Microcalcifications detection in mammograms based on Ant Colony Optimization and Markov Random Field

Mammography constitutes a credible technique for the detection of breast cancer. Early detection of microcalcifications in breast tissue, which is an indication of developing breast cancer, facilitates prompt intervention averting fatalities associated with this type of disease. It is, however, difficult for practitioners to pinpoint effectively the affected regions of the breast. This paper proposes a novel method for microcalcifications detection based on a hybrid metaheuristic approach using Ant Colony Optimization and Markov Random Field approaches. Markov Random Fields are used to model spatial relation between different neighboring pixels via an energy function. This energy function is then optimized using an Ant colony so to find its minimum value. The proposed algorithm is tested on the mini-MIAS mammogram database. The obtained results, evaluated using Borsotti criterion, show the good accuracy and efficiency of our approach.