AI-assisted segmentation extension

Announcements
,
42

Is there a python script to run Boundary-point based segmentation?

43

See example of running NVidia AI segmentation from Python script here: https://www.slicer.org/wiki/Documentation/Nightly/ScriptRepository#How_to_run_segment_editor_effects_from_a_script

44

Thank you for your response. I’ve run through these examples but was hoping to find a script specific to Nvidia’s AI-assisted Boundary-points based segmentation mode. Something similar to the example you provided for doing brain tumor segmentation using Nvidia here. If you are aware of any other communities/sources I could use to search for something like this that would be a huge help.

45

If you have a look at the source code of the AIAA module then is should not be too hard to figure out how to change the current example.

46

Where could I find the source code you are referring to? Do you mean the brain tumor segmentation code I linked to above or do you mean code that exists within the extension folder for Nvidia?
I am also having trouble running the script you’ve given for brain tumor segmentation. I am running the script from the python console within Slicer and am getting the following error: Screen Shot 2020-09-03 at 2.53.14 PM

The rest of the error message contains the following:

Traceback (most recent call last):
File "/Applications/Slicer.app/Contents/Extensions-29345/NvidiaAIAssistedAnnotation/lib/Slicer-4.11/qt-scripted-modules/SegmentEditorNvidiaAIAALib/SegmentEditorEffect.py", line 472, in onClickAnnotation
result_file = self.logic.dextr3d(in_file, session_id, model, pointSet)
File "/Applications/Slicer.app/Contents/Extensions-29345/NvidiaAIAssistedAnnotation/lib/Slicer-4.11/qt-scripted-modules/SegmentEditorNvidiaAIAALib/SegmentEditorEffect.py", line 1053, in dextr3d
session_id=session_id)
File "/Applications/Slicer.app/Contents/Extensions-29345/NvidiaAIAssistedAnnotation/lib/Slicer-4.11/qt-scripted-modules/NvidiaAIAAClientAPI/client_api.py", line 265, in dextr3d
points, crop = AIAAUtils.image_pre_process(image_in, cropped_file, point_set, pad, roi_size)
File "/Applications/Slicer.app/Contents/Extensions-29345/NvidiaAIAssistedAnnotation/lib/Slicer-4.11/qt-scripted-modules/NvidiaAIAAClientAPI/client_api.py", line 558, in image_pre_process
points[::, 0] = points[::, 0] - x1
IndexError: too many indices for array: array is 1-dimensional, but 2 were indexed

I am running the most updated version of Slicer 4.11 on a MacOS Mojave

47

I mean that you can read the AIAA effect’s source code and update the AIAA example script accordingly.

If you find any errors in NVidia AIAA segmentation then please report to https://github.com/NVIDIA/ai-assisted-annotation-client/issues.

1 Like
48

Hello,

I am interested in this 3D Slicer extension for academic purposes (non-commercial), and I was wondering how you would like it to be cited/referenced?

Regards

49

You can cite 3D Slicer as described here and in addition to that you can refer to NVidia Clara by this link.

1 Like
50

Hi all,
I was wondering if there is a list of ready to use servers/services for different applications.
Something like:

  • server 123.456.678 -> CT segmentation
  • server test.test -> prostate MR segmentation
    …
    …

Thanks a lot.
Paolo

51

There is only one publicly available server (the default server - when you don’t specify any server address). It would be great if others would set up various servers, too, but there is not much incentive for anyone to do so. If grant funding agencies or journals started to require networks to be publicly available then this may change, but I don’t expect this to happen in the near future.

52

Perfect, I got it.

Thanks

53

Thank you Andras. Do you happen to know how I might be able to run the AIAA example script from terminal? I’m not sure exactly where I should have the file saved/which folder to run the code from using terminal and also how to include the correct folder path. If you knew roughly what command to use from terminal to run this that would be a huge help.

54

You can use --python-script argument to specify a script and command-line arguments to a script that you want to run in Slicer’s Python environment. See a complete example here: https://github.com/SlicerRt/SlicerRT/tree/master/BatchProcessing

55

Prof Lasso -

Many thanks for your work and sharing it with the community. We’re actually pondering a server setup to use within our small hospital group, but are a bit overwhelmed on where to start. Would you recommend going through the CLARA documentation to start and go from there? Is there any resource you would suggest as well?

Thanks,
Alb

56

2 posts were split to a new topic: NVidia AIAA segment editor effect returns “URL not found” error

57

Hi there, very great tool! Are there specific requirements to the CT-dataset (specific aquisition parameters like slice thickness …) Thank yu very much!

58

Most models that NVidia created are trained on medical segmentation decathlon data sets. You’ll get the best results if you use similar resolution and cropping. Fully automatic segmentation models are more sensitive and slower, so if you have trouble using those then you may switch to the ones that also take boundary points as inputs.

59

Hi, trying to connect, but the server is not responded. any solution regarding this?

thank you.

60

The segmentation server address had changed, upgrade to the latest Slicer Stable Release or latest Slicer Preview Release and also update (or uninstall and install) the NVidia AI assisted segmentation extension.

61

A post was merged into an existing topic: Nvidia AI-assisted segmentation (brain tumor)