Extracting Regions of Interest in Biomedical Images

Abstract

Regions of interest (ROI) usually means the meaningful and important regions in the images. The use of ROI can avoid the processing of irrevelent image points and accelerate the processing. Extraction of regions of interest from images is an important and unsolved topic in the image processing area, especially in biomedical image processing area. In this paper, a feasible and fast constrast-based method is proposed to extract regions of interest in biomedical images. Motivated biologically, this approach simulates the bottom-up human visual selective attention mechanism, computes the global contrast of each pixel and constructs the saliency map.Then by segmentating the saliency map using a dynamic threshold, the regions of interest can be extracted. This approach has been tested on many medical images. Computer experimental results show that the proposed method can extract regions of interste in biomedical images rapidly and precisely, and indicate that the proposed approach is effective and practicable.