Normal and Abnormal Masses Detection in Mammography Images Using Deep Convolutional Neural Network (DCNN)

One of the most important and influential ways to diagnose breast cancer, especially in the early stages of the disease, is mammography. Mammography images are usually of low quality due to the complexity of breast tissues, the similarity between cancerous masses and normal tissues, the different sizes and shapes of the masses, and X-ray radiation. Therefore, it is very difficult to detect lesions, especially in the early stages; Because some mass lesions are embedded in natural tissues and have weak margins or vague margins. The proposed method in this study is to present an architecture based on a deep convolutional neural network to detect cancerous masses in mammography images, which ultimately leads to classifying the masses into normal and abnormal classes. The training of the proposed network begins with the modification of the images in the pre-processing stage in order to perform more accurate drawings with high resolution on the images and finally to improve the accuracy and sensitivity of separating the mass from the breast tissue for correct diagnosis. Python programming language and TensorFlow library have been used in the Windows environment to implement the proposed method. To ensure the performance of the proposed method, the cross-validation method was used and the obtained results were evaluated by the criteria of precision, accuracy, and sensitivity. The results obtained with an accuracy of 97.67% indicate the improvement of the diagnosis accuracy and the cost reduction in the diagnosis process