Demetics Protects AI-Based Medical Innovation with Intel SGX

Demetics Medical Technology Co. Ltd. is using Intel® Software Guard Extension (Intel® SGX) and Intel® oneAPI Math Kernel Library (oneMKL) to protect its medical artificial intelligence (AI) algorithms and intellectual property (IP) in medical devices at the edge.

News

author-image

By

What’s New: Demetics Medical Technology Co. Ltd. is using Intel® Software Guard Extension (Intel® SGX) and Intel® oneAPI Math Kernel Library (oneMKL) to protect its medical artificial intelligence (AI) algorithms and intellectual property (IP) in medical devices at the edge. A pioneer in China of AI-based ultrasonography, Demetics accelerated adoption of DE-Light, its independently developed deep-learning framework that has shown outstanding performance and improved the accuracy of thyroid nodule detection under an open source framework by 30% to 40%.1

“Due to the nature of the medical industry, some artificial intelligence vendors have to deploy their solutions on the client side. Therefore, protecting the core intellectual property of manufacturers’ AI models has become an urgent need. The trusted execution environment technology based on Intel SGX technology has become the most cost-effective and ultimate solution for runtime protection. With the support of Intel, they [Demetics] strategically apply SGX technology to a series of AI products, including its ultrasonic AI products, greatly enhancing its market competitiveness and innovation capabilities.”
–Yali Liang, Intel vice president in the Sales, Marketing and Communications Group and general manager of PRC Business Consumption

Why It Matters: Ultrasonic imaging diagnosis and analysis are widely used and an important part of current healthcare diagnosis and treatment; however, imaging examination depends heavily on manual analysis and can be inefficient.

One of AI’s greatest values in medicine is its quick analysis of a large amount of data and its ability to obtain accurate results. By using intelligent imaging diagnostics, AI technology in diagnosis and analysis can help medical institutions accelerate detection and treatment. Doctors can improve the efficiency of film reading, reduce the probability of misdiagnosis and obtain diagnostic assistance, while patients can receive more accurate diagnoses and personalized treatment recommendations.

Demetics has successfully deployed AI medical application products in more than 400 hospitals across China, and it plans to deploy in more than 1,000 hospitals in the coming year.

How It Works: Demetics launched the AI-SONIC™ computer-aided diagnosis system for ultrasonic imaging and also developed several other AI products that use Intel SGX to protect algorithms. In the ultrasonic AI model solution, to ensure that the algorithm in Intel SGX can make full use of the computing power of Intel® processors, the project team, with the help of the Intel® C++ Compiler, manually optimized the core calculation of the algorithm floating-point matrix multiplication modeled on oneMKL.

AI-powered ultrasonic model protection solution based on Intel architecture.

» Click for full image

oneMKL is a highly optimized and parallel library of mathematical functions for Intel and its compatible processors. In different operating environments, oneMKL can automatically run runtime processor detection and specifically optimized programs on different processors, thus enabling better performance across all its processors. Intel® oneAPI Math Kernel Library (oneMKL) and Intel® oneAPI Deep Neural Network Library (oneDNN) provide underlying support for commonly used deep learning frameworks and numerous self-developed AI applications.

Intel SGX is a set of instructions that increase the security of application code and data, offering them more protection from disclosure or modification. Developers can partition sensitive information into a hardware-based trusted execution environment (TEE) or enclave — an area of memory with a higher level of security protection. The technology helps ensure the root of trust is limited to a small portion of the central processing unit’s hardware, better protecting the confidentiality and integrity of code and data.

The Small Print:
1 The test results are quoted from internal evaluation of Demetics. For more details, please contact Demetics.