Segmentation of ROIs/VOIs from small animal images for functional analysis

Abstract

Segmentation of anatomical structures and tumors is an important prerequisite for the interpretation and further quantitative analysis of tomographic images. A method for segmenting regions-of-interest or volumes-of-interest (ROIs or VOIs) from small animal MRI images for functional analysis of co-registered PET images was developed. The method consists of automatic image enhancement, semi-automatic segmentation and a coded contour-based interface to region-based functional analysis software. Two semi-automatic segmentation approaches, a dynamic programming-based and an active contour-based approach, were compared regarding user interaction and segmentation consistency. The results obtained from six MRI data sets of different resolution and image quality show that both semi-automatic approaches are a good choice to deal with different kind of images, image quality and regions.