Brain Cancer Classification using MR Images Based on VGG-19 Feature Extraction and Ensemble Classifier

One of the most studied disorders in modern medicine is the brain tumor. A correct cancer type diagnosis enables the professionals to select the best course of action and save the patient's life. The significance of an automated tumor classification system incorporating image processing should go without saying. The classification of different forms of brain tumors using magnetic resonance images (MRI) has been improved in this paper utilizing deep learning and a group of machine learning algorithms. The Brain Tumor MRI Dataset can be divided into four groups, which include three types of brain cancer (Glioma, Meningioma, and Pituitary), as well as non-cancerous, or normal type, tumors. o extract detailed information from the MRIpictures, a VGG-19 convolutional neural network is created. Then, to classify among various cancer kinds, these filtered deep characteristics are put into multi-class ML classifiers. The outputs of each ML classifier are thencombined to improve performance using a weighted average ensemble of classifiers. A total of 7022 MRI scansfrom four classes make up the dataset for the system. The accuracy of the Brain Tumor MRI dataset has improved, with a a Meningioma class accuracy of 97.88%, a Pituitary class accuracy of 98.76%, Glioma class accuracy of 99.78%, a normal class accuracy of 99.00% and an overall accuracy of 98.87%.