Multi-stage deep learning artifact reduction for parallel-beam computed tomography.

IF 2.5 3区 物理与天体物理
Journal of Synchrotron Radiation Pub Date : 2025-03-01 Epub Date: 2025-02-17 DOI:10.1107/S1600577525000359
Jiayang Shi, Daniël M Pelt, K Joost Batenburg
{"title":"Multi-stage deep learning artifact reduction for parallel-beam computed tomography.","authors":"Jiayang Shi, Daniël M Pelt, K Joost Batenburg","doi":"10.1107/S1600577525000359","DOIUrl":null,"url":null,"abstract":"<p><p>Computed tomography (CT) using synchrotron radiation is a powerful technique that, compared with laboratory CT techniques, boosts high spatial and temporal resolution while also providing access to a range of contrast-formation mechanisms. The acquired projection data are typically processed by a computational pipeline composed of multiple stages. Artifacts introduced during data acquisition can propagate through the pipeline and degrade image quality in the reconstructed images. Recently, deep learning has shown significant promise in enhancing image quality for images representing scientific data. This success has driven increasing adoption of deep learning techniques in CT imaging. Various approaches have been proposed to incorporate deep learning into computational pipelines, but each has limitations in addressing artifacts effectively and efficiently in synchrotron CT, either in properly addressing the specific artifacts or in computational efficiency. Recognizing these challenges, we introduce a novel method that incorporates separate deep learning models at each stage of the tomography pipeline - projection, sinogram and reconstruction - to address specific artifacts locally in a data-driven way. Our approach includes bypass connections that feed both the outputs from previous stages and raw data to subsequent stages, minimizing the risk of error propagation. Extensive evaluations on both simulated and real-world datasets illustrate that our approach effectively reduces artifacts and outperforms comparison methods.</p>","PeriodicalId":48729,"journal":{"name":"Journal of Synchrotron Radiation","volume":" ","pages":"442-456"},"PeriodicalIF":2.5000,"publicationDate":"2025-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11892890/pdf/","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of Synchrotron Radiation","FirstCategoryId":"101","ListUrlMain":"https://doi.org/10.1107/S1600577525000359","RegionNum":3,"RegionCategory":"物理与天体物理","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"2025/2/17 0:00:00","PubModel":"Epub","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0

Abstract

Computed tomography (CT) using synchrotron radiation is a powerful technique that, compared with laboratory CT techniques, boosts high spatial and temporal resolution while also providing access to a range of contrast-formation mechanisms. The acquired projection data are typically processed by a computational pipeline composed of multiple stages. Artifacts introduced during data acquisition can propagate through the pipeline and degrade image quality in the reconstructed images. Recently, deep learning has shown significant promise in enhancing image quality for images representing scientific data. This success has driven increasing adoption of deep learning techniques in CT imaging. Various approaches have been proposed to incorporate deep learning into computational pipelines, but each has limitations in addressing artifacts effectively and efficiently in synchrotron CT, either in properly addressing the specific artifacts or in computational efficiency. Recognizing these challenges, we introduce a novel method that incorporates separate deep learning models at each stage of the tomography pipeline - projection, sinogram and reconstruction - to address specific artifacts locally in a data-driven way. Our approach includes bypass connections that feed both the outputs from previous stages and raw data to subsequent stages, minimizing the risk of error propagation. Extensive evaluations on both simulated and real-world datasets illustrate that our approach effectively reduces artifacts and outperforms comparison methods.

并行束计算机断层扫描的多阶段深度学习伪影减少。
使用同步辐射的计算机断层扫描(CT)是一种强大的技术,与实验室CT技术相比,它提高了高空间和时间分辨率,同时也提供了一系列对比度形成机制。获取的投影数据通常由由多个阶段组成的计算管道进行处理。在数据采集过程中引入的伪影会在管道中传播,降低重建图像的图像质量。最近,深度学习在提高代表科学数据的图像的图像质量方面显示出了巨大的希望。这一成功推动了深度学习技术在CT成像中的应用。已经提出了将深度学习纳入计算管道的各种方法,但每种方法在同步加速器CT中有效和高效地处理伪影方面都有局限性,无论是在正确处理特定伪影还是在计算效率方面。认识到这些挑战,我们引入了一种新颖的方法,该方法在层析管道的每个阶段(投影、正弦图和重建)中集成了单独的深度学习模型,以数据驱动的方式在局部处理特定的工件。我们的方法包括旁路连接,将前一阶段的输出和原始数据馈送到后续阶段,从而最大限度地降低错误传播的风险。对模拟和现实世界数据集的广泛评估表明,我们的方法有效地减少了人为影响,优于比较方法。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
求助全文
约1分钟内获得全文 求助全文
来源期刊
Journal of Synchrotron Radiation
Journal of Synchrotron Radiation INSTRUMENTS & INSTRUMENTATIONOPTICS&-OPTICS
CiteScore
5.60
自引率
12.00%
发文量
289
审稿时长
1 months
期刊介绍: Synchrotron radiation research is rapidly expanding with many new sources of radiation being created globally. Synchrotron radiation plays a leading role in pure science and in emerging technologies. The Journal of Synchrotron Radiation provides comprehensive coverage of the entire field of synchrotron radiation and free-electron laser research including instrumentation, theory, computing and scientific applications in areas such as biology, nanoscience and materials science. Rapid publication ensures an up-to-date information resource for scientists and engineers in the field.
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:604180095
Book学术官方微信