Investigating MONAI Deploy in a series of incrementally more complex examples, ending in a working pipeline: Orthanc → MONAI Deploy Express → our MAPs → linked DICOM SRs back in Orthanc.
| Folder | What |
|---|---|
Simple_Image_Processing/, MedNIST_Classifier_App/, MedNIST_Classifier_App_Pre-Built_Model/ |
App SDK tutorials (below) |
AIREADI_Laterality_App/ |
laterality model: training code, model weights, tutorial-style MAP |
maps/ |
deployable MAPs and their shared operators |
MONAI_Deploy_Express/ |
upstream Express 0.6.0 running the MAPs; our changes in LOCAL_SETUP.md |
shadow_mode/ |
parked: a custom orchestrator explored before Express |
Examples explored:
- Creating a Simple Image Processing App with MONAI Deploy App SDK
- Deploying a MedNIST Classifier App with MONAI Deploy App SDK
- Deploying a MedNIST Classifier App with MONAI Deploy App SDK (Prebuilt Model)
- AIREADI Laterality Classifier App with MONAI Deploy App SDK:
- Trained a ResNet18 on the AI-READI dataset to classify the laterality of colour fundus photographs (left / right eye)
- Deployed it as a MAP; the production version is
maps/aireadi_laterality