Pseudoprogression estimator does not work on either GUI or CLI

[sarthakpati]

Describe the bug Followed documentation on sample data and didn't get any results.

To Reproduce Steps to reproduce the behavior:

1. Select the "Training Directory"
2. Select the "Output Directory"
3. Click on 'Confirm'.
4. Output directory is empty (for both training and testing)

Command for Training:

./PseudoProgressionEstimator.exe -t 0 -i C:/PseudoprogressionInputDirectory -o C:/outputDir

Command for Testing:

./PseudoProgressionEstimator.exe -t 1 -i C:/PseudoprogressionInputDirectory -m C:/PseudoprogressionInputModelDirectory -o C:/outputDir

Ref: https://cbica.github.io/CaPTk/Glioblastoma_Pseudoprogression.html

CaPTk Version 1.8.0.Alpha

[ashishsingh18]

@sarthakpati Can you please attach screenshots of what you did via the GUI and cmd for this? And also of the error you are getting. It is difficult to understand what you did and what results you get with the information you have.

[sarthakpati]

I don't have any screenshots but I have added more information.

[ashishsingh18]

I want to try this on GUI side and want to know what error you are getting and in what case. Please post screenshots, without which I can't proceed. Also for the cmd line please post screenshots of your cmd window.

[sarthakpati]

The results folder is empty, as I have said in the report. Do you want a screenshot of an empty directory?

I just pasted the full command I used, which is far more helpful than posting screenshots of the terminal.

[ashishsingh18]

I want a screenshot of what you did on the GUI so I can reproduce.

And the same thing on the command window. This should take you about 10 sec each.

[sarthakpati]

GUI:

I am still not sure how this will help compared to the actual commands and steps I have illustrated on my original report.

[ashishsingh18]

For the cmd your screenshot doesn't show what happens when you run the commands.

However, this helped for the GUI side. I was able to run this and can confirm that this works and produces results. I am not sure what is not working on your side or if you forgot to click on 'confirm'.

[sarthakpati]

On the CLI:

On the GUI, I just saw the message "finished successfully" without any files in the output directory in either GUI or CLI mode.

I am not sure what I have forgotten in this particular scenario.

[saimarathore]

@sarthakpati based on the screenshot that you have provided (Glioblastoma Pseudo-Progression Estimator), it seems like that you flipped model and test data folders when you ran from the GUI.

[sarthakpati]

@sarthakpati based on the screenshot that you have provided (Glioblastoma Pseudo-Progression Estimator), it seems like that you flipped model and test data folders when you ran from the GUI.

That was a typo when I was trying to take a screenshot in the middle of doing 10 other things. Anyway, here is what I actually used:

[sarthakpati]

The PR #1185 does not fix this (using the new model with the provided sample data):

(base) PS C:\CaPTk_Full\1.8.0.Alpha2\bin> .\PseudoProgressionEstimator.exe -m "C:\PseudoProgressionEstimator_PretrainedModel_new\" -i "C:\PseudoProgressionEstimator" -o "C:\test_psp" -t 1
Input directory name:C:\PseudoProgressionEstimator
Output directory name:C:\test_psp
Model directory name:C:\PseudoProgressionEstimator_PretrainedModel_new\
Module loaded: Pseudoprogression Estimation on Existing Model:
2ABEM
ABFC
Number of subjects with required input: 2
(base) PS C:\CaPTk_Full\1.8.0.Alpha2\bin> ls C:\test_psp
(base) PS C:\CaPTk_Full\1.8.0.Alpha2\bin> 

[sarthakpati]

One of the reasons why testing is not working is this line:

https://github.com/CBICA/CaPTk/blob/master/src/applications/PseudoProgressionEstimator.cpp#L276

Which looks for the file chiharu_featurefile.csv in the output directory, which it should not.

[sarthakpati]

One of the reasons why testing is not working is this line:

https://github.com/CBICA/CaPTk/blob/master/src/applications/PseudoProgressionEstimator.cpp#L276

Which looks for the file chiharu_featurefile.csv in the output directory, which it should not.

This has now changed. Will post after I do a train session on the cluster.

[saimarathore]

this has been fixed