Analysis Hybrid Approach for Multi-Objective Optimization-based segmentation of fuzzy contours in mammogram using artificial plant optimization algorithm

Abstract

It is observed that importance is given to the identification of tumor in case of medical image diagnosis. It has been found that in women breast cancer is increasing and due to which it has an impact in the death rate also. In medical domain radiologist generally uses the computer aided detection methods (CAD ). Medical image often contains the noise and some fuzziness in the picture border so the identification of the tumor is having problem and it becomes challenging task. by Owing to the noise and fuzziness of the picture border, identifying tumors on mammograms is challenging. Even though Active contour is an effective method for the segmentation but It is having problem related to local minima. The proposed method is a hybrid which is consist of multi objective optimization and Gravitational search algorithm (GSA) used for segmentation after that results are given to an Artificial plant optimization algorithm (APOA). The operations involve in this process includes some subprocesses primarily contour initialization is done for ROI followed by segmentation through GSA and multi-objective optimization and then APOA is applied. To evaluate the results two different databases are used mini-MIAS and DDSM databases. Results are also compared with other research work carried out earlier. It is observed that our proposed method gives better output then other.