A Novel Method for Brain Tumor Classification Using Capsule Network

Brain tumours are one of the most fatal and common cancers in both children and adults. As a result, early detection of the exact type of brain tumour is critical for developing a precise treatment plan and predicting the patient's reaction to the treatment. In this field, there has recently been a spike of interest in using Convolution Neural Networks (CNN) to classify various kinds of brain cancer. However, CNNs require a large amount of training data and are incapable of managing input changes. Capsule networks (also known as CapsNets) are relatively new machine learning structures that were designed to address the weaknesses of CNNs. They are expected to improve deep learning solutions. Capsule networks are stable to rotation and affine translation, and therefore require much less training data, as is the case when processing medical image datasets such brain Magnetic Resonance Imaging (MRI) images, the proposed model provides better results.