Skip to content

A Hybrid Machine Learning Approach using MRI Scans This project implements a hybrid interpretable machine learning pipeline for brain tumor classification and severity detection using MRI images.

Notifications You must be signed in to change notification settings

anubhav9369/Brain-Tumor-classification-and-severity-detector

Repository files navigation

🧠 Brain Tumor Classification & Severity Detection

A Hybrid Machine Learning Approach using MRI Scans

This project implements a hybrid interpretable machine learning pipeline for brain tumor classification and severity detection using MRI images. Unlike deep learning “black-box” models, this system focuses on feature-engineered, explainable, and computationally efficient diagnosis — making it suitable for real clinical decision support.

The system performs two tasks:

Tumor Classification → Glioma, Meningioma, Pituitary, No Tumor Severity Detection → Low, Mild, High, No Tumor

Demo

image image image image

📂 Dataset Description

👉 BRISC2025 MRI Dataset

Contains MRI images + segmentation masks.

Fine-grained classes (4-class): 0: Glioma 1: Meningioma 2: Pituitary 3: No Tumor

Coarse classes (2-class): Tumor-present Tumor-absent

Masks allow extraction of ROI-based geometric and intensity features.

📌 Key Features

Masks allow extraction of ROI-based geometric and intensity features. 🔹 Hybrid ML Pipeline (ROI + Global Features)

Uses a combination of: Geometric features (Area, Perimeter, Solidity, Circularity) Intensity statistics (Mean, Std, Min, Max) Texture features (LBP, HOG, Gabor filters) Gradient & edge-based features

🔹 Random Forest–based Dual-Classifiers

Two ML models trained using extracted features: Model Task Accuracy Random Forest Tumor Type Classification 90.0% Random Forest Severity Classification 59.7%

🔹 Interactive Gradio Deployment

The user can upload MRI images and receive: Tumor Presence Tumor Type Severity Visual & color-coded explanation

🚀 How to Run the Project

1️⃣ Install Dependencies pip install -r requirements.txt

2️⃣ Run the Gradio App python app.py

3️⃣ Upload an MRI Image

You will receive: Tumor detection Classification Severity level Explanation

🧪 Tech Stack

Python Scikit-learn (Random Forest) OpenCV (Image Processing) Albumentations (Mask Augmentation) Skimage (HOG, LBP) Gabor Filters Numpy, Pandas Gradio (Deployment UI)

About

A Hybrid Machine Learning Approach using MRI Scans This project implements a hybrid interpretable machine learning pipeline for brain tumor classification and severity detection using MRI images.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages