Abstract:This paper introduces a framework for the application of constrained non-negative matrix factorization (cNMF) to estimate the statistically distinct neural responses in a sequence of functional magnetic resonance images (fMRI). While an improved objective function has been defined to make the representation suitable for task-related brain activation detection, in this paper we explore various methods for better detection and efficient computation, placing particular emphasis on the initialization of the constrained NMF algorithm. The K-means algorithm performs this structured initialization and the information theoretic criterion of minimum description length (MDL) is used to estimate the number of clusters. We illustrate the method by a set of functional neuroimages from a motor activation study.

Clustered Cnmf for Fmri Data Analysis