Breast Cancer Detection Using Cnn On Mammogram Images

The most commonly occurring form of cancer is breast cancer and thereby requires urgent attention. In our paper, we work on these CT images of breast cancer to detect cancer in early stages. These images are classified into benign (non-cancerous) and malignant (cancerous) using our proposed method of Convolution Neural Network (CNN) following which medical care can be given. In our paper, image segmentation along with the image preprocessing is used which provides enhanced images. This helps in increasing the overall accuracy of the system. To conclude, the paper compares the trade-off between accuracy and time to train the network obtained in different execution environment i.e. Tensorflow and Matlab. Tensorflow gave 87.98% accuracy with training time of 6 h and Matlab gave 84.02% accuracy with training time of 45 min when the neural network's architecture remains the same.