BRAIN TUMOR CLASSIFICATION USING DEEP LEARNING MODELS

Brain tumor detection is a vital step in diagnosing life-threatening neurological conditions.Manual analysis of brain MRI scans is often time-consuming, subjective, and prone to human error.With recent advancements in artificial intelligence, particularly deep learning, automated methods can be leveraged to assist radiologists and improve diagnostic efficiency and accuracy.This project presents a deep learning-based solution for the automatic detection of brain tumors using a dynamic ensemble of Convolutional Neural Networks (CNNs).The ensemble integrates multiple powerful architectures, including MobileNetV2, EfficientNet, ResNet50, Inception, and DenseNet, enabling the system to adapt effectively to diverse MRI scan inputs.To enhance model performance, image preprocessing and feature extraction techniques such as intensity normalization and Region of Interest (ROI) segmentation are applied.A dynamic model selection mechanism evaluates the prediction confidence of each CNN and optimally combines their outputs to generate the final classification result.The system is trained and validated using a publicly available brain MRI dataset containing both tumor and nontumor images.Performance evaluation using metrics such as accuracy, precision, recall, and F1-score demonstrates that the ensemble approach significantly outperforms individual CNN models.Furthermore, Grad-CAM visualizations are employed to improve model explainability by highlighting the regions of the MRI scans that most influence the model's decisions.Overall, this work demonstrates that deep learning, when combined with ensemble learning and explainable AI techniques, can serve as a powerful and reliable tool for medical image analysis and early brain tumor detection.