Modeling fiber type grouping by a binary Markov random field.

A new approach to the quantification of fiber type grouping is presented, in which the distribution of histochemical type in a muscle cross section is regarded as a realization of a binary Markov random field (BMRF). Methods for the estimation of the parameters of this model are discussed. The first order BMRF, which is used in this article, contains 2 parameters: alpha and beta. The parameter beta is of prime importance, as it is an interaction parameter which governs the degree of type grouping. The value of this parameter is estimated for 9 muscle biopsies. The interpretation of the results is discussed.