Performance Evaluation of Wavelet-Based Compression Techniques in Medical Imaging

The field of medical imaging is becoming popular and prone to research day-byday and there is an exponential rise of medical images. In telemedicine applications, we need to store and transmit the diagonastic images much frequently, so we need to compress the images. DICOM uses image compression standards like JPEG-LS, EZW etc to compress the medical images. There is always a tradeoff between quality and amount of storage (or bandwidth) needed. We want bitrate as low as possible and quality as high as possible. As there is a tradeoff, we may have to sacrifice one for the other. But in medical images, we cannot sacrifice quality too much as it may lead to wrong diagnosis. Presently we have very good compression techniques based on wavelets. In this paper we have analysed some wavelet based compression techniques like Embedded Zerotree Wavelet (EZW), Set Partitioning in Hierarchical Trees (SPIHT) and JPEG2000. These techniques are analysed on some medical test images. The performance evaluation is made on the basis of objective parameters like PSNR, BER etc. Some other parameters despite being objective in nature, take care of visual and structural quality, these parameters are WSNR, VIF and SSIM.