Analysis of CT-images

2022: A CT platform has been released that allows you to look inside the vessels for the treatment of heart disease

At the end of February 2022, Dyad Medical announced the launch of its Libby IAAA Cardiac Imaging Analysis Platform. The platform uses quantitative evaluation of intravascular optical coherence tomography images. More details here.

2018: Philips uses Xeon processors to accelerate CT snapshot analysis multiple times

On August 14, 2018, Philips Healthcare published the results of tests of machine learning algorithms conducted on Intel Xeon Scalable processors with a package of OpenVINO programs for automatic recognition of dynamic images. The researchers evaluated algorithms of two types: one to model changes in bone structure with age from radiological images, and the other to determine the boundaries of lung segments from CT scans.

The use of the Intel Xeon Scalable processor accelerated the analysis of X-rays by 188 times (the original result of 1.42 images per second, obtained - 267.1 images per second). The analysis speed for the lung segmentation model increased 38 times - the algorithm processed 71.7 images per second, while previously the processing speed was 1.9 images per second.

Intel Xeon Scalable Processor Accelerates X-ray Analysis by 188 Times

Until recently, only one tool was used to accelerate deep learning - improving the performance of the graphics processor. These processors work perfectly with images, but when creating some models, researchers have to take into account the built-in memory limitations.

Central processors - in this case, Intel Xeon Scalable processors - do not have such memory limitations and can perform complex, hybrid algorithms, including working with larger, memory-intensive models usually used in the field of medical imaging. When analyzing large subsets using AI, Intel Xeon Scalable processors better meet the needs of researchers than graphics-based systems.

The use of central processors for deep learning analysis is a direct benefit to companies such as Philips because it allows them to offer artificial intelligence-based services without increasing end-user costs.^[1]

Notes