Encoding of CT Image by Predicting PSNR Based on LSSVM

Abstract

With the development of remote telemedicine, the research of medical image compression with limited band is very important. In this paper, we analyze the features of kidney CT image and build LSSVM model to predict PSNR of ROI(Region of Interest) and BR(Background Region) of CT image. Then we propose a new method of encoding of CT image by predicting PSNR based on LSSVM, which has lower computational complexity and improves about 1/3 in encoding efficiency compared to the ROI encoding algorithm based on ISA-DWT. Besides, it can also achieve the effect of balancing ROI and BR as ISA-DWT algorithm.