Deep Learning-based Hybrid CNN-VGG16 Model for Brain MRI Tumor Classification

This paper discusses the use of the VGG-16 model and CNN architecture for classifying brain tumors in MRI images. The focus is enhancing the existing accuracy attained in the diagnosis of tumors through imaging modalities. The model differentiates MRI images into four types: glioma, meningioma, pituitary tumors, and no tumors. Features are extracted using the VGG16 model with CNN layers that are custom-built for Brain Tumor (BT) image features. This hybrid model can take advantage of transfer learning, which automates feature extraction and enables the model to greatly benefit from not requiring manual feature extraction. The model was trained and validated on the Kaggle BT-MRI dataset, which required some preprocessing, such as normalization, image resizing, and augmentation, because the dataset was small. To optimize the model, we tested the Adam optimizer with the cross-entropy loss function and a batch size of 32 for 10 epochs to allow proper convergence. The results indicate that the hybrid model has better performance in terms of accuracy and robustness when tested against the CNN and VGG16 models. This research advances the development of computer-aided diagnostic (CAD) systems by enabling better detection and classification of brain tumors, thus making the systems more useful in clinical environments.