A Comparison of JPEG and Wavelet Compression Applied to CT Images

Abstract

A study of image compression is becoming more important since an uncompressed image requires a large amount of storage space and high transmission bandwidth. This paper focuses on the quantitative comparison of lossy compression methods applied to a variety of 8-bit Computed Tomography (CT) images. Joint Photographic Experts Group UPEG) and Wavelet compression algorithms were used on a set of CT images, namely brain, chest, and abdomen. These algorithms were applied to each image to achieve maximum compression ratio (CR). Each compressed image was then decompressed and quantitative analysis was performed to compare each compressed-then-decompressed image with its corresponding original image. The Wavelet Compression Engine (standard edition 2.5), and ]pEG Wizard (Version 1.1.7) were used in this study. The statistical indices computed were mean square error (MSE) , signal-to-noise ratio (SNR) and peak signal-to-noise ratio (PSNR). Our results show that Wavelet compression yields better compression quality compared with ]pEG for higher compression. From the numerical values obtained we observe that the PSNR for chest and abdomen images is equal to 24 dB for compression ratio up to 31:1 by using ]pEG and 18 dB for compression ratio up to 33:1 by using wavelet. For brain image the PSNR is equal to 26 to 30 dB for compression ratio between 40 to 125:1 by using ]pEG, whereas by using wavelet the PSNR is equal to 22 to 34 dB for compression ratio between 52 to 240:1. The degree of compression was also found dependent on the anatomic structure and the complexity of the CT images.