News
NVIDIA Introduces VISTA-3D NIM Microservice for Advanced CT Scan Analysis
Terrill DickiJul 28, 2024 07:58
NVIDIA introduces VISTA-3D NIM microservice, enhancing CT scan analysis with advanced organ and disease segmentation capabilities.
More than 300 million computed tomography (CT) scans are performed each year worldwide, including 85 million in the United States alone. Radiologists are continually looking for ways to speed up their workflow and produce accurate reports. To meet this need, NVIDIA Research has developed a new base model, VISTA-3D, that is integrated into a streamlined microservice called NVIDIA NIM, designed for scalable deployment, according to NVIDIA Tech Blog.
VIEW-3D Model
The VISTA-3D (Versatile Imaging Segmentation and Annotation) model is trained on over 12,000 volumes, covering 127 types of human anatomical structures and various lesions, including lung nodules, liver tumors, and bone lesions. It offers precise, out-of-the-box segmentation and state-of-the-art zero-shot interactive segmentation, making it a versatile tool for medical imaging.
The model features three main workflows:
- Segment everything: It allows a complete exploration of the body, helping to understand complex pathologies that affect multiple organs.
- Segment using class: It provides detailed class-based visualizations, essential for targeted disease analysis.
- Segment Point Requests: Improve segmentation accuracy through user-driven selection, accelerating the creation of accurate ground-truth data.
The VISTA-3D architecture includes an encoder layer followed by two parallel decoder layers, one for automatic segmentation and another for point prompts. This structure ensures high accuracy and adaptability in different anatomical areas.
Microservice NIM VISTA-3D
Hosted on the NVIDIA API catalog, the VISTA-3D NIM microservice allows users to test its capabilities with sample data. It can segment over 100 specific organs or classes of interest, providing views in axial, coronal, or sagittal planes.
Using NIM Microservices
Users can run VISTA-3D on their own data by signing up for a personal key from NVIDIA, which provides 1,000 free credits to try out any NIM microservice. Detailed instructions on generating an API key and running the model are available, along with sample code in various programming languages.
For those who want to run VISTA-3D on their own data, it is necessary to set up an FTP server to serve the medical images. This approach is suitable for the large size of medical images, which are typically too large to be sent directly in API payloads.
Running NIM Microservices On-Premises
To run NIM microservices locally, users must request NVIDIA NIM access. Once approved, they will receive a Docker container to run the VISTA-3D NIM microservice on their preferred hardware. Prerequisites include installing Docker, Docker Compose, and NVIDIA drivers.
To help users get started quickly, a sample Docker Compose file is provided, along with instructions for setting up an NGINX server for image deployment.
Conclusion
NVIDIA’s VISTA-3D core model represents a significant advance in medical imaging, providing accurate segmentation of over 100 organs and various pathologies in CT scans. The NVIDIA NIM microservice simplifies the deployment and use of this powerful model, improving radiologists’ workflow and accuracy.
Interested parties can request access to the VISTA-3D NIM microservice to leverage its capabilities on their own hardware, streamlining medical imaging processes.
Image source: Shutterstock