Accurate liver cancer classification is vital for effective treatment and patient prognosis. This project utilizes SimSiam, a self-supervised learning approach, to improve classification accuracy of liver tumors. We integrate SimSiam with three CNN classifiers - Inception, Xception, and ResNet152 - and pretrain them using two loss functions: MSE and COS. Our tests show consistent improvements in classification accuracy across all models. The dataset includes CT scans of 460 patients with various types of liver cancers.

• Clone the repository: git clone https://github.com/your_username/liver-cancer-classification.git
• Navigate to the project directory: cd liver-cancer-classification
• Install dependencies: pip install -r requirements.txt
• Train and evaluate the model: python train.py
• Evaluate the model: python evaluate.py

• Python 3.8+
• TensorFlow 2.x
• NumPy
• scikit-learn

Feel free to open an issue or submit a PR for improvements. To access the datasets, please contact at [email protected]