BAT Optimization-Based Vector Quantization Algorithm for Compression of CT Medical Images

Abstract

Purpose The role of image compression is inevitable in telemedicine for the transfer of medical images. The diagnostic quality of the reconstructed image plays a vital role, and hence, the choice of compression algorithm is necessary. The medical images are analysed by the physicians for the diagnosis of anomalies. Procedures Linde Buzo Gray (LBG) is a classical algorithm for vector quantization; it produces a local optimum codebook that results in low compression efficiency. The efficiency of the vector quantization (VQ) relies on the appropriate codebook; hence, many research works based on optimization have developed for the generation of global codebook. This paper couples LBG with BAT optimization algorithm which generates an appropriate codebook. The optimization algorithm is employed not only for codebook design but also for the codebook size selection. Results The proposed BAT-LBG with dynamic codebook selection was compared with the classical LBG-VQ, JPEG lossless and BAT-LBG with static codebook selection. The algorithms were tested on abdomen CT images of five datasets, and performance evaluation was done by performance metrics like peak signal-to-noise ratio (PSNR) and mean square error (MSE). Conclusions The BAT-LBG compression with dynamic codebook size selection was found to produce efficient results when compared with JPEG lossless, classical LBG-VQ and BAT-LBG with static codebook selection. The quality of the reconstructed image was found to be good for BAT-LBG with dynamic codebook selection, and hence, it is efficient for transfer of medical images in telemedicine.