This repository provides deep learning models developed for CT-free attenuation correction (AC), Monte Carlo-based scatter correction (SC), and combined attenuation and scatter correction (ASC) in 90Y bremsstrahlung SPECT imaging, using a Swin UNETR architecture. These models were trained on dose maps to improve voxel-wise and organ-level dosimetry for patients undergoing radioembolization (RE) or Selective Internal Radiation Therapy (SIRT).
🧠 Read the refernce publication for more information::
Mansouri Z., Salimi Y., Bianchetto Wolf N., Mainta I., Zaidi H.
CT-free attenuation and Monte-Carlo based scatter correction-guided quantitative 90Y-SPECT imaging for improved dose calculation using deep learning
Eur J Nucl Med Mol Imaging, 2025
DOI:10.1007/s00259-025-07191-5
90Y-SPECT is widely used for post-therapy verification in liver cancer patients treated with radioembolization. However, its quantitative accuracy is compromised by photon attenuation and scatter, especially in CT-less settings. Traditional Monte Carlo methods are accurate but slow and hardware-dependent.
This work introduces fast and robust deep learning-based alternatives using dose-domain training on a relatively large patient dataset (n=190), leveraging a Swin UNETR model with transformer-based encoding.
These models are beneficial in:
Standalone SPECT systems without CT
Hospitals lacking GPU access for MC simulations
Reducing processing time (from ~80 min to ~20 sec/GPU or ~6 min/CPU)
https://drive.google.com/drive/folders/1FfzE_-_-mxiG6lyi4ap_JN6VlnThFfsd?usp=sharing
To install this repository, simply run:
pip install git+https://github.com/ZahraMansouriMedPhys/DL-90Y-SPECT-Correction.git
We welcome any feedback, suggestions, or contributions to improve this project!
for any furtehr question please email me at: zahra.mansouri@unige.ch