During screening, diagnosis or therapy, patients may undergo examinations with mammography systems from different vendor. Due to proprietary and unique post processing algorithms the images can show large variations in their multi scale contrast appearance. These differences cannot be adjusted with common window level handling tools.
We use a wavelet-based filter method for an automatic homogenization of the multiscale contrast of mammograms. All image information is preserved, while the appearance is adjusted. Thereby our solution supports current prior comparison tasks in clinical settings when different systems were used for acquisition.
- Easy inclusion into a standard reading workflow.
- Fast computation during preprocessing; below 3 seconds per image.
- Can also be used to adapt the image to individual viewing customs or tasks.
User preference study
- In a study with experienced readers the homogenized setting has always been chosen over the unprocessed setting.