RSNA 2005 

Abstract Archives of the RSNA, 2005


SSJ21-02

Discrete Cosine Transform JPEG Compression versus 2D JPEG2000 Compression: Visual Discrimination Model Image Quality Analysis

Scientific Papers

Presented on November 29, 2005
Presented as part of SSJ21: Radiology Informatics (Information Technology Variety PAC: Strategies and Tools)

Participants

Khan Mohammad Siddiqui MD, Presenter: Nothing to Disclose
Eliot Lawrence Siegel MD, Abstract Co-Author: Nothing to Disclose
Jeffrey Johnson PhD, Abstract Co-Author: Nothing to Disclose
Bruce Ian Reiner MD, Abstract Co-Author: Nothing to Disclose

PURPOSE

This study compares image quality as measured by just-noticeable differences (JNDs) and peak signal-to-noise ratio (PSNR) between discrete cosine transform (DCT) JPEG encoding and JPEG2000 wavelet encoding.

METHOD AND MATERIALS

The JPEG2000 compression standard is increasingly being adopted by industry in lieu of DCT JPEG encoding as a preferred “standard” method for 2D image compression despite a relative paucity of literature supporting this on an image quality basis.. In this study, 4 CT and 6 computed radiolography studies were compressed using the International JPEG Group DCT JPEG encoder and JPEG2000 (kakadu) standard compression with compression ratios ranging from 3:1 to 300:1. Image quality was measured in terms of JNDs, as assessed by the visual discrimination model (VDM) and PSNR.

RESULTS

DCT JPEG compression showed consistently lower image distortions (higher image quality) at lower compression ratios, whereas JPEG2000 compression was superior at higher compression ratios (>50:1). The crossover occurred at compression ratios that varied among the images included in the studies. The magnitude of any advantage of DCT compression at low ratios was often rather small (delta mean JNDs < 0.5). This advantage of DCT JPEG compression at lower ratios was generally not observed when image quality was measured in PSNR.

CONCLUSION

VDM analysis suggests that DCT JPEG compression may outperform JPEG2000 compression at ratios used in routine medical imaging practice. If a constant compression ratio were used then the difference could be observable with the DCT images noticeably superior at compression ratios of less than 50:1 in a side-by-side comparison. In continuing work, we are evaluating similar findings with other encoders.

DISCLOSURE

J.J.: Jeffrey Johnson works for Siemens.

Cite This Abstract

Siddiqui, K, Siegel, E, Johnson, J, Reiner, B, Discrete Cosine Transform JPEG Compression versus 2D JPEG2000 Compression: Visual Discrimination Model Image Quality Analysis.  Radiological Society of North America 2005 Scientific Assembly and Annual Meeting, November 27 - December 2, 2005 ,Chicago IL. http://archive.rsna.org/2005/4420285.html