At SIIM 2020, the annual meeting of the Society for Imaging Informatics in Medicine, NVIDIA announced updates to the Clara Deploy Application Framework to expand its capabilities to address COVID-19, digital pathology and improved developer experience.
The Clara Deploy Application Framework allows you to ingest massive amounts of data from popular data sources like PACS (using a DICOM adapter), encapsulate your medical imaging business logic on operators (i.e. image segmentation) as part of a pipeline, an execute those jobs at scale using cloud-native paradigms based on containers and Kubernetes orchestration.
Below is an overview of the updates to the latest public release 6. For more details please visit the updated Deploying Healthcare AI Workflows with the NVIDIA Clara Deploy Application Framework blog post.
- Reference pipeline: Detection of COVID-19 in CT datasets. The pipeline reads from a single aixial DICOM series of a lung CT scan, segments the Lung with a new AI operator, classifies the segment regions for presence or absence of COVID-19, and optionally sends the results to a DICOM device.
- Digital Pathology (DP) new reference pipeline and operators, showcasing tiling, processing and stitching. Several improvements have been included in the Render Server to visualize multi-resolution data in TIFF formats and distance measurement in 2D, among others.
- Developer experience enhancements:
- DICOM adapter can now be accessed via REST APIs, facilitating the pre-processing and selection of the best DICOM series for processing with a new reference pipeline and parser operator.
- Multi AI CT pipelines use the Shared Memory FastIO interface, both in read and write mode.
