Non-linear quality evaluation index for mammograms

Mammographic image enhancement algorithms form an essential part of Computer Aided Detection Techniques (CAD) for breast cancer. The performance measures for evaluation (of these enhancement algorithms) should be such that they not only assess the degree of improvement of image from original in terms of pixels related values but also measure the utility of mammogram in terms of its improvement for diagnostic purposes. This paper introduces a new quality index for mammogram enhancement algorithms, which could be used as tool to evaluate the performance of CAD techniques (for breast cancer). The evaluation procedure involves extraction of statistical parameters from the foreground and background regions of the targeted region-of interest in mammograms. The extracted features are then combined using non-linear operators to yield the proposed quality index. Mammograms from MIAS database are used to validate the performance of the proposed index during simulations.