Thanks to visit codestin.com
Credit goes to github.com

Skip to content

Claritas is Python tools for denoising PET images using SwinUNETR and UNET-based neural networks. It includes scripts for preprocessing .nrrd PET data, training models with or without tumor masks, and evaluating performance. The repository is designed for research purposes only. The developed models can be implemented in SlicerPETDenoise module.

License

Notifications You must be signed in to change notification settings

4burakfe/Claritas

Repository files navigation

Claritas

Claritas is a collection of standalone Python tools for denoising PET images using SwinUNETR-based neural networks. It includes scripts for preprocessing .nrrd PET data, training models with or without tumor masks, and evaluating performance. The repository is designed for research purposes only. The developed models can be directly implemented in SlicerPETDenoise Extension which is available from here: https://github.com/4burakfe/SlicerPETDenoise


🛠️ Script Descriptions

Claritas_Train_UI_single_wmasks.py

GUI application for training PET denoising models with tumor masks.

Features:

  • Supports SwinUNETR and UNet
  • Fully customizable loss functions (MSE, MAE, SSIM, tumor-weighted L1)
  • Real-time training plots and logging
  • Uses .npy files for training and validation
  • The input files shold have be in pairs in seperate folders with exact same names

Claritas_Train_UI_single_nomask.py

Same as above but for training without tumor segmentation masks.

batch_nrrd_to_npy_fullvol.py

Resamples .nrrd PET files to 2×2×2 mm Crops central 256×256 area Saves full volumes as .npy

batch_nrrd_to_npy_gridpatch.py

Extracts non-overlapping patches of size 64×64×64 Saves each patch as a separate .npy file

🧠 Model Descriptions

Models are designed to use with SlicerPETDenoise Extension which is available from here: https://github.com/4burakfe/SlicerPETDenoise

Filename Description
pet_denoiser_std_char.pth Trained with standard Charbonnier loss for general denoising.
pet_denoiser_x19w1_5.pth Tumor-weighted loss model (1.5x penalty for underestimation).
pet_denoiser_x14w3.pth Strong tumor-focused model (3x weight for underestimation).

📦 Requirements Python 3.9+ PyTorch MONAI SimpleITK numpy matplotlib einops

For installation of torch please visit https://pytorch.org For other dependencies you can use command: "pip install monai einops SimpleITK matplotlib numpy"

Disclaimer: This software is for research purposes only and is not certified for clinical use.

About

Claritas is Python tools for denoising PET images using SwinUNETR and UNET-based neural networks. It includes scripts for preprocessing .nrrd PET data, training models with or without tumor masks, and evaluating performance. The repository is designed for research purposes only. The developed models can be implemented in SlicerPETDenoise module.

Resources

License

Stars

Watchers

Forks

Languages