Fractal analysis segmentation of medical images: Application to breast cancer mammography image

Abstract

This study discusses a new approach to the classification of mammographic medical images of the breast based on the calculation of the fractal dimension by the method of planar triangles of images segmented by Markovian modeling. The relaxation algorithm used is Iterated Conditional Modes (ICM). To do this, and based on the criterion of the energy of the neighborhood system cliques and some of the cluster cliques while relying on an energy function adapted to the irregular neighborhood models of t image. This new approach pushed us to orient this study towards the detection of pathologies to extract the tumoral part in the patient such that the fractal dimension is a useful tool in this study for the characterization and segmentation of mammographic images in the medical field. This approach has been used for the characterization of both healthy and pathological states of breast cancer. In order to apply this new approach, we used a database of mammographic medical images from the Kenitra-Morocco reproductive health reference center (CRSRKM) which contains mammographic images of the breast of normal and pathological cases.