Title: Functional features in brain imaging
Authors: Donatello Telesca - UCLA (United States) [presenting]
Abstract: Brain imaging techniques produce data which can be fruitfully interpreted as the realization of stochastic processes over functions of one or several evaluation domains. We propose a modeling approach for the identification of functional features, which combine to define the complete pattern observed for several statistical units. A probabilistic interpretation of the Karhunen-Loeve construction is merged with a prior on finite binary feature allocation matrices. The approach examined in the context of several case studies involving both benchmark datasets used in functional data analysis and brain imaging studies of developmental neurocognition.