Symmetric Wavelet Analysis Toward MRI Image Compression

Abstract

Every year trillions of medical imaging data are generated through different imaging techniques. Demand for communication of this data through telecommunication networks using the internet is growing rapidly for various applications, hence requires a lot of memory space for storage, and high bandwidth for transmission. Efficient storage and transmission of this data is a problem to be resolved. Compression of this huge medical imaging data would be a suggested solution to this which primarily considers neighboring pixel redundancy. The proposed work includes calculation of compression on a set of ten different real time MRI (512 × 512) images acquired by the neural department of the hospital. The compression of the MRI image includes decomposing the image by a chosen discrete wavelet transform’s (DWT) mother wavelet function which is followed by its encoding through set portioning in the hierarchal tree (SPIHT). After applying each of the compression methods their performance is quantified in terms of Compression Ratio (CR), Peak Signal to Noise Ratio (PSNR) and Bits per pixel (BPP). The calculated performance parameters have been compared in order to get a higher compression ratio with considerably high PSNR values from all of the applied mother wavelets. DWT (Discrete Wavelet Transform)-based symmetric mother wavelets like Haar, bior4.4, bior6.8, sym4 and sym20 have been tested and compared in virtue of comparative analysis of them. Among all of the used wavelets, sym4 and sym20 have not been explored yet for compression. MATLAB and wavelet toolbox have been used for implementation of algorithms. Haar wavelet gives a higher compression ratio and a good PSNR value after compression. The images compressed with these takes less time for uploading and downloading also can be reconstructed without a loss of diagnostic details for utilization of telemedicine services.