THE INFLUENCE OF SPATIAL RESOLUTION AND FIBER DENSITY ON THE FRACTAL DIMENSION OF CEREBRAL FIBER TRACTS

Abstract

The determination of the generalized fractal dimensions (FDs) from MRI tractograms is affected by image resolution and the number of fiber tracts. In this work, we demonstrate that only certain combinations of image resolution, number of fiber tracts and exponent of the generalized dimensions are able to capture structural changes in human cerebral white matter. MRI tractography was carried out for a different number of fiber tracts and was discretized on different grid sizes. Generalized FDs were evaluated for two groups of healthy subjects with different age distribution. For the box-counting dimension ([Formula: see text]) the highest difference between two age groups was found for a matrix size of 20483 pixel with a strong dependence on the number of fiber tracts. The correlation dimension found the highest differences for a resolution of 10243 pixel, largely independent of the number of fiber tracts. The correlation dimension evaluated for tractograms discretized on a 10243 grid is a robust measurement to capture structural changes in human cerebral white matter by means of FD.