Transfer Learning Technique for Brain Tumor Classification from MRI Images

Brain cancer has the greatest death rate of any cancer, and it is difficult to classify tumors in medical imaging due to their aberrant proliferation of cells, forms, orientation, and location. Therefore, using computer-assisted methods is crucial to overcoming these limitations. Here, we proposed a model for classifying brain tumors from MRI images using deep transfer learning models. This work centers on three pre-trained models, specifically InceptionResNetV2, ResNet101V2, and MobileNetV2. We added extra layers after the base model output layer. From the InceptionResNetV2 model, we got the best accuracy with a low loss value. Our model's top validation accuracy was 98.86% and testing accuracy was 99.15%. Based on a comparison with different models, our proposed model has a high level of reliability in the early identification of brain tumors.