Secure Medical Image Analysis with CrypTFlow

TL;DR

CRYPTFLOW is a system that simplifies converting TensorFlow models into secure multi-party computation protocols, enabling privacy-preserving medical image analysis with significant speedups and supporting complex tasks like 3D image segmentation.

Contribution

We introduce an end-to-end compiler from TensorFlow to MPC protocols and an improved semi-honest 3-party protocol, enabling efficient secure inference for complex medical imaging tasks.

Findings

Secure inference of neural networks like DENSENET121 and 3D-UNet demonstrated.

First evaluation of secure 3D image segmentation.

Largest secure inference task to date.

Abstract

We present CRYPTFLOW, a system that converts TensorFlow inference code into Secure Multi-party Computation (MPC) protocols at the push of a button. To do this, we build two components. Our first component is an end-to-end compiler from TensorFlow to a variety of MPC protocols. The second component is an improved semi-honest 3-party protocol that provides significant speedups for inference. We empirically demonstrate the power of our system by showing the secure inference of real-world neural networks such as DENSENET121 for detection of lung diseases from chest X-ray images and 3D-UNet for segmentation in radiotherapy planning using CT images. In particular, this paper provides the first evaluation of secure segmentation of 3D images, a task that requires much more powerful models than classification and is the largest secure inference task run till date.