CLASSIFICATION OF BRAIN CANCER TYPE USING MACHINE LEARNING

Abstract

The Brain cancer is the most dangerous and found commonly in multitude of people in the younger stage and the adolescent stages. The early stage identification about the tumors in the brain and the appropriate type of the cancer would help the physicians in deciding the accurate treatments and further analyzing based on the responses from the patients to the treatment done. The paper puts forth the capsule neural network, the machine learning system that can be trained using a less number of dataset unlike convolutional neural network and is sturdy against the rotation or the affine conversions, to identify the type of cancerous tumors in brain at its early stage. The evaluation of the training and the testing accuracy of the proposed method for classification of the brain cancer type using the capsule neural network proves that Caps Net based classification have outperformed the convolutional networks.