Meeting Banner
Abstract #4045

k-Means and Graph Cuts Clustering of Diffusion MRI in Rat STN

Ellen Brunenberg1, Erik Pelgrim1, Bart ter Haar Romeny1, Bram Platel2

1Biomedical Engineering, Eindhoven University of Technology, Eindhoven, Netherlands; 2Biomedical Engineering, Maastricht University Medical Center, Maastricht, Netherlands


Deep Brain Stimulation of the STN for Parkinsons alleviates motor symptoms, but often causes cognitive or emotional side effects. In this abstract, we present the results of different clustering algorithms in order to separate the rat STN motor and cognitive/emotional parts. We performed k-means and graph cuts clustering on spherical harmonics coefficients for each voxel in an atlas-based ROI around the STN. Graph cuts clustering results in a medial STN cluster, which might correspond to the cognitive/emotional part of the rat STN. Future research could focus on other HARDI distance measures and on fiber tracking projections as clustering input.