Performance evaluation of lossless ROI coding methods for telemedicine applications

A very small part of medical images, called Region of Interest (ROI), carries diagnostically important data. Thus, for telemedicine applications, we can transmit the diagnostically important data first and then transmit the remaining data in serial order, so that the ROI is reconstructed directly and the background is reconstructed gradually and doctors can stop at any instant, based on internet speed and clarity of background necessary to make diagnoses. This lossless region of interest functionality, called Maxshift method, has been added to JPEG 2000. However, the compression ratios that can be achieved are very limited. In this paper, we propose different possible ways to add ROI functionality to the wavelet–based methods JPEG 2000 and SPIHT and make comparative analyses between these methods and the Maxshift method.