Abstract
Nowadays, manually examining the enormous volume of MRI (magnetic resonance imaging) images and identifying a brain tumour is an extremely time-consuming and imprecise task. It might have an impact on the patient's appropriate medical care. Again, because there are so many image datasets involved, it can take a very long time. The segmentation of tumour locations becomes challenging due to the apparent similarities between brain tumour cells and normal tissue. In this study, we presented a method for segmenting brain tumours from 2D MRIs using a convolutional neural network, which is then followed by conventional classifiers and deep learning techniques. To properly train the model, we have collected a wide range of MRI pictures with variable tumour sizes, locations, forms, and image intensities. To further validate our work, we used SVM classifier and additional activation algorithms (such as softmax, RMSProp, sigmoid, etc.). We use TensorFlow and Keras to build our suggested solution because Python is an effective programming language for quick work. The accuracy of the brain tumour detection in MRI images made possible by our CNN-based model would greatly speed up the pace of treatment.