QUANTIFICATION OF MR BRAIN IMAGES BY MIXTURE DENSITY AND PARTIAL VOLUME MODELING

This paper addresses the problem of automatic quantification of brain tissue by utilizing single-valued (single echo) MRI brain scans. It is shown that this problem can be solved without classification or segme ntation, a method that may be particularly useful in quantifying white matter lesions where the range of values associated with the lesions and the white matter may heavily overlap. The general technique utiliz es a statistical model of the noise and partial volume effect together with a finite mixture density description of the tissues. The quantif ication is then formulated as a minimization problem of high order wit h up to six separate densities as part of the mixture. This problem is solved by tree annealing with and without partial volume utilized, th e results compared, and the sensitivity of the tree annealing algorith m to various parameters is exhibited. The actual quantification is per formed by two methods: a classification-based method called Bayes quan tification, and parameter estimation. Results from each method are pre sented for synthetic and actual data.