This project demonstrates a hybrid Quantum + Classical deep learning model for brain tumour classification using MRI images. The architecture uses EfficientNet-B0 + a 4-qubit quantum variational layer.
Accurate brain tumour detection from MRI scans can speed up diagnosis and improve patient outcomes. This project explores a hybrid approach — classical convolutional feature extractors enhanced by a small quantum variational layer — to improve classification robustness and generalization on medical images.
This project is based on the following published research paper:
Title: Brain Tumour Detection Using Quantum Convolutional Neural Networks (QCNN)
Conference: IEEE Odisha International Conference on Information Technology (OCIT 2025)
Publisher: IEEE
DOI: https://doi.org/10.1109/OCIT66168.2025.11400476
This repository contains the implementation corresponding to the published work.
- Hybrid QCNN architecture (EfficientNet + Quantum Layer)
- Binary brain tumor classification
- Quantum circuit built using PennyLane
- Gradio UI for predictions
- Safe-cleaning of model checkpoints
- GPU supported
- Achieved high accuracy on binary MRI tumour classification
- Quantum variational layer improves feature separability
- EfficientNet-B0 extracts strong spatial representations
- Stable training with smooth loss convergence
- Works efficiently on GPU (recommended)
- Produces confident predictions for both tumour and no tumour classes
- Lightweight architecture suitable for deployment
Deep Learning:
- PyTorch
- EfficientNet-B0
Quantum:
- PennyLane
- Strongly Entangling Layers (4-qubit)
Utilities:
- NumPy, Pillow, Scikit-Learn, Torchvision, Gradio
Kaggle dataset by the author: https://www.kaggle.com/datasets/skarthik112/karthik-braindataset-mri
Folder structure:
brain_Tumor_karr/
├── yes/ → tumour present
└── no/ → no tumour
| Class | Precision | Recall | F1-Score |
|---|---|---|---|
| Yes (Tumour) | 0.98 | 0.97 | 0.98 |
| No (Normal) | 0.99 | 0.99 | 0.99 |
Overall Accuracy: 0.996
Brain-Tumor-QCNN/
│
├── train_qcnn.py
├── predict_qcnn.py
├── README.md
├── requirements.txt
├── .gitignore
│
└── assets/
├── yes_result.png
└── no_result.png
S. Karthik
Developer & Research Student
Brain Tumour Detection using Quantum Convolutional Neural Networks (2025)
Follow the steps below to set up and run the QCNN model.
git clone https://github.com/Karthik7661/Brain-Tumor-QCNN.git
cd Brain-Tumor-QCNN
# 2 Install Dependencies
pip install -r requirements.txt
# 3 Train the Model
python train_qcnn.py
# 4 Run the Gradio Prediction App
python predict_qcnn.py