wandiyajames-arch/brain-tumor-classification-teams-gedeon-james

deep learning competition at AIms senegal # 🧠 Brain Tumor Classification from MRI Scans > Deep learning pipeline for brain tumor detection and classification using MRI images — from raw data cleaning to transfer learning with performance comparison. --- ## 📌 Project Overview Brain tumors are abnormal cell growths within the brain. Due to the rigid structure of the skull, any abnormal growth can increase intracranial pressure and potentially lead to severe neurological damage. **Early detection and accurate classification are critical** for selecting appropriate treatment strategies and improving patient outcomes. This project builds a complete end-to-end deep learning pipeline to classify brain MRI scans into 4 categories: | Class | Description | |---|---| | 🔴 Glioma | Tumor originating from glial cells | | 🟠 Meningioma | Tumor arising from the meninges | | 🟡 Pituitary | Tumor in the pituitary gland | | 🟢 No Tumor | Healthy brain scan | --- ## 📂 Dataset **Source:** Brain Tumor MRI Dataset — Kaggle ``` dataset/ ├── Training/ │ ├── glioma/ (1,400 images) │ ├── meningioma/ (1,400 images) │ ├── pituitary/ (1,400 images) │ └── notumor/ (1,400 images) └── Testing/ ├── glioma/ (300 images) ├── meningioma/ (300 images) ├── pituitary/ (300 images) └── notumor/ (300 images) ``` ### ⚠️ Real-World Data Challenges This dataset reflects real clinical imaging conditions: - **Noisy images** — MRI artifacts, varying orientations, different slice planes - **Blurry or poorly framed** scans - **Inconsistent resolutions** across images - **Class imbalance** between tumor and non-tumor samples These challenges are treated explicitly in the preprocessing pipeline. --- ## 🛠️ Pipeline ### Step 1 — Data Collection & Exploration - Download dataset from Kaggle - Analyze class distribution - Identify and visualize problematic images (blurry, corrupt, artifacts) ### Step 2 — Cleaning & Preprocessing - Remove corrupted or overly blurry images - Resize all images to ** …