Fast Clustering for Interactive Tractography Segmentation

We developed a novel interactive system for human brain tractography segmentation to assist neuroanatomists in identifying white matter anatomical structures of interest from diffusion magnetic resonance imaging (dMRI) data. The difficulty in segmenting and navigating tractographies lies in the very large number of reconstructed neuronal pathways, i.e. the streamlines, which are in the order of hundreds of thousands with modern dMRI techniques. The novelty of our system resides in presenting the user a clustered version of the tractography in which she selects some of the clusters to identify a superset of the streamlines of interest. This superset is then re-clustered at a finer scale and again the user is requested to select the relevant clusters. The process of re-clustering and manual selection is iterated until the remaining streamlines faithfully represent the desired anatomical structure of interest. In this work we present a solution to solve the computational issue of clustering a large number of streamlines under the strict time constraints requested by the interactive use. The solution consists in embedding the streamlines into a Euclidean space and then in adopting a state-of-the art scalable implementation of the $k$-means algorithm. We tested the proposed system on tractographies from amyotrophic lateral sclerosis (ALS) patients and healthy subjects that we collected for a forthcoming study about the systematic differences between their corticospinal tracts.