Integrate Sonador with MONAI label: add MONAILabel OHIF plugin to Sonador build, update permissions to allow for multi-user support
MONAILabel is a server-client system that facilitates interactive medical image annotation by using AI. It is an open source and easy to install set of components that facilitates the creation of Deep Learning models for targeted clinical segmentation tasks (examples: spleen segmentation, left atrium segmentation, MRI knee segmentation, MRI spine segmentation).
Demonstration: https://www.youtube.com/watch?v=m2rYorVwXk4
Models are hosted (and incrementally trained) on a server process. A client helps to visualize the model and allows for human in the loop feedback to improve the model performance.
Development tasks:
-
MONAILabel POC - Checkout MONAILabel sources, install plugin and sample application
- MONAI Server: https://github.com/Project-MONAI/MONAILabel
- Slicer Plugin: https://github.com/Project-MONAI/MONAILabel/tree/main/plugins/slicer
- Sample Application (Left Atrium Segmentation): https://github.com/Project-MONAI/MONAILabel/tree/main/sample-apps/segmentation_left_atrium
- Successfully train a left atrium segmentation model (or alternative)
- Re-train a segmentation model to another challenge (such as MRI based knee segmentation or heart segmentation)
- Checkout MONAILabel sources, install plugin and sample application
-
Determine modifications needed to Sonador OHIF to enable the MONAILasbel OHIF plugin to be embedded in the Sonador OHIF build (or a custom PWA) - What kind of security is required to protect the MONAILabel server?
- How does the MONAILabel server need to persist its models so that it can work in a cloud environment?
-
Create a cloud implementation of the server/client components (Slicer/OHIF/MONAILabel server) that demonstrates MRI knee segmentation