Performance Analysis of Wavelet Based Roi Medical Image Compression Using Optimum Subband Shift Coding

This paper presents an approach for an Enhanced Image Compression Method using Optimal sub-band Shift based SPIHT (Set partitioning in Hierarchal Trees)Algorithm. This is based on the progressive image compression algorithm, SPIHT which is an extension of Shapiro's embedded Zero tree Wavelet Algorithm. The proposed Optimal Sub-band Shift (OSS) Algorithm overcomes the difficulty of Embedded zero Wavelet EZW that loses its efficiency in transmitting lower bit planes. In this paper, we include integer wavelet transformation and region of interest coding to Partial EZW and hence make it more superior to EZW and SPIHT Algorithm and hence proved from the results. Medical Images are to be compressed in such a way that diagnostically significant regions (ROIs) are compressed in lossless manner where as the rest of the image can be compressed using lossy techniques. In this paper we present an efficient method of separation of ROI in Medical Images and apply lossless wavelet compression to ROI and lossy compression in Non ROI regions. The arbitrary shaped Region of Interest (ROI) in medical image is separated using Hierarchical Pyramid Neural Network. The ROI is compressed using lossless Integer Wavelet Transform, where as the remaining image (Non ROI) is compressed using Integer Wavelet Transform with OSS. Comparisons of these methods are brought out in terms of Peak Signal to Noise ratio (PSNR), Mean Square Error (MSE) and Compression Ratio (CR). Experimental results show that the image compression using ROI IWT-OSS achieves better results than the conventional lifting based wavelet transform with OSS.