Integrating Deep and Machine Learning Techniques for Brain Tumor Detection

The traditional method to identify cancerous brain tumors is magnetic resonance imaging (MRI). These days, radiologists can diagnose brain tumors with greater accuracy because of the advancements in diagnostics aided by computers. In this study, a dataset of 7023 MRI scans of the human brain, broken down into four kinds - Glioma (G), Meningioma (M), Pituitary (P), and No tumor (N) was employed. Various trials to detect brain cancers using Deep Learning (DL) models, Machine Learning (ML) techniques, and a combination of DL and ML approaches were conducted. In the hybrid approach, the VGG16 model and SVM algorithm were combined. The dataset was trained and features in the image were extracted using the deep learning model VGG16. Then, SVM (Linear, Polynomial, and RBF) was utilized to classify the image into four classes. Every system has produced better outcomes. With 91% accuracy, the hybrid approach VGG16 and SVM (linear) model performed the best and detected tumors successfully.