- About XNAT
- News & Events
- XNAT Marketplace
- Contact Us
The latest version of XNAT is now available, and includes several powerful enhancements as well as UI improvements, upgrades to foundational technology, and improved support for containerized processing and machine learning workflows. There are also related updates to core plugins such as the Container Service and XNAT-OHIF Viewer.
What Can You Do With XNAT 1.8?
The big new functional addition is the XNAT Event Service, which provides a way to set up and fine-tune automated tasks in response to data events in XNAT. Event subscriptions can be tied to container executions and other actions.
XNAT 1.8 also incorporates and hardens support for the Machine Learning capabilities introduced in the XNAT ML demo. Additionally, there are improvements to the image uploading process, including asynchronous processing of archive actions and enhanced administrative controls for custom DICOM project routing. This release also includes improvements to custom resource uploading, improved handling of project-level experiments, and native generation of image snapshots rather than using the Autorun pipeline.
Upgrading to XNAT 1.8
XNAT 1.8 is built on Java 8, PostgreSQL 10, and removes the dependency on the XNAT Pipeline Engine, providing more flexibility in data processing. The upgrade process for moving from previous versions of XNAT to XNAT 1.8 is otherwise unchanged.
Released alongside XNAT 1.8, new versions of core XNAT plugins are also available.
The latest version of Container Service serves as an "Action Provider" to the Event Service in XNAT 1.8, allowing for native event subscriptions that trigger container processing launches. Also, Container Service 3.0 offers improved support for Docker Swarm, which adds container request queuing and processing resource management.
As a companion to the Container Service, the Batch Launch Plugin offers the ability to launch containers or pipelines in bulk, and manage those operations in a centralized Processing Dashboard.
The latest version of the XNAT OHIF Viewer focuses on defining regions-of-interest, with the "ROI Collection" datatype now a part of core XNAT, and on integration with machine learning capabilities.
The set of core XNAT ML Beta plugins necessary to launch model training has been updated with hardened UI feature support and refactored datatype definitions. This includes the ability to train and fine-tune deep learning models using NVIDIA Clara Train, as well as import/export models, configure model training parameters, and review training results. Pre-trained seed models for image segmentation, annotation and classification are available for download from NVIDIA NGC.
The XNAT Project Sync plugin has also been updated for compatibility with XNAT 1.8, as well as adding better support for handling synchronization of large XAR files and custom relabeling schemes.