CCHMC Ped Abd CT Seg Example App #525
Conversation
Signed-off-by: bluna301 <[email protected]>
Signed-off-by: bluna301 <[email protected]>
![github-advanced-security[bot]](https://avatars.githubusercontent.com/in/57789?s=60&v=4){kind=small}
bluna301
changed the title
Signed-off-by: bluna301 <[email protected]>
bluna301
changed the title
|
HI @bluna301, Thanks for the pull request. I will review it by the end of the week. |
Thanks Vikash! |
Signed-off-by: bluna301 <[email protected]>
examples/apps/hugging_face_integration_app/med_image_generation/app.py
Outdated
Show resolved
Hide resolved
| "Modality": "(?i)CT", | ||
| "ImageType": ["PRIMARY", "ORIGINAL", "AXIAL"], | ||
| "SeriesDescription": "(?i)^(?!.*(cor|sag)).*$", | ||
| "SliceThickness": [3, 5] |
There was a problem hiding this comment.
This range support looks good once we publish the next version of MD SDK package (as your PR had already been merged). In the meantime, the editable SDK with the recent code will be needed. Planning to do the next release of MD App SDK later in the month and bumping the version to 3.0, following the release of holoscan v3.0, but we can discuss if a minor 2.x version needs to be squeezed in.
There was a problem hiding this comment.
Great point Ming, I was weighing the same considerations when I was drafting the PR. If it's fine with you, I'm fine with waiting to merge this PR until App SDK V3.0.0 is released, so that we don't have to include most recent Series Selector Operator updates as a Python file within this example app. Let me know what you think.
examples/apps/cchmc_ped_abd_ct_seg_app/mongodb_entry_creator_operator.py
Show resolved
Hide resolved
|
Another thing worth mentioning is that there are more advanced operator execution conditions and soon dynamic workflow support in the SDK, which can be used for conditional inclusion and execution of some branches of the workflow, e.g. the Mongo DB integration. Examples of their use are not in the repo yet, and we can think about adding such in the next version of this app. |
Signed-off-by: bluna301 <[email protected]>
|
@MMelQin thanks for your review. Your latest PR with the typo, formatting, and other fixes has been pulled in. I have addressed the spelling and dependency concerns you highlighted in your code review in my latest commit. I am fine with waiting to merge this PR until after v3.0.0 of App SDK is released so that users will have access to the updated DICOM Series Selector range matching. Also happy to explore implementing the dynamic workflow support for the MongoDB write in a future version of this example app after holoscan v3.0.0 is available. |
Signed-off-by: bluna301 <[email protected]>
|
This MAP pipeline is based on the CCHMC Pediatric Abdominal CT Segmentation MONAI Bundle.
Some unique features of this MAP pipeline include:
AbdomenSegOperatorenables either PyTorch or TorchScript model loadingDICOMSecondaryCaptureWriterOperatorwrites a DICOM SC with organ contoursapp.pyMongoDBEntryCreatorOperatorandMongoDBWriterOperator) allow for writing to the MongoDB database associated with MONAI Deploy ExpressAs described in the
README.md, when executing the pipeline pythonically, a.envfile that contains the relevant MongoDB connection information is recommended. However, I am unsure of the best way to implement this sensitive information into the MAP without hardcoding.