Progress with an “all-wavelet” approach to image enhancement and de-noising of direct digital thorax radiographic images

Abstract

This paper describes the current status of our program of work in the area of digital image enhancement using wavelet-based multi-scale processing. We are developing an all-wavelet image processing algorithm to enhance the quality of direct digital thorax images, by the manipulation of the data within scale-specific sub-bands. This method avoids the presentation compromises which may result from the global application of unsharp mask based image enhancement methods which are commonly used in medical imaging. This is achieved by applying specific processing to image components according to their scale. In particular contrast enhancement, de-noising and sharpening stages are all tailored to the noise and feature characteristics of Thoravision digital chest X-ray images. Whilst our experiments to date convince us that processing digital X-ray images within the wavelet domain is a useful tool for improving diagnostic image quality, our concerns now focus on the subtlety of this processing, since it can be prone to artefact generation if applied incorrectly. We are also interested in optimising the way in which wavelet based image enhancement must be presented to our clinical colleagues to meet their diagnostic needs.