Failed to run inference in MONAI Label Server

[aleoneshahvar]

I am able to grab the next sample from my studies folder however this error pops up as soon as the image is grabbed:
Traceback (most recent call last):
File "C:/Users/Ariela/AppData/Local/NA-MIC/Slicer 4.13.0-2021-10-03/NA-MIC/Extensions-30282/MONAILabel/lib/Slicer-4.13/qt-scripted-modules/MONAILabel.py", line 1549, in onClickSegmentation

result_file, params = self.logic.infer(model, image_file, self.getParamsFromConfig("infer", model))

File "C:/Users/Ariela/AppData/Local/NA-MIC/Slicer 4.13.0-2021-10-03/NA-MIC/Extensions-30282/MONAILabel/lib/Slicer-4.13/qt-scripted-modules/MONAILabel.py", line 1832, in infer

result_file, params = client.infer(model, image_in, params, label_in)

File "C:\Users\Ariela\AppData\Local\NA-MIC\Slicer 4.13.0-2021-10-03\NA-MIC\Extensions-30282\MONAILabel\lib\Slicer-4.13\qt-scripted-modules\MONAILabelLib\client.py", line 111, in infer

"Status: {}; Response: {}".format(status, form),

MONAILabelLib.client.MONAILabelException: (2, "Status: 500; Response: b'Internal Server Error'")

Once I dismiss the error, I am still able to segment the image and submit it to the model. I can see that the model does in fact update and that the training bar progresses. I am unsure what this error means though and am wondering if it will cause errors down the line. Please let me know what you think!

[SachidanandAlle]

Please check the logs... either logs directory in your apps folder.. app.log in your app/logs folder
Or you can access online.. http://server:port/logs?lines=-1

After fetching the sample, slicer tries to run auto-segmentation (depending on the model/availability) and looks like it didn't get any label from server.... so you can check what happened at the server in its' logs

[aleoneshahvar]

Hello! Could you clarify what you mean by this? I've trained the model but when I go to test the automated segmentation, the same error appears. The link you provided also does not lead to anything. Thanks!

[aleoneshahvar]

What exactly am I looking for in the logs?

[SachidanandAlle]

Can you attach the log file?

[aleoneshahvar]

Here are screenshots from the last entry in my logs!

[SachidanandAlle]

Do you have GPU with CUDA enabled?

[aleoneshahvar]

When I load the images into ImageJ it says "wrong size, 512z104 expected, 512x512 found", I'm not sure if that helps at all!

[diazandr3s]

You can use Slicer or ITK-SNAP to check the image details (image size, modalities, spacing, etc) and also to visualize the images along with the labels. Currently, MONAI Label doesn't support multimodality datasets :/
Is it the dataset public? Would it be possible to share an image with their corresponding label? I'm happy to check them :)

[aleoneshahvar]

I'm having a hard time getting the files to attach here on github, do you have an email I can send them to by chance?

[diazandr3s]

[email protected]

[lassoan]

I'm having a hard time getting the files to attach here on github, do you have an email I can send them to by chance?

In general: upload anywhere (dropbox, onedrive, google drive, etc.) and post the link.

[SachidanandAlle]

@diazandr3s can you help..

[aleoneshahvar]

Just wanted to close this discussion! @diazandr3s was able to help me resolve the issues! The primary issue had to do with my manual segmentations. I used the paint tool on one of the older versions of MONAI Label to segment my dataset (pregnant guinea pigs), however, I segmented each 2D slice independently (this is how manual segmentations are done in my lab) and then submitted it to the model. Andres showed me how to conduct the segmentations using the 'grow from seeds' tool in 3D slicer and suggested that I update to the latest version of MONAI and slicer. Additionally, Andres showed me that the "Failed to run inference on MONAI Label server" error was popping up because I had "Auto-run pre-trained model" checked off under slicer's application settings (Edit -> Application Settings -> MONAI Label) while I am not using a pre-trained model. Thank you to everyone who responded and helped me with these issues!