通过深度学习辅助迭代算法进行分层相位检索。

IF 6.1 3区 材料科学 Q1 Biochemistry, Genetics and Molecular Biology Journal of Applied Crystallography Pub Date : 2024-08-19 eCollection Date: 2024-10-01 DOI:10.1107/S1600576724006897
Koki Yamada, Natsuki Akaishi, Kohei Yatabe, Yuki Takayama
{"title":"通过深度学习辅助迭代算法进行分层相位检索。","authors":"Koki Yamada, Natsuki Akaishi, Kohei Yatabe, Yuki Takayama","doi":"10.1107/S1600576724006897","DOIUrl":null,"url":null,"abstract":"<p><p>Ptychography is a powerful computational imaging technique with microscopic imaging capability and adaptability to various specimens. To obtain an imaging result, it requires a phase-retrieval algorithm whose performance directly determines the imaging quality. Recently, deep neural network (DNN)-based phase retrieval has been proposed to improve the imaging quality from the ordinary model-based iterative algorithms. However, the DNN-based methods have some limitations because of the sensitivity to changes in experimental conditions and the difficulty of collecting enough measured specimen images for training the DNN. To overcome these limitations, a ptychographic phase-retrieval algorithm that combines model-based and DNN-based approaches is proposed. This method exploits a DNN-based denoiser to assist an iterative algorithm like ePIE in finding better reconstruction images. This combination of DNN and iterative algorithms allows the measurement model to be explicitly incorporated into the DNN-based approach, improving its robustness to changes in experimental conditions. Furthermore, to circumvent the difficulty of collecting the training data, it is proposed that the DNN-based denoiser be trained without using actual measured specimen images but using a formula-driven supervised approach that systemically generates synthetic images. In experiments using simulation based on a hard X-ray ptychographic measurement system, the imaging capability of the proposed method was evaluated by comparing it with ePIE and rPIE. These results demonstrated that the proposed method was able to reconstruct higher-spatial-resolution images with half the number of iterations required by ePIE and rPIE, even for data with low illumination intensity. Also, the proposed method was shown to be robust to its hyperparameters. In addition, the proposed method was applied to ptychographic datasets of a Simens star chart and ink toner particles measured at SPring-8 BL24XU, which confirmed that it can successfully reconstruct images from measurement scans with a lower overlap ratio of the illumination regions than is required by ePIE and rPIE.</p>","PeriodicalId":14950,"journal":{"name":"Journal of Applied Crystallography","volume":"57 Pt 5","pages":"1323-1335"},"PeriodicalIF":6.1000,"publicationDate":"2024-08-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11460392/pdf/","citationCount":"0","resultStr":"{\"title\":\"Ptychographic phase retrieval via a deep-learning-assisted iterative algorithm.\",\"authors\":\"Koki Yamada, Natsuki Akaishi, Kohei Yatabe, Yuki Takayama\",\"doi\":\"10.1107/S1600576724006897\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<p><p>Ptychography is a powerful computational imaging technique with microscopic imaging capability and adaptability to various specimens. To obtain an imaging result, it requires a phase-retrieval algorithm whose performance directly determines the imaging quality. Recently, deep neural network (DNN)-based phase retrieval has been proposed to improve the imaging quality from the ordinary model-based iterative algorithms. However, the DNN-based methods have some limitations because of the sensitivity to changes in experimental conditions and the difficulty of collecting enough measured specimen images for training the DNN. To overcome these limitations, a ptychographic phase-retrieval algorithm that combines model-based and DNN-based approaches is proposed. This method exploits a DNN-based denoiser to assist an iterative algorithm like ePIE in finding better reconstruction images. This combination of DNN and iterative algorithms allows the measurement model to be explicitly incorporated into the DNN-based approach, improving its robustness to changes in experimental conditions. Furthermore, to circumvent the difficulty of collecting the training data, it is proposed that the DNN-based denoiser be trained without using actual measured specimen images but using a formula-driven supervised approach that systemically generates synthetic images. In experiments using simulation based on a hard X-ray ptychographic measurement system, the imaging capability of the proposed method was evaluated by comparing it with ePIE and rPIE. These results demonstrated that the proposed method was able to reconstruct higher-spatial-resolution images with half the number of iterations required by ePIE and rPIE, even for data with low illumination intensity. Also, the proposed method was shown to be robust to its hyperparameters. In addition, the proposed method was applied to ptychographic datasets of a Simens star chart and ink toner particles measured at SPring-8 BL24XU, which confirmed that it can successfully reconstruct images from measurement scans with a lower overlap ratio of the illumination regions than is required by ePIE and rPIE.</p>\",\"PeriodicalId\":14950,\"journal\":{\"name\":\"Journal of Applied Crystallography\",\"volume\":\"57 Pt 5\",\"pages\":\"1323-1335\"},\"PeriodicalIF\":6.1000,\"publicationDate\":\"2024-08-19\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11460392/pdf/\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Journal of Applied Crystallography\",\"FirstCategoryId\":\"88\",\"ListUrlMain\":\"https://doi.org/10.1107/S1600576724006897\",\"RegionNum\":3,\"RegionCategory\":\"材料科学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"2024/10/1 0:00:00\",\"PubModel\":\"eCollection\",\"JCR\":\"Q1\",\"JCRName\":\"Biochemistry, Genetics and Molecular Biology\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of Applied Crystallography","FirstCategoryId":"88","ListUrlMain":"https://doi.org/10.1107/S1600576724006897","RegionNum":3,"RegionCategory":"材料科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"2024/10/1 0:00:00","PubModel":"eCollection","JCR":"Q1","JCRName":"Biochemistry, Genetics and Molecular Biology","Score":null,"Total":0}
引用次数: 0

