Role of Capsule Network in Brain Tumor Prediction: A Novel Approach

In today's time several deadly tumors are present like breast cancer, liver cancer, oral cancer, etc. Brain tumor is also one such disease. It is a biological condition in which the brain's normal functioning is disturbed by uncontrolled growth of the cells in the brain. It can cause lifetime impairment and if not detected and treated on time it can cause death also. That is the reason it becomes important to detect it on time. MRI is the most common way to collect the image dataset. Inspecting MRI images manually is very tedious work and error-prone. There is a continuous search for intelligent methods that can assist tumor detection. Deep Learning is a subfield under machine learning that helps medical experts read the patterns of brain images and make predictions accurately. Several Deep-learning methods are present today, used to read the images i.e. AlexNet, DenseNet, ResNet, VGG16, VGG19, etc. Convolutional Neural Network (CNN) is one of the most popular techniques used to analyze MRI images and detect the tumor. However, the major drawback of CNN is its MaxPooling layer. It is translation invariant. That is the reason there is a need for a technique that can overcome this drawback. This paper will propose the technique of Capsule Network to detect the tumor and will be compared with the existing methods for comparison.