Deep learning-based spatio-temporal fusion for high-fidelity ultra-high-speed X-ray radiography

Full-field ultra-high-speed (UHS) X-ray imaging experiments have been well established to characterize various processes and phenomena. However, the potential of UHS experiments through the joint acquisition of X-ray videos with distinct configurations has not been fully exploited. In this paper, we...

Full description

Saved in:
Bibliographic Details
Main Authors: Songyuan Tang, Tekin Bicer, Tao Sun, Kamel Fezzaa, Samuel J. Clark
Format: Article
Language:English
Published: International Union of Crystallography 2025-03-01
Series:Journal of Synchrotron Radiation
Subjects:
Online Access:https://journals.iucr.org/paper?S1600577525000323
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850252588030623744
author Songyuan Tang
Tekin Bicer
Tao Sun
Kamel Fezzaa
Samuel J. Clark
author_facet Songyuan Tang
Tekin Bicer
Tao Sun
Kamel Fezzaa
Samuel J. Clark
author_sort Songyuan Tang
collection DOAJ
description Full-field ultra-high-speed (UHS) X-ray imaging experiments have been well established to characterize various processes and phenomena. However, the potential of UHS experiments through the joint acquisition of X-ray videos with distinct configurations has not been fully exploited. In this paper, we investigate the use of a deep learning-based spatio-temporal fusion (STF) framework to fuse two complementary sequences of X-ray images and reconstruct the target image sequence with high spatial resolution, high frame rate and high fidelity. We applied a transfer learning strategy to train the model and compared the peak signal-to-noise ratio (PSNR), average absolute difference (AAD) and structural similarity (SSIM) of the proposed framework on two independent X-ray data sets with those obtained from a baseline deep learning model, a Bayesian fusion framework and the bicubic interpolation method. The proposed framework outperformed the other methods with various configurations of the input frame separations and image noise levels. With three subsequent images from the low-resolution (LR) sequence of a four times lower spatial resolution and another two images from the high-resolution (HR) sequence of a 20 times lower frame rate, the proposed approach achieved average PSNRs of 37.57 dB and 35.15 dB, respectively. When coupled with the appropriate combination of high-speed cameras, the proposed approach will enhance the performance and therefore the scientific value of UHS X-ray imaging experiments.
format Article
id doaj-art-c9c76b08a8b34108b4aef6b52037026a
institution OA Journals
issn 1600-5775
language English
publishDate 2025-03-01
publisher International Union of Crystallography
record_format Article
series Journal of Synchrotron Radiation
spelling doaj-art-c9c76b08a8b34108b4aef6b52037026a2025-08-20T01:57:35ZengInternational Union of CrystallographyJournal of Synchrotron Radiation1600-57752025-03-0132243244110.1107/S1600577525000323tv5068Deep learning-based spatio-temporal fusion for high-fidelity ultra-high-speed X-ray radiographySongyuan Tang0Tekin Bicer1Tao Sun2Kamel Fezzaa3Samuel J. Clark4Advanced Photon Source, Argonne National Laboratory, Lemont, IL 60439, USAAdvanced Photon Source, Argonne National Laboratory, Lemont, IL 60439, USADepartment of Mechanical Engineering, Northwestern University, Evanston, IL 60208, USAAdvanced Photon Source, Argonne National Laboratory, Lemont, IL 60439, USAAdvanced Photon Source, Argonne National Laboratory, Lemont, IL 60439, USAFull-field ultra-high-speed (UHS) X-ray imaging experiments have been well established to characterize various processes and phenomena. However, the potential of UHS experiments through the joint acquisition of X-ray videos with distinct configurations has not been fully exploited. In this paper, we investigate the use of a deep learning-based spatio-temporal fusion (STF) framework to fuse two complementary sequences of X-ray images and reconstruct the target image sequence with high spatial resolution, high frame rate and high fidelity. We applied a transfer learning strategy to train the model and compared the peak signal-to-noise ratio (PSNR), average absolute difference (AAD) and structural similarity (SSIM) of the proposed framework on two independent X-ray data sets with those obtained from a baseline deep learning model, a Bayesian fusion framework and the bicubic interpolation method. The proposed framework outperformed the other methods with various configurations of the input frame separations and image noise levels. With three subsequent images from the low-resolution (LR) sequence of a four times lower spatial resolution and another two images from the high-resolution (HR) sequence of a 20 times lower frame rate, the proposed approach achieved average PSNRs of 37.57 dB and 35.15 dB, respectively. When coupled with the appropriate combination of high-speed cameras, the proposed approach will enhance the performance and therefore the scientific value of UHS X-ray imaging experiments.https://journals.iucr.org/paper?S1600577525000323x-ray imagingdeep learninghigh-speed imagingspatio-temporal fusionfull-field x-ray radiography
spellingShingle Songyuan Tang
Tekin Bicer
Tao Sun
Kamel Fezzaa
Samuel J. Clark
Deep learning-based spatio-temporal fusion for high-fidelity ultra-high-speed X-ray radiography
Journal of Synchrotron Radiation
x-ray imaging
deep learning
high-speed imaging
spatio-temporal fusion
full-field x-ray radiography
title Deep learning-based spatio-temporal fusion for high-fidelity ultra-high-speed X-ray radiography
title_full Deep learning-based spatio-temporal fusion for high-fidelity ultra-high-speed X-ray radiography
title_fullStr Deep learning-based spatio-temporal fusion for high-fidelity ultra-high-speed X-ray radiography
title_full_unstemmed Deep learning-based spatio-temporal fusion for high-fidelity ultra-high-speed X-ray radiography
title_short Deep learning-based spatio-temporal fusion for high-fidelity ultra-high-speed X-ray radiography
title_sort deep learning based spatio temporal fusion for high fidelity ultra high speed x ray radiography
topic x-ray imaging
deep learning
high-speed imaging
spatio-temporal fusion
full-field x-ray radiography
url https://journals.iucr.org/paper?S1600577525000323
work_keys_str_mv AT songyuantang deeplearningbasedspatiotemporalfusionforhighfidelityultrahighspeedxrayradiography
AT tekinbicer deeplearningbasedspatiotemporalfusionforhighfidelityultrahighspeedxrayradiography
AT taosun deeplearningbasedspatiotemporalfusionforhighfidelityultrahighspeedxrayradiography
AT kamelfezzaa deeplearningbasedspatiotemporalfusionforhighfidelityultrahighspeedxrayradiography
AT samueljclark deeplearningbasedspatiotemporalfusionforhighfidelityultrahighspeedxrayradiography