On the thresholding of probabilistic fibre tracking maps.
Fonteijn H, Verstraten F, Norris D
Helmholtz Institute, F.C. Donders Institute for Cognitive Neuroimaging
Fibre tracking has been established in the last decade as a promising technique to reveal anatomical connectivity in vivo. Recently, developments have been made too quantify the confidence in a fibre track, based on local probabilistic models of the principal diffusive direction. However, generally the probability of a connection decreases with distance, due to the fact that the uncertainty of the fibre direction locally adds cumulatively to the uncertainty of the global track. We present a new method in which the probability is normalized for the number of voxels it has hit and we demonstrate its feasibility with an example.