Hi, I would like to use my own labeled dataset to kickstart monai label training.
Background:
Followed the youtube installation and scribbler tutorial, i was able to download and view “Task09_Spleen” in 3D slicer using:
monailabel start_server --app apps/radiology --studies datasets/Task09_Spleen/imagesTr --conf models deepedit
Problem:
I have a LABELED dataset “Task013_Abdomen” set up exactly like example “Task09_Spleen”. I want to train a new model to detect liver tumours, liver, and kidney. I want to start the training procedure with existing Task013_abdomen images and labels, and then use the model to iteratively correct and add more segmentations to my model on Slicer 3D.
Specific issues:
-
Slicer 3D does not see my dataset. error message “no unlabeled images found”. How do I prime it to expect an already labeled dataset for training?
-
What configuration to start monai label server in? I am currently using the following terminal prompt:
monailabel start_server --app apps/radiology --studies
datasets/Task013_Abdomen/imagesTr --conf models segmentation -
How do I generate the dataset.json and datastore_v2.json that appears in example “Task09_Spleen”. I think this is required for slicer to see my MRI images and their respective labels right?
Any help is appreciated, this is taking me a while to set up correctly.
Cheers,
Ulvi