Brain Tumor Classification using a Convolutional Neural Network and Different Optimizers

In this paper, we designed a brain tumor classifier using a convolutional neural network (CNN) which is a subfield of deep learning. The proposed framework aims to classify three different brain tumor classes which are meningioma, glioma, and pituitary from the publicly available Figshare brain tumor dataset. In this regard, VGG-16 CNN architecture is trained for three brain tumor classes and the results are very promising. This architecture is experimented with six different optimizers namely Adaptive Moment Estimation (Adam), Stochastic Gradient Descent (SGD), Root Mean Square Propagation (RMSprop), Adaptive Gradient (Adagrad), Adadelta, and Adamax. Simulation results demonstrate that the proposed VGG-16 architecture achieved 97.15% on the test dataset with respect to 70% training, 15% validation, and 15% testing data.