Randomized iterative spherical-deconvolution informed tractogram filtering

Antonia Hain, Daniel Jörgens, Rodrigo Moreno

Abstract: Highlights•Tractogram filtering methods remove faulty connections from tractograms.•One method, SIFT, is not suitable for assessing the compliance of individual streamlines with the acquired data.•We apply SIFT to randomly selected subsets of the tractogram.•Streamlines with only positive or negative votes are used to train classifiers.•Inconsistent streamlines are a mixture of streamlines that comply and do not comply with the acquired diffusion data.