Skip to content

Fixes #18 Update the AI Model Service method to include the modality information sent from the viewer and perform model selection

  • Created a method 'model-selection' that calls modelSelection.py
  • Receives data on image, dicom tags & model configuration information from the viewer end
  • Utilized the DICOM tags information to perform model selection (Chest, Mammo, etc)
  • Once a model is selected its respective task and endpoint information captured from the viewer is used to call the corresponding endpoint. Right now since only cheXnet is available, it calls the 'boundin-box' method

Tested with postman, image

Tested complete model selection integrity by deploying both viewer and AI model service. Uploaded a DICOM to check for inference. image

Merge request reports