Data Analysis Strategies
Nuclear Potential Clustering
Motivation and Concept
Cluster analysis is one of the standard methods blindly group data into different categories. As we investigate the disease course of multiple sclerosis (MS), such categories might be patients with different disease grade or, more importantly, patients responding different to a given therapy. However, most of the neuroimaging MRI parameters in a multiparametric acquisition are corralated, rendering the use of standard clustering algorithms for analysis of such data difficult.Therefore, the method newly developed in our team follows a different strategy based on a concept adapted from nuclear physics. To partition the data we model the dynamic behaviour of electrically even charged nucleons interacting in an N-dimensional space. Hereby, data points that are densely distributed in space fuse to build nuclei (clusters) whereas single point clusters repel each other.
For more Information click the following links:
