clustering of their shape features. We study the correlation between the fiber shape patterns and the folding patterns of cortical regions that the fibers connect, and find that these patterns are highly correlated. Also, we investigate the shape patterns of the WM fibers that connect different brain regions such as cortical lobes and subcortical regions, and find that the shape patterns of the fibers connecting different brain regions are quite specific. These results might provide insights into the structural and functional architecture of the human brain.