⚡ Quick Summary
This study introduces a novel Federated Learning (FL) framework integrated with blockchain technology to enhance privacy and security in retinal imaging. The proposed system demonstrated robust performance against adversarial attacks, achieving an AUC of 0.868 for myopic macular degeneration detection and 0.970 for OCT macular disease classification.
🔍 Key Details
- 📊 Dataset: 27,145 retinal images from Singapore, China, and Taiwan
- 🧩 Features used: Fundus photographs and optical coherence tomography (OCT) scans
- ⚙️ Technology: Federated Learning combined with blockchain
- 🏆 Performance: AUC of 0.868 for MMD detection, 0.970 for OCT classification
🔑 Key Takeaways
- 🔒 Privacy preservation is crucial in collaborative AI model development.
- 🤖 Federated Learning allows model training without transferring raw data.
- 🌐 Blockchain integration enhances security during model updates.
- 📈 Robust performance against adversarial attacks was demonstrated.
- ⏱️ Minimal impact on model development time with blockchain (approx. 5 seconds per epoch).
- 🌍 Study conducted across multiple cohorts in Asia.
- 🆔 PMID: 39424148.
📚 Background
The integration of artificial intelligence (AI) in healthcare has the potential to revolutionize diagnostics and treatment. However, the collaboration required for developing robust AI models often raises concerns about data privacy. Traditional methods of data sharing can lead to privacy breaches, making it essential to explore technologies that allow for secure collaboration without compromising patient confidentiality.
🗒️ Study
This retrospective multicohort study utilized a dataset of 27,145 retinal images to develop a novel FL aggregation method aimed at detecting myopic macular degeneration (MMD) and other macular diseases. The study specifically addressed challenges posed by non-independent and identically distributed (non-i.i.d.) data, which is common in healthcare settings. The incorporation of blockchain technology served as a proof of concept for secure model updates across collaborating sites.
📈 Results
The results of the study were promising, with the FL model achieving an AUC of 0.868±0.009 for MMD detection and 0.970±0.012 for OCT macular disease classification. In the face of adversarial attacks, the FL model maintained an AUC of 0.861±0.019 during label flipping attacks, comparable to centralized models. In clean label attacks, it achieved an AUC of 0.878±0.006, demonstrating superior performance compared to other state-of-the-art FL models.
🌍 Impact and Implications
The findings from this study highlight the potential of blockchain-enabled Federated Learning as a trusted platform for collaborative health AI research. By addressing privacy concerns and enhancing security, this approach could pave the way for more effective and ethical AI applications in healthcare, ultimately improving patient outcomes and fostering innovation in medical research.
🔮 Conclusion
This study showcases a significant advancement in the field of AI and healthcare, demonstrating that the integration of Federated Learning and blockchain technology can effectively address privacy challenges while maintaining robust performance against adversarial attacks. As we move forward, further research and development in this area could lead to transformative changes in how we approach collaborative health AI research.
💬 Your comments
What are your thoughts on the integration of blockchain with Federated Learning in healthcare? We would love to hear your insights! 💬 Leave your comments below or connect with us on social media:
Privacy Preserving Technology using Federated Learning and Blockchain in protecting against Adversarial Attacks for Retinal Imaging.
Abstract
PURPOSE: Collaboration provides valuable data for robust artificial intelligence (AI) model development. Federated learning (FL) is a privacy enhancing technology that allows collaboration while respecting privacy via the development of models without raw data transfer. However state-of-the-art FL models still face challenges in non-independent and identically distributed (non-i.i.d.) health care settings and remain susceptible to privacy breaches. We propose a novel FL framework coupled with blockchain technology to address these challenges.
DESIGN: Retrospective multicohort study SUBJECTS AND METHODS: 27,145 images from Singapore, China and Taiwan were used to design a novel FL aggregation method for the detection of myopic macular degeneration (MMD) from fundus photographs and macular disease from optical coherence tomography (OCT) scans in feature distribution skew and label distribution imbalance scenarios. We further performed adversarial attacks (label flipping and clean label). As proof of concept, blockchain was incorporated into FL to demonstrate secure transfer of model updates across collaborating sites.
MAIN OUTCOME MEASURES: We evaluated our FL model performance in MMD and OCT classification and compared our model against state-of the-art FL and centralized models.
RESULTS: Our FL model showed robust performance with areas under the receiving operating characteristic curves (AUC) of 0.868±0.009 for MMD detection and 0.970±0.012 for OCT macular disease classification. In label flipping attack, our FL model had an AUC of 0.861±0.019, similar to the centralized model (AUC 0.856± 0.015) and higher than other FL models (AUC 0.578-0.819) In clean label attack, our FL model had an AUC of 0.878±0.006 which was comparable to the centralized model (AUC 0.878±0.001) and superior to other state-of-the-art FL models with AUC of 0.529-0.838. Simulation showed that the additional time with blockchain in one global epoch was around 5 seconds. The addition of blockchain to the FL framework was feasible with a minimal impact on model development time.
CONCLUSIONS: Our proposed FL algorithm overcomes the shortcoming of the traditional FL in non i.i.d. situations and remains robust to against adversarial attacks. The integration of blockchain adds further security during the transfer of model updates. Blockchain-enabled FL can be a trusted platform for collaborative health AI research.
Author: [‘Teo ZL’, ‘Zhang X’, ‘Yang Y’, ‘Jin L’, ‘Zhang C’, ‘Jieh Poh SS’, ‘Yu W’, ‘Chen Y’, ‘Jonas JB’, ‘Wang YX’, ‘Wu WC’, ‘Lai CC’, ‘Liu Y’, ‘Mong Goh RS’, ‘Wei Ting DS’]
Journal: Ophthalmology
Citation: Teo ZL, et al. Privacy Preserving Technology using Federated Learning and Blockchain in protecting against Adversarial Attacks for Retinal Imaging. Privacy Preserving Technology using Federated Learning and Blockchain in protecting against Adversarial Attacks for Retinal Imaging. 2024; (unknown volume):(unknown pages). doi: 10.1016/j.ophtha.2024.10.017