End-to-end privacy preserving deep learning on multi-institutional medical imaging

By Georgios Kaissis, A. Ziller, Jonathan Passerat-Palmbach, T. Ryffel, Dmitrii Usynin, Andrew Trask, Ionésio Lima, Jason V. Mancuso, F. Jungmann, Marc-Matthias Steinborn and 4 more. Nature Machine Intelligence, 2021

478 citations recorded by the source corpus.

Using large, multi-national datasets for high-performance medical imaging AI systems requires innovation in privacy-preserving machine learning so models can train on sensitive data without requiring data transfer. Here we present PriMIA (Privacy-preserving Medical Image Analysis), a free, open-source software framework for differentially private, securely aggregated federated learning and encrypted inference on medical imaging data. We test PriMIA using a real-life case study in which an expert-level deep convolutional neural network classifies paediatric chest X-rays; the resulting model’s classification performance is on par with locally, non-securely trained models. We theoretically and empirically evaluate our framework’s performance and privacy guarantees, and demonstrate that the protections provided prevent the reconstruction of usable data by a gradient-based model inversion attack. Finally, we successfully employ the trained model in an end-to-end encrypted remote inference scenario using secure multi-party computation to prevent the disclosure of the data and the model. Gaining access to medical data to train AI applications can present problems due to patient privacy or proprietary interests. A way forward can be privacy-preserving federated learning schemes. Kaissis, Ziller and colleagues demonstrate here their open source framework for privacy-preserving medical image analysis in a remote inference scenario.

Subfields: Artificial Intelligence, Health Informatics, Radiology, Nuclear Medicine and Imaging

View Paper: https://www.semanticscholar.org/paper/9bbb61d9bcfa1a2b63efbdb8a521af8c8a2e94ec

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Dacheng Tao
2. Bo Du
3. Mang Ye
4. Zsolt Kira
5. Yen-Chang Hsu
6. Qinbin Li
7. Bingsheng He
8. James Smith

In Health Informatics

1. Bingsheng He
2. Qinbin Li

In Radiology, Nuclear Medicine and Imaging

1. Zsolt Kira
2. Dacheng Tao
3. Bingsheng He
4. Bo Du
5. Dimitris N. Metaxas
6. Shaoting Zhang
7. Pong Chi Yuen
8. Mang Ye

Citations in the corpus

• A Closer Look at Rehearsal-Free Continual Learning * (2022, background): “, 2021) and inverting images from a trained model can also violate the same data-privacy concerns (Kaissis et al., 2021).”
• On the Challenges and Perspectives of Foundation Models for Medical Image Analysis (2023, methodology): “Furthermore, sharing multi-cohort knowledge in foundation models is also feasible through the use of federated learning Li et al. [2020b], Kaissis et al. [2021], which enables training on data distributed across multiple institutions or devices without the data ever leaving the local machines.”
• Federated learning for medical image analysis: A survey (2023, background): “The Privacy-preserving Medical Image Analysis (PriMIA) (Kaissis et al., 2021) is an open-source framework for privacy-preserving decentralized deep learning with medical images.”
• A Survey on Federated Learning Systems: Vision, Hype and Reality for Data Privacy and Protection (2019, methodology): “other, and an FLS can adopt multiple methods to enhance the privacy guarantees [60], [78], [189].”
• Heterogeneous Federated Learning: State-of-the-art and Research Challenges (2023, background): “Federated learning has drawn increasing attention in both academia and industry owing to its potential utility in large-scale applications [48, 85, 185], which has been widely explored in the fields of healthcare [39, 178, 214], medical analysis [54, 97], and data security [205], etc.”