ImagingDataCommons / IDC-Tutorials

Self-guided notebook tutorials to help get started with using IDC
BSD 3-Clause "New" or "Revised" License
27 stars 14 forks source link

/bin/bash: line 1: plastimatch: command not found #46

Open TrinhNguyenP opened 8 months ago

TrinhNguyenP commented 8 months ago

Hi IDC,

I am following the tutorial named lung_nodules_demo.ipynb located in IDC-Tutorials/notebooks/deprecated. I encounter an error as below:

Screenshot 2024-01-12 at 11 18 45 AM

Can you please guide me how to fix it?

Thanks, Trinh

fedorov commented 8 months ago

@TrinhNguyenP we fixed that issue for another notebook that you can see here: https://github.com/ImagingDataCommons/IDC-Tutorials/issues/30.

@denbonte is the developer of the notebook that doesn't work for you, and the one who identified the solution for plastimatch installation. Would you be willing to update this notebook as well Dennis?

denbonte commented 8 months ago

Thanks for opening this issue @TrinhNguyenP.

@fedorov I will look into the notebook ASAP!

denbonte commented 8 months ago

In the meantime, @TrinhNguyenP, since this will require a bit of work on our side (due to the deprecation, for the current Google Colab, of multiple components we used in that notebook), you might be interested in taking a look at this repository - which stores a series of notebooks we wrote, running the classification model from Hosny et al. on IDC data [1][2][3].

Let me know if I can help with anything else - I will try to fix the notebook you linked in the following days!

TrinhNguyenP commented 8 months ago

In the meantime, @TrinhNguyenP, since this will require a bit of work on our side (due to the deprecation, for the current Google Colab, of multiple components we used in that notebook), you might be interested in taking a look at this repository - which stores a series of notebooks we wrote, running the classification model from Hosny et al. on IDC data [1][2][3].

Let me know if I can help with anything else - I will try to fix the notebook you linked in the following days!

Hi @denbonte, I was able to install plastimatch and pyplastimatch. However, I am comimg up with another error with this line: model = keras.models.model_from_config(model_json) This is from the section Data processing, which loads the pre-trained weights for DeepPrognosis and run the last part of the pipeline.). I will shoot you a message if I can’t resolve the error following the suggestion from @deepakri201. Thanks, Trinh

denbonte commented 8 months ago

Hey @TrinhNguyenP,

I was able to install plastimatch and pyplastimatch.

Awesome!

However, I am comimg up with another error with this line: model = keras.models.model_from_config(model_json) This is from the section Data processing, which loads the pre-trained weights for DeepPrognosis and run the last part of the pipeline.).

Yes. Unfortunately, Colab dropped the support for Tensorflow 1.x - which means that many models trained with TF 1.x (or Keras based on TF 1.x) were left in this state.

The way we solved this, as you can see in the repo I linked to you above, was by converting the Deep Prognosis model into the open ONNX format and using the ONNX engine to run the inference phase. We made the model available as part of the idc-radiomics-reproducibility repository, and have examples of how to run that in the notebooks.

since this will require a bit of work on our side (due to the deprecation, for the current Google Colab, of multiple components we used in that notebook),

This is exactly what I was planning to do, by the way. Clean up the notebook, pull the 22.04LTS binaries for Plastimatch, switch to the ONNX implementation of the model, and solve any other dependency issues. I'm still planning to tackle that whenever I have a moment, but if you want to give it a try following the other notebooks, do let me know how it goes!

Cheers, Dennis.