Detection of breast pathologies in digital mammography images by thresholding and mathematical morphology

Résumé

This paper proposes an algorithm for mass and micro-calcification detection by manual thresholding and prewitt detector. This algorithm has been tested using mammography images of different densities from multiple databases of a health clinic and images taken from the internet (40 images in total). The results are very accurate, allowing better detection of breast pathologies (mass and micro-calcification). Finally, the detection of breast pathologies was performed using as input a detection algorithm specially designed for this purpose. After segmentation by manual thresholding, morphological opening, morphological dilatation and Prewitt contour detection we have a demarcation of the masses and breast micro-calcification. The results obtained show the robustness of the proposed manual thresholding method. In order to evaluate the efficiency of our pathology detector, we compared our results with those in the literature and performed a qualitative evaluation with a rate of 98.04% for the detection of breast pathologies.  A radiologist from the health clinic evaluated the results and considers them acceptable to the CAD.