Preliminary evaluation of adaptive histogram equalisation for digital mammography
KJ Kearfott, Sy Lin, JS Yakoubian · 1994
The present research is focused on (he application of adaptive histogram equalisation (AHE) to digital mammography. The effects of varying the enhancement factors and the contextual region size on the output histograms for mammographic phantom and breast images are studied. Phantom images are subjected to observer performance tests and the resulting pseudo-receiver operator characteristic (ROC) curves analysed. This analysis shows that the best performance is achieved with AHE, followed by illuminated films, and nonenhanced digital images, respectively. Subjective evaluation of four breast images by two radiology residents resulted in the observation that AHE may increase the difficulty of detection of some tumours but improve visualisation of others. Performance of the method should be unproved with the availability of a higher spatial resolution, lower noise image capture and display system.