Segmentation of 3D MicroPET Images Using The Mixture Method

Abstract

The segmentation of 3D microPET image is one of the most important issues in tracing and recognizing the gene activity in vivo. In order to discover and recover the dynamic activity of gene expression, reconstruction techniques with higher precision and fewer artifacts are necessary. To improve the resolution on microPET images, the maximum likelihood estimate (MLE) by the EM algorithm is applied. In addition, advanced statistical technique based on the mixture model is developed to segment the reconstructed images. In this study, the new proposed method is evaluated with simulation and empirical studies. The performance shows that the proposed method is feasibly promising.