pathologydatascience / mutils_panoptic Goto Github PK

Hello
I could not find region_model_configs.py file in the repo, which is rewuired to run MuTILsTrainer.py

I'd like to replicate the results you've shown in your recent publication on PanopTILs. (Table 1 to start with but ideally every table and figure.) How would you recommend doing this?

My guess would be to start by running MuTILsTrainer.py, adjusting variables like BASEPATH to match the directory from which I've downloaded and unzipped PanopTILs bootstrapped set to. However, I noticed that the Trainer main calls utils.GeneralUtils.save_configs, which has been commented out in the GeneralUtils file. Should I avoid using it?

Once I have a trained model, would I use MuTILsWSIRunner or MuTILsAnalyticalValidation to evaluate my trained model on the manually labeled test set across the five folds?

Thank you for your help.