Unsupervised Clustering of Neural Pathways

Sergio Medina (1, 2)

1 LNAO - Laboratoire de Neuroimagerie Assistée par Ordinateur
2 PARIETAL - Modelling brain structure, function and variability based on high-field MRI data

Diffusion-weighted Magnetic Resonance Imaging (dMRI) can unveil the microstructure of the brain white matter. The analysis of the anisotropy observed in the dMRI contrasted with tractography methods can help to understand the pattern of connections between brain regions and characterize neurological diseases. Because of the amount of information produced by such analyses and the errors carried by the reconstruction step, it is necessary to simplify this output. Clustering algorithms can be used to group samples that are similar according to a given metric. We propose to explore the well-known K-means clustering algorithm and QuickBundles, an alternative algorithm that has been proposed recently to deal with tract clustering. We propose an efficient procedure to associate K-means with Point Density Model, a recently proposed metric to analyze geometric structures. We analyze the performance and usability of these algorithms on simulated data and a database of 10 subjects.

Mots clés: Tract Clustering; Unsupervised Learning; Point Density Model; DWI Imaging; DTI Clustering

Domaines: Apprentissage [cs.LG]

Sergio Medina. Unsupervised Clustering of Neural Pathways. Machine Learning [cs.LG]. 2014. ⟨hal-00908433v1⟩