Siemens Healthineers Uses Medical Imaging AI with MONAIs Deploy.
MONAI Deploy
The Siemens Healthineers Digital Marketplace now offers MONAI integration, making it simple to incorporate AI into clinical procedures.
3.6 billion. In order to diagnose, track, and treat a variety of illnesses, that is about how many medical imaging tests are carried out globally each year.
Improving health outcomes and assisting physicians in managing their workloads require expediting the processing and analysis of all these X-rays, CT scans, MRIs, and ultrasounds.
For this reason, NVIDIA unveiled MONAI, an open-source platform for research and development of AI applications for use in medical imaging and other fields. In order to develop deep learning models and deployable apps for medical AI processes, MONAI brings together data scientists and physicians to harness the potential of medical data.
Siemens Healthineers has adopted MONAI Deploy, a module within MONAI that bridges the gap from research to clinical production, to speed up and improve the efficiency of integrating AI workflows for medical imaging into clinical deployments, NVIDIA announced this week at the annual meeting of the Radiological Society of North America, or RSNA.
The Siemens Healthineers Syngo Carbon and syngo.via enterprise imaging systems, which are installed in more than 15,000 medical devices worldwide, assist physicians in better interpreting and deriving meaning from medical pictures from a variety of sources.
A range of frameworks are commonly used by developers for creating AI applications. Because of this, implementing their apps in clinical settings is difficult.
MONAI Deploy creates AI apps that operate anywhere with only a few lines of code. It is a tool for creating, packaging, testing, implementing, and executing clinically produced medical AI applications. It simplifies the process of creating and incorporating AI solutions for medical imaging into clinical processes.
The Siemens Healthineers platform’s MONAI Deploy has greatly sped up the AI integration process. Instead of taking months, customers can now import trained AI models into actual clinical situations with a few clicks. This facilitates the faster delivery of apps to radiologists by researchers, entrepreneurs, and startups.
“Thousands of clinical researchers worldwide can access AI-driven advancements directly on their syngo.via and Syngo Carbon imaging platforms with MONAI Deploy, which enables researchers to swiftly customize AI models and translate innovations from the lab to clinical practice.”
These platforms can greatly expedite AI integration when enhanced with apps created by MONAI. The Siemens Healthineers Digital Marketplace makes it simple to provide and utilize these applications, allowing users to peruse, choose, and effortlessly incorporate them into their clinical processes.
MONAIs Ecosystem Boosts Innovation and Adoption
As it approaches its fifth anniversary, MONAI has amassed over 3.5 million downloads, 220 global contributors, recognition in more than 3,000 articles, 17 MICCAI challenge victories, and usage in a wide range of clinical products.
Researchers and clinicians now have even more options to benefit from MONAIs innovations and contribute to Siemens Healthineers Syngo Carbon, syngo.via, and the Siemens Healthineers Digital Marketplace with to the newest release of MONAI v1.4.
New foundation models for medical imaging have been added to MONAI v1.4 and associated NVIDIA products. These models may be tweaked in MONAI and implemented as NVIDIA NIM microservices. As NIM microservices, the following models are now widely accessible:
MAISI (Medical AI for Synthetic Imaging): A latent diffusion generative AI foundation model called MAISI (Medical AI for Synthetic Imaging) can replicate full-format, high-resolution 3D CT scans and associated anatomic segmentations.
VISTA-3D: A CT image segmentation foundation model, VISTA-3D provides accurate out-of-the-box performance across more than 120 main organ classes. Additionally, it has zero-shot capabilities and efficient adaptability for learning to segment new structures.
The new MONAI Multi-Modal Model, or M3, is now available via MONAIs VLM GitHub repository in addition to the main features of MONAI 1.4. M3 is a framework that adds medical AI specialists, such trained AI models from MONAIs Model Zoo, to any multimodal LLM. The VILA-M3 foundation model, which is already accessible on Hugging Face and provides cutting-edge radiological image copilot performance, exemplifies the strength of this new architecture.
MONAI Bridges Hospitals, Healthcare Startups and Research Institutions
MONAI is being adopted and advanced by top healthcare organizations, university medical institutes, startups, and software companies worldwide, including:
- German Cancer Research Center: The benchmark and metrics working group of MONAI, headed by the German Cancer Research Center, develops measures to assess AI performance as well as recommendations on when and how to apply them.
- Nadeem Lab at Memorial Sloan Kettering Cancer Center (MSK): It was the first to use MONAI to deploy several AI-assisted annotation pipelines and inference modules for pathology data on the cloud.
- University of Colorado School of Medicine: MONAI-based ophthalmology technologies were created by academics at the University of Colorado School of Medicine to identify retinal disorders utilizing a range of imaging modalities. Additionally, the institution spearheads some of the initial advancements in federated learning and clinical MONAI demonstrations.
- MathWorks: With the integration of MONAI Label into its Medical Imaging Toolbox, MathWorks has made medical imaging AI and AI-assisted annotation available to thousands of MATLAB users working in academia and industry on medical and biomedical applications.
- GSK: For image segmentation, GSK is investigating MONAI foundation models including VISTA-3D and VISTA-2D.
- Flywheel: It provides a platform that grows to meet the demands of research institutions and life sciences enterprises. This platform includes MONAI, which simplifies imaging data administration, automates research workflows, and enables AI creation and analysis.
- Alara Imaging: At the 2024 Society for Imaging Informatics in Medicine conference, Alara Imaging presented their work on merging LLMs like Llama 3 with MONAI foundation models like VISTA-3D.
- RadImageNet: It is investigating the use of MONAIs M3 framework to create state-of-the-art vision language models that provide superior radiological reports by using MONAIs expert image AI models.
- Kitware: It offers expert software development services related to MONAI, assisting in the integration of MONAI into regulatory-approved devices and unique processes for device makers.
To develop and implement scalable AI systems, researchers and businesses are also utilizing MONAI on cloud service providers. AWS Health Imaging, Google Cloud, Precision Imaging Network, a component of Microsoft Cloud for Healthcare, and Oracle Cloud Infrastructure are among the cloud platforms that offer MONAI access.
