Enhanced Brain Tumor Classification using Ensemble Fusion Method

The human brain is the primary control center of the nervous system, and brain tumors are among the most fatal illnesses.Magnetic resonance imaging (MRI) has become the leading technique for early brain cancer diagnosis due to its exceptional ability to visualize soft tissues.Manual interpretation of brain MRIs is susceptible to errors and is significantly influenced by the radiologist's fatigue and experience.In the past decade, deep convolutional neural networks (CNNs) have been an effective instrument for brain tumor classification in medical imaging.This research is presented an innovative fusion approach that combines pre-trained deep CNNs with a transformer-based model to improve brain tumor classification.By integrating three separate MRI datasets into one unified dataset, the study has been provided with a comprehensive resource that has included normal brain scans and three categories of brain tumors: glioma, meningioma, and pituitary.The ensemble method employed three pre-trained models (VGG19, InceptionV3, and ResNet50), achieving a 93.5% accuracy rate.Finally, after applying the ensemble technique, the Vision Transformer model is utilized and has increased the accuracy to 98.32%.This fusion technique shows great potential for enhancing brain tumor diagnosis and patient care.