Convolution Neural Network for Detecting Histopathological Cancer Detection

Mostly in Malaysia, the most fearful and deadlier cancer to women is breast cancer, thus it needs early detection for better diseases prevention. There are many ways of detecting the cancer and histopathological approach is one of methods in detecting the cancer cell which is known as mitosis. Mitosis is commonly being misclassified as non-cancerous cell due to its smaller size, jagged and dissimilarity texture and structure. Identification of such detail features requires domain-specific expert knowledge and is not an easy process. Therefore, advanced detecting methods are needed to ensure the cancerous cells are being detected for tiny scale cells. Lately, deep learning algorithms have been proposed to solve such complex images for better solutions in medical applications. In this study, the structure of deep learning of convolution neural network (CNN) is developed to improve the detection of cancer through histopathological images. The deep learning algorithm model is implemented using more than 3 connecting hidden layers such as Convolutional, ReLU, and Softmax layer. The performance evaluation of deep learning of convolution neural network (CNN) was been compared to other classifier algorithm and the result was better accuracy on detecting the mitosis and non-mitosis.