Breast Cancer Detection by Artificial Intelligence Technologies

One of the cancer that gives high in fatality rates is breast cancer. Current method used to detect the breast cancer need the existing radiologist which makes it costly and time consuming. Possible solution for this cancer is to detect it early which can be done by Artificial Intelligence (AI). This research will propose an end-to-end system that could increase the accuracy in detecting and in classifying the breast masses. From the mammographic images, it is first undergone the pre-processing stages to eliminate background and noise. Then the pre-processed image was segmented by the active contour and followed by deep learning convolutional neural networks for features extraction. Principle Component Analysis (PCA) technique will be applied to select the necessary features as input to the Support Vector Machine (SVM) for determining the class of cells (normal or abnormal). Lastly, 5-fold cross validation techniques were executed to validate the results and obtain the average reading for both training and testing dataset. Proposed system was tested on the DDSM and MIAS dataset and obtain 77%, 90% and 65% for accuracy, sensitivity and specificity respectively. Percentage accuracy was greatly affected by the low percentage of specificity shows by this proposed system which is 65% only.