Performance Analysis of Compression Techniques Using LM Algorithm and SVD for Medical Images

Abstract

This paper presents the techniques for compression and decompression of MRI, CT and X-ray images. These medical images are the pictorial representation of inner parts of human body which are used for the analysis of critical diseases. As the vast amount of data is required to store medical images for future reference of the patients and for the transmission, there is need to use image compression methods without reducing the quality of information. Two methods of compression i.e. back propagation neural network with LM training algorithm (BPNNLM) and Singular Value Decomposition (SVD) are used in this paper. The results of these two techniques are compared with respect to the performance metrics of Peak Signal to Noise Ratio (PSNR), Mean Squared Error and Structural Similarity Index Measurement (SSIM). From the results, it is observed that SVD image compression technique based on singular values provides more PSNR, less MSE and better SSIM values compared to BPNNLM Technique.