摘要

层析成像技术是一种功能强大的计算成像技术,具有显微成像能力和对各种标本的适应性。要获得成像结果,需要一种相位检索算法,其性能直接决定了成像质量。最近,有人提出了基于深度神经网络(DNN)的相位检索,以改善基于普通模型的迭代算法的成像质量。然而,基于 DNN 的方法有一些局限性,因为它对实验条件的变化很敏感,而且很难收集到足够的测量标本图像来训练 DNN。为了克服这些局限性,我们提出了一种将基于模型和基于 DNN 的方法相结合的分层相位检索算法。该方法利用基于 DNN 的去噪器来帮助 ePIE 等迭代算法找到更好的重建图像。这种 DNN 与迭代算法的结合使测量模型可以明确地纳入基于 DNN 的方法中,从而提高其对实验条件变化的鲁棒性。此外,为了避免收集训练数据的困难,建议不使用实际测量的标本图像,而是使用公式驱动的监督方法来系统地生成合成图像,从而训练基于 DNN 的去噪器。在基于硬 X 射线层析成像测量系统的模拟实验中,通过与 ePIE 和 rPIE 比较,评估了所提方法的成像能力。这些结果表明,所提出的方法能够重建更高的空间分辨率图像,而所需的迭代次数仅为 ePIE 和 rPIE 的一半,即使是低照度数据也是如此。而且,所提出的方法对其超参数具有鲁棒性。此外,该方法还被应用于在 SPring-8 BL24XU 上测量的 Simens 星图和墨粉颗粒的双色图像数据集,结果表明,该方法可以成功地从测量扫描中重建图像,且照明区域的重叠率低于 ePIE 和 rPIE 的要求。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
查看原文
分享 分享
微信好友 朋友圈 QQ好友 复制链接
本刊更多论文
Ptychographic phase retrieval via a deep-learning-assisted iterative algorithm.

Ptychography is a powerful computational imaging technique with microscopic imaging capability and adaptability to various specimens. To obtain an imaging result, it requires a phase-retrieval algorithm whose performance directly determines the imaging quality. Recently, deep neural network (DNN)-based phase retrieval has been proposed to improve the imaging quality from the ordinary model-based iterative algorithms. However, the DNN-based methods have some limitations because of the sensitivity to changes in experimental conditions and the difficulty of collecting enough measured specimen images for training the DNN. To overcome these limitations, a ptychographic phase-retrieval algorithm that combines model-based and DNN-based approaches is proposed. This method exploits a DNN-based denoiser to assist an iterative algorithm like ePIE in finding better reconstruction images. This combination of DNN and iterative algorithms allows the measurement model to be explicitly incorporated into the DNN-based approach, improving its robustness to changes in experimental conditions. Furthermore, to circumvent the difficulty of collecting the training data, it is proposed that the DNN-based denoiser be trained without using actual measured specimen images but using a formula-driven supervised approach that systemically generates synthetic images. In experiments using simulation based on a hard X-ray ptychographic measurement system, the imaging capability of the proposed method was evaluated by comparing it with ePIE and rPIE. These results demonstrated that the proposed method was able to reconstruct higher-spatial-resolution images with half the number of iterations required by ePIE and rPIE, even for data with low illumination intensity. Also, the proposed method was shown to be robust to its hyperparameters. In addition, the proposed method was applied to ptychographic datasets of a Simens star chart and ink toner particles measured at SPring-8 BL24XU, which confirmed that it can successfully reconstruct images from measurement scans with a lower overlap ratio of the illumination regions than is required by ePIE and rPIE.

求助全文
通过发布文献求助,成功后即可免费获取论文全文。 去求助
来源期刊
CiteScore
10.00
自引率
3.30%
发文量
178
审稿时长
4.7 months
期刊介绍: Many research topics in condensed matter research, materials science and the life sciences make use of crystallographic methods to study crystalline and non-crystalline matter with neutrons, X-rays and electrons. Articles published in the Journal of Applied Crystallography focus on these methods and their use in identifying structural and diffusion-controlled phase transformations, structure-property relationships, structural changes of defects, interfaces and surfaces, etc. Developments of instrumentation and crystallographic apparatus, theory and interpretation, numerical analysis and other related subjects are also covered. The journal is the primary place where crystallographic computer program information is published.
期刊最新文献
SUBGROUPS: a computer tool at the Bilbao Crystallographic Server for the study of pseudo-symmetric or distorted structures. Characterization of sub-micrometre-sized voids in fixed human brain tissue using scanning X-ray microdiffraction. Electronic angle focusing for neutron time-of-flight powder diffractometers. Link between b.c.c.-f.c.c. orientation relationship and austenite morphology in CF8M stainless steel. In situ counter-diffusion crystallization and long-term crystal preservation in microfluidic fixed targets for serial crystallography.
×
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
现在去查看 取消
×
提示
确定
0
微信
客服QQ
Book学术公众号 扫码关注我们
反馈
×
意见反馈
请填写您的意见或建议
请填写您的手机或邮箱
已复制链接
已复制链接
快去分享给好友吧!
我知道了
×
扫码分享
扫码分享
Book学术官方微信
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术
文献互助 智能选刊 最新文献 互助须知 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 Book学术 All rights reserved.
ghs 京公网安备 11010802042870号 京ICP备2023020795号-1