Ensemble of Deep Learning Models and Machine Learning Classifiers for the Classification of Brain Tumors

Brain tumors are typically one of the more complex health issues in the world. It could result in death if not addressed in the beginning stages. It is important to adopt an automated system to diagnose tumors. Here this article proposes a model to automate the system. The proposed model unites two separate Convolutional Neural Network (CNN) architectures, VGG16 and ResNet-50, to improve the accuracy and robustness of brain tumor detection. The model is trained and evaluated on a comprehensive dataset of brain tumors for encompassing three distinct tumor types: pituitary, meningiomas, and glioma tumors. The proposed model uses MRI images for preprocessing and feature extraction through pre-trained CNNs. Various machine learning classifiers like SVM, KNN, LR, and RF are applied. Finally, an ensemble of classifiers, including LR, RF, and XGBoost, is used for tumor detection and classification. The model is evaluated on brain tumor dataset which is publicly available. The output demonstrates that the ensemble approach outperforms the individual classifiers, achieving an overall accuracy of 93.43%. The primary aim is to equip doctors with high-quality software for tumor identification and understanding their origins. This software is designed to save patients' time, offer timely solutions during the early stages, and enable swift consultations.