Project Overview
This project addresses one of the most critical challenges in medical imaging — early and accurate detection of brain tumors. Using deep convolutional neural networks trained on thousands of labeled MRI scans, the system can classify tumor types with over 95% accuracy.
The pipeline covers image preprocessing, model inference, explainability via Grad-CAM heatmaps, and a Django REST API that medical professionals can query in real time. The frontend dashboard provides visual overlays on MRI images, confidence scores, and patient report generation.
- 95.4%
- Accuracy
- 10K+
- Dataset Size
- CNN
- Model Type
- <2s
- Inference
Gallery
Key Features
- Custom CNN architecture with residual connections
- Transfer learning with EfficientNetB4
- Grad-CAM explainability for medical validation
- DICOM image format support
- REST API with JWT authentication
Technology Stack
PythonTensorFlowKerasOpenCVDjangoReactPostgreSQLDocker
