Brain Tumor Detection Using Convolutional Neural Network

One of the most significant and difficult issues in medical image processing is brain tumor segmentation (BTS) as human classification might cause improper diagnosis and prognosis. Furthermore, when a lot of data needs to be contributed it is an annoying task. It becomes challenging to identify tumor regions in images as brain tumors (BT) exhibit a great level of visual variation as well as mimic healthy tissues. Using the fuzzy C-Means clustering (FCM) approach, we proposed in the current work to exclude brain cancers from the two dimensional MRI. Next came convolutional neural networks (CNNs) and traditional classifiers. Utilizing a real-time dataset which had a diversity of tumor locations, sizes, shapes, image intensities, this experimental study was carried out. Next, we discussed CNNs, which are built with Keras and Tensorflow since they outperform the conventional ones. In this investigation, CNN attained an amazing accuracy rate that is 87.87%. This study's primary objective is to make the differentiation between normal and aberrant pixels utilizing statistical along with texture-based criteria.