Christian Herz

``` { "name":"MONAILabel - Radiology", "description":"DeepLearning models for radiology", "version":"2.0", "labels":[ "model" ], "models":{ "segmentation_organ":{ "type":"segmentation", "labels":[ "organ" ], "dimension":3, "description":"A pre-trained model for volumetric (3D) segmentation of the organ...

It would still be interesting to know what the reason for this issue is. Is there anything cashed by MONIA Label, that we could delete instead of reinstalling everything?

It seems that I found the reason for this happening in our case: We implemented https://github.com/JolleyLab/DeepHeart which is a SegmentEditorEffect using UI components of MONAILabel (same names) in a new...

@ntdb works for me fine. Great improvement!

Hi @seziegler, RAM usage is about 160GB during training. 1TB RAM installed on the machine.

To narrow it down, it's happening somewhere here: https://github.com/Slicer/Slicer/blob/d7080d6a519df05983f9aebf9e0abd7cbfbc02e6/Base/Python/slicer/parameterNodeWrapper/wrapper.py#L199-L202 During the creation of that dictionary, the selected value is overwritten.

@jcfr Thanks a lot for the help today. Is it uncommon to use the slicer.dicomDatabase in tests? If so, are there any examples how to handle such situations?

@cpinter I just checked with the nightly build on Windows. It takes 1 minutes until the surfaces appear in the 3D view. When using SegmentEditor only, this issue doesn't occur....

I figured that it has to do with observing events of the segmentation Commenting the following code out, fixes the performance issue. https://github.com/QIICR/QuantitativeReporting/blob/905436c0c2939e77e7de4fe480e32e5d209214a8/QuantitativeReporting/QuantitativeReporting.py#L513-L522 I will investigate a bit more to...

@cpinter Not sure why, but for showing the 3D surface, the `SegmentModified` event is invoked for each segment to be displayed in 3D. Is that on purpose and if so,...