Determination of the Region of Interest in the Compression of Biomedical Images

Compression of medical images should be lossless to preserve details and to avoid wrong diagnosis. However, to achieve high compression ratios while maintaining image quality, only regions of interest (ROI). i.e., abnormal regions need to be losslessly compressed. The rest of the image if lossy compressed can lead to a reasonable overall compression ratios. In this paper, two techniques for determining and extracting the ROI of an MRI brain image are proposed. The first technique is based on the mean squared error (MSE) between a normal and an abnormal brain images. The region that shows high MSE is the ROI. In the other technique, selection of the ROI is based on the gray levels of the image. Any region that possesses a gray level that noticeably differs from the surroundings is the ROI.