Wavelet versus JPEG (Joint Photographic Expert Group) and fractal compression. Impact on the detection of low-contrast details in computed radiographs.

RATIONALE AND OBJECTIVES

The aim of this study was to evaluate different lossy image compression algorithms in direct comparison.

METHODS

Computed radiographs were reviewed after compression with Wavelet, Fractal, and Joint Photographic Expert Group (JPEG) algorithms. For receiver operating characteristic (ROC) analysis, 54 thoracic computed radiographs (31 showing pulmonary nodules) were compressed with a ratio of 1:60. Five images of a test-phantom were coded at 1:13. All images were reviewed on a PC. Uncompressed images were reviewed at a PC and at a radiologic workstation (with image processing).

RESULTS

For thorax images, decrease of diagnostic accuracy was significant with Wavelets. Fractal performed worse than Wavelets. No ROC curve was observed for JPEG due to poor image quality. No diagnostic loss was noted comparing PC and Workstation review. For low-contrast details of the phantom, results of Wavelet compression were equal to uncompressed images. Fewer true positives and increased true negatives were noted with Wavelets though. Wavelets were superior to JPEG, and JPEG images were superior to Fractal. Workstation review was superior to PC review.

CONCLUSIONS

Only Wavelets provided accurate review of low-contrast details at a compression of 1:13. Frequency filtering of Wavelets affects contrast even at a low compression ratio. JPEG performed better than Fractal at low and worse at high compression ratio.