This repository houses the filter backprojection codes, which have been crafted in CUDA/C. The codes are then seamlessly wrapped within the TensorFlow Op framework (works for TF<2).
The implementation primarily serves as the forward projection module in the research paper:
A Quality‐checked and Physics‐constrained Deep Learning Method to Estimate Material Basis Images from Single‐KV Contrast‐enhanced Chest CT Scans published in Medical Physics.
- Languages & Technologies Used: CUDA/C
- Integration: TensorFlow Op
Li Y*, Tie X*, Li K, Zhang R, Qi Z, Budde A, Grist TM, Chen GH. A quality-checked and physics-constrained deep learning method to estimate material basis images from single-kV contrast-enhanced chest CT scans. Med Phys. 2023 Jun;50(6):3368-3388. doi: 10.1002/mp.16352. Epub 2023 Mar 23. PMID: 36908250; PMCID: PMC10330050.