Image Segmentation using Mask R-CNN for Tumor Detection from Medical Images

One of the dreadful diseases that the world encounters today is brain tumor. When abnormal cells form in the brain, it is called a brain tumor. There are lot of variations in sizes and location of tumor, and hence this makes it really hard for a complete understanding of tumor. Radiologists can easily diagnose the disease with the help of medical image techniques, but making this process automatic is obviously useful. Magnetic Resonance Imaging (MRI) is the most effective method for detecting brain tumors where, MRI images are trained and tested in order to detect the tumor. The automated system would be able to detect and pin-point the exact location of the tumor in an MRI image. In this project, the automated system is built using Mask Regional-based Convolution Neural Network (Mask R-CNN) which segments the abnormal tissues in the brain and mask is applied over the segmented region. Mask R-CNN has the best performance compared to other methods such as R-CNN (Regional-based CNN), Fast R-CNN and Faster R-CNN.