This repository contains Perception Fused Iterative Tomography Reconstruction Engine (PFITRE), which integrates a convolutional neural network (CNN) with perceptional knowledge as a smart regularizer into an iterative solving engine to correct limited-angle induced artifacts in X-ray tomography images. We demonstrate the effectiveness of the proposed approach using various experimental datasets obtained with different x-ray microscopy techniques.
The model weights can be downloaded using the following link: https://drive.google.com/file/d/1rqop4dAZ5QSjZluPkQnnMj5Qkmn5gtKo/view?usp=drive_link
We provide an approach for using PFITRE for iterative correction, as well as a one-time post-correction option with the pretrained network for testing and comparison.
PFITRE for 2D and 3D tomography images
If you use PFITRE model, we would appreciate your references to our paper.
Zhao, C., Ge, M., Yang, X. et al. Limited-angle x-ray nano-tomography with machine-learning enabled iterative reconstruction engine. npj Comput Mater 11, 240 (2025).
@article{PFITRE,
title = {Limited-angle x-ray nano-tomography with machine-learning enabled iterative reconstruction engine},
author = {Chonghang Zhao, Mingyuan Ge, Xiaogang Yang, Yong S. Chu, Hanfei Yan},
url = {https://www.nature.com/articles/s41524-025-01724-0},
doi = {https://doi.org/10.1038/s41524-025-01724-0},
year = {2025}
}
