Brain Tumor Classification Using Transfer Learning

One type of deadly disease is a brain tumor. To determine the presence of a brain tumor, it can be seen from an MRI image. In this research, we classified brain tumor MRI. The classification system uses transfer learning because only a few datasets are used. The Pre-Trained models used to extract features are VGG-16 and ResNet-50. Tests are carried out using several different parameters such as different batch sizes, optimizers, and learning rates. We evaluate the results using the confusion matrix. VGG-16 got the best accuracy of 0.96 using the Adam optimizer and ResNet-50 got the best accuracy of 0.94 using the RMSprop optimizer. From several different parameter variations, there is a relationship between parameter selection and accuracy results.