Interactive Multi-scale Contrast Enhancement of Previously Processed Digital Mammograms

摘要

Manufacturers of digital mammography systems include proprietary and unique post-processing algorithms to enhance the image contrast for diagnostic presentation. We use a wavelet-based enhancement method to both automatically and interactively adjust the contrast of these previously processed mammograms. Apart from enhancing small but relevant structures in dense breast tissue, this method allows for the automatic homogenization of cross-vendor contrast appearance, supporting current-prior comparison tasks in the screening and diagnostic setting when different systems were used for acquisition. An optional smooth crossfading between originally processed and enhanced images can be used interactively to adapt the presentation view to individual situations, for example to facilitate the detection or interpretation of mammographic features in dense breast tissue. A user preference study with experienced readers revealed that our work might positively impact prior-current comparisons and the diagnosis of mammographic images.