Siemens Healthcare and Intel application of AI technology to show real-time diagnostic cardiac MRI

Intel and Siemens Medical Solutions (Siemens Healthineers) * are collaborating to develop a groundbreaking artificial intelligence-based cardiac MRI (magnetic resonance imaging) segmentation and analysis model, it is expected to provide real-time diagnosis of cardiovascular disease. Intel and Siemens Medical use of the second generation of Intel Xeon processor scalability artificial intelligence reasoning, reasoning provides results in real time magnetic resonance imaging (MRI) for technical specialists, cardiologists and radiologists. David Ryan, general manager of Life Sciences and Health Department Division of Things Intel Corporation said: “Intel and Siemens Medical Solutions have a common goal – the use of artificial intelligence technology, to further improve the level of medical treatment at the edge by deploying integrated Intel acceleration technology and depth of learning. Intel Distribution of OpenVINO Toolkit second generation Intel Xeon processor scalability, the data is immediately after the acquisition for analysis in order to achieve real-time cardiac MRI applications, “why this technology is so important: in the United States, cardiovascular disease It led to a third of the mortality rate – 34 cases per minute, up to 18 million cases per year 1. Cardiac MRI evaluation of cardiac function has become the gold standard ventricular volume and myocardial tissue 1. Cardiologist manual or semiautomatic tools generally used to extract a quantitative measure of the cardiac magnetic resonance imaging (CMR) but this step is very time consuming and error prone and susceptible to subjectivity in the interpretation of the image. Dorin Comaniciu senior vice president of Siemens Medical Solutions, said: “Based on the Intel Xeon processor scalability, we are now able to develop multiple real-time use of medical imaging and critical use cases, such as cardiac MRI, etc., and without additional cost and complexity of hardware accelerators. “using artificial intelligence model of cardiac cardiologist will save more time, so they do not need a manual for image segmentation ventricular myocardium and the blood pool. After the scanner generates image slices Copyright Control Engineering , based on image segmentation performed immediately at the edge of artificial intelligence, to make the computing system deployed at the edge of the real-time data capture generated – this artificial intelligent reasoning brings low latency and high throughput speeds and other advantages , so that medical institutions can be safely diagnostic services to more patients each day.What are the benefits of this technology: life sciences and health industry is undergoing medical digital revolution, through the use of artificial intelligence, speed up clinical workflow, and improve the accuracy of diagnosis, while reducing hospital costs, provide greater support for medical research. Artificial intelligence can quickly provide visualization of anatomical systems, and identify abnormal conditions, which helps clinicians to further focus on patient care. Technical: the moment Control Engineering Copyright , most systems are deployed Siemens Healthcare uses Intel processors, which enable Siemens Medical Solutions to take advantage of the existing infrastructure to run CPU-based artificial intelligence reasoning work load. Siemens Healthcare and Intel uses Intel Distribution of OpenVINO toolkit to optimize, quantify and execution model. The results show that the final presentation, to achieve a speed of more than 5 fold increase , almost no loss in accuracy and 2. Intel depth of learning a new technology is accelerating embedded processor technology , is possible to accelerate the learning example of depth. It Intel? AVX-512 extended instruction set a new neural network instructions vector (VNNI), built in the second generation instruction scalable Intel Xeon processors. In the past, such tasks such as convolution usually requires three instructions, are only one instruction can be completed. This technique can be applied to the target workload comprising image recognition, image segmentation, speech recognition, language translation and object detection.

Related articles