Design a Computer Based Detection System For Medical Images To Detect Cancer
Fatemeh Maleki, Mehdi Nooshyar, Gholamreza Zare Fatin, Homayun Ebrahimian, A Agha Golzadeh · 2015
One of the serious reasons of death among women is breast cancer. Early diagnosis can help to women’s quality of life. In this context, mammography images have a most significant effectiveness. In fact, they are known as the main test used for screening and early diagnosis. Recently, computer aided mass detection techniques was an active field of research and some of these studies showed a promising future. In this study, we present a method based on the harmony search algorithm (HSA). HSA is a new optimization algorithm which is inspired from musicians developing new harmonies while playing. The proposed method employs random samples as candidate solutions from the search space inside the image histogram. Here, we used this algorithm for segmenting mammography images and skin lesion images by considering to the objective function that is utilized with the Kapur’s method. This objective values are achieved until an optimal solution is found. Experimental results show the high accuracy of the proposed approach for the breast cancer segmentation and skin lesion. To analyse the efficiency of the proposed method, three metrics (CDR, FAR and FRR) are employed considering the coincidences between the segmented and the original images and the results show good efficiency for the proposed algorithm. Keywords: Image segmentation, Harmony search algorithm, Kapur’s method, breast cancer