An automated string-based approach to White Matter fiber-bundles clustering

White Matter fibers play an important role in the working of brain. In order to improve their analysis, it is important to cluster them in homogeneous bundles. In this activity, the amount of data to process is huge, and an automated approach to carrying out this task is in order. Since fiber clustering should consider the position of fibers in the three-dimensional space, we are in presence of a multi-dimensional clustering problem. In this paper, we propose an automated approach to solving it. Our approach is based on a particular string representation of fibers and on a new string dissimilarity metric. Thanks to these two novelties, we can reduce the complex problem of White Matter fiber clustering to a much simpler and well-known string clustering problem. Interestingly, this way of proceeding can be extended to define other multi-view data applications, as well as to integrate (possibly heterogeneous) data coming from different domains.