Open Source Privacy Preserving Inference in Medical Imaging

Explore privacy-preserving machine learning techniques for medical imaging in this 56-minute conference talk from the Toronto Machine Learning Series. Dive into the critical challenges of secure prediction-as-a-service and secure model testing in healthcare scenarios. Learn about CrypTFlow, an innovative system that compiles TensorFlow/ONNX inference code into secure computation protocols. Discover how this technology enables patients to receive prognoses without revealing sensitive medical data while protecting hospital models. Examine real-world case studies demonstrating CrypTFlow's potential in detecting lung diseases, predicting doctor visit frequency for Wet-AMD patients, and segmenting 3D CT scans for radiotherapy planning. Gain insights from Microsoft Research experts Dr. Divya Gupta and Rahul Raina on making cryptography practical, usable, and performant in healthcare applications.

Open Source Privacy Preserving Inference in Medical Imaging