Multiple-subjects connectivity-based parcellation using hierarchical Dirichlet process mixture models.

摘要: We propose a hierarchical infinite mixture model approach to address two issues in connectivity-based parcellations: (i) choosing the number of clusters, and (ii) combining data from different subjects. In Bayesian setting, we voxel-wise anatomical connectivity profiles as an multivariate Gaussian distributions, with Dirichlet process prior on cluster parameters. This type allows us conveniently clusters estimate its posterior distribution directly data. An important benefit using modelling is extension multiple subjects clustering via processes. Data are used infer class parameters classes at individual group level. Such method accounts for inter-subject variability, while still benefiting yield more robust estimates clusterings.