Interpolating multi-fiber models by Gaussian mixture simplification

Multi-fiber models have been introduced to leverage the accuracy of the diffusion representation in crossing fiber areas. The improved accuracy may, however, be impaired by poor processing of the multi-fiber models. In particular, interpolating multi-fiber models proves challenging, while it is a pervasive and recurrent task in many processes. The error accumulated from iterating a poor interpolation may yield significantly corrupted global results. In this paper, we propose an interpolation scheme based on gaussian mixture simplification and demonstrate its benefits over a heuristic approach in terms of spatial normalization and tractography results.