Transfer Learning Based Brain Tumor Multi-Class Classification

Brain is an important organ which monitors the functions of all other organs and regulates all the activities of a human. Uncontrolled cell development condition in a brain is considered as brain tumor, can disrupt the brain's normal functionality. Exact prediction of brain tumor is still a challenge, due to the noise and indistinguishable intensity with the neighbouring cells. Precise and early diagnosis of brain tumor will save the life of person. In this research, transfer learning based brain tumour detection, localization and classification models are proposed by using VGG16 and ResNet50 on MRI images. The accuracy obtained with VGG16 and ResNet50 models are 96.36% and 97.72% respectively. The performance of ResNet50 is 1.36% better than the VGGnet16.