Expectation maximization (EM) algorithm has been used widely for calculating the maximum likelihood (ML) parameters in the statistical segmentation of brain magnetic resonance (MR) images. Since standard EM algorithm is time and computer memory consuming, which makes the segmentation impractical in many real-world situations. In order to overcome this, a novel statistical histogram based expectation maximization (SHEM) algorithm is presented in this paper. The method is developed for segmentation of the single-channel brain MR image data by combining the SHEM algorithm and the region-growing algorithm, which is used to provide the priori knowledge for the segmentation. The performance of the SHEM based method is compared with that of popular applied fuzzy c-means (FCM) segmentation. The experimental results show that the proposed method is robust and can reduce the computing time and computer memory largely.