Tumor edge detection in mammography images using quantum and machine learning approaches

Automatic processing and analysis of medical images may provide to doctor valuable assistance for diagnostic and therapeutic practice. In this work, the problem of breast cancer edge detection is addressed. We are faced with a challenging task considering the breast tissue specificities and the inevitable mammogram noise. To meet this challenge, we propose novel approaches involving quantum genetic algorithm and support vector machines. The first method uses the quantum genetic algorithm to solve a multilevel thresholding problem based on Tsallis entropy. In the second method, the support vector machines are trained, in different ways, on a simulated image in order to be able to detect breast cancer edge. The proposed approaches are compared to some standard methods of edge detection on a sample of mammographic images taken from a well-known benchmark databases. The evaluation results obtained by PSNR, SSIM and FSIM metrics demonstrated the effectiveness of the proposed approaches.