Deep neural networks enabled isotropic quantitative differential phase contrast microscopy

An Cin Li; Yu Hsiang Lin; Hsuan Ming Huang; Yuan Luo

doi:10.1117/12.2615974

Deep neural networks enabled isotropic quantitative differential phase contrast microscopy

An Cin Li
, Yu Hsiang Lin
, Hsuan Ming Huang
, Yuan Luo

National Taiwan University
National Tsing Hua University

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

Isotropic quantitative differential phase contrast (iDPC) microscopy based on pupil engineering has made significant improvement in reconstructing phase image of weak phase objects. In previous researches, the pupil designs have been investigated for enhancing the data acquisition efficiency. To further improve the phase retrieval procedure in iDPC, we adapt deep neural networks to achieve isotropic phase distribution from half-pupil based quantitative differential phase contrast (qDPC) microscopy. In this study, we utilized U-net model for mapping from 1-axis phase reconstruction to 12- axis one. The results show that the deep neural network we proposed achieved expecting performance. The final testing loss value of our model after 1000 epochs of training achieved 6.7e-5 after normalized. The peak signal to noise ratio improvement is from 26dB to 30dB.

Original language	English
Title of host publication	Biomedical Imaging and Sensing Conference 2021
Editors	Toyohiko Yatagai, Yoshihisa Aizu, Osamu Matoba, Yasuhiro Awatsuji, Yuan Luo
Publisher	SPIE
ISBN (Electronic)	9781510647190
DOIs	https://doi.org/10.1117/12.2615974
State	Published - 2021
Externally published	Yes
Event	Biomedical Imaging and Sensing Conference 2021 - Virtual, Online, Japan Duration: 20 04 2021 → 22 04 2021

Publication series

Name	Proceedings of SPIE - The International Society for Optical Engineering
Volume	11925
ISSN (Print)	0277-786X
ISSN (Electronic)	1996-756X

Conference

Conference	Biomedical Imaging and Sensing Conference 2021
Country/Territory	Japan
City	Virtual, Online
Period	20/04/21 → 22/04/21

Bibliographical note

Publisher Copyright:
© COPYRIGHT SPIE. Downloading of the abstract is permitted for personal use only.

Keywords

Deep neural network
Isotropic quantitative differential phase contrast microscopy
Patch-based
Phase retrieval

Access to Document

10.1117/12.2615974

Cite this

Li, A. C., Lin, Y. H., Huang, H. M., & Luo, Y. (2021). Deep neural networks enabled isotropic quantitative differential phase contrast microscopy. In T. Yatagai, Y. Aizu, O. Matoba, Y. Awatsuji, & Y. Luo (Eds.), Biomedical Imaging and Sensing Conference 2021 Article 119251P (Proceedings of SPIE - The International Society for Optical Engineering; Vol. 11925). SPIE. https://doi.org/10.1117/12.2615974

Li, An Cin ; Lin, Yu Hsiang ; Huang, Hsuan Ming et al. / Deep neural networks enabled isotropic quantitative differential phase contrast microscopy. Biomedical Imaging and Sensing Conference 2021. editor / Toyohiko Yatagai ; Yoshihisa Aizu ; Osamu Matoba ; Yasuhiro Awatsuji ; Yuan Luo. SPIE, 2021. (Proceedings of SPIE - The International Society for Optical Engineering).

@inproceedings{dc7b5231d5004159a615e10e7116bec9,

title = "Deep neural networks enabled isotropic quantitative differential phase contrast microscopy",

abstract = "Isotropic quantitative differential phase contrast (iDPC) microscopy based on pupil engineering has made significant improvement in reconstructing phase image of weak phase objects. In previous researches, the pupil designs have been investigated for enhancing the data acquisition efficiency. To further improve the phase retrieval procedure in iDPC, we adapt deep neural networks to achieve isotropic phase distribution from half-pupil based quantitative differential phase contrast (qDPC) microscopy. In this study, we utilized U-net model for mapping from 1-axis phase reconstruction to 12- axis one. The results show that the deep neural network we proposed achieved expecting performance. The final testing loss value of our model after 1000 epochs of training achieved 6.7e-5 after normalized. The peak signal to noise ratio improvement is from 26dB to 30dB.",

keywords = "Deep neural network, Isotropic quantitative differential phase contrast microscopy, Patch-based, Phase retrieval",

author = "Li, \{An Cin\} and Lin, \{Yu Hsiang\} and Huang, \{Hsuan Ming\} and Yuan Luo",

note = "Publisher Copyright: {\textcopyright} COPYRIGHT SPIE. Downloading of the abstract is permitted for personal use only.; Biomedical Imaging and Sensing Conference 2021 ; Conference date: 20-04-2021 Through 22-04-2021",

year = "2021",

doi = "10.1117/12.2615974",

language = "英语",

series = "Proceedings of SPIE - The International Society for Optical Engineering",

publisher = "SPIE",

editor = "Toyohiko Yatagai and Yoshihisa Aizu and Osamu Matoba and Yasuhiro Awatsuji and Yuan Luo",

booktitle = "Biomedical Imaging and Sensing Conference 2021",

address = "美国",

}

Li, AC, Lin, YH, Huang, HM & Luo, Y 2021, Deep neural networks enabled isotropic quantitative differential phase contrast microscopy. in T Yatagai, Y Aizu, O Matoba, Y Awatsuji & Y Luo (eds), Biomedical Imaging and Sensing Conference 2021., 119251P, Proceedings of SPIE - The International Society for Optical Engineering, vol. 11925, SPIE, Biomedical Imaging and Sensing Conference 2021, Virtual, Online, Japan, 20/04/21. https://doi.org/10.1117/12.2615974

Deep neural networks enabled isotropic quantitative differential phase contrast microscopy. / Li, An Cin; Lin, Yu Hsiang; Huang, Hsuan Ming et al.
Biomedical Imaging and Sensing Conference 2021. ed. / Toyohiko Yatagai; Yoshihisa Aizu; Osamu Matoba; Yasuhiro Awatsuji; Yuan Luo. SPIE, 2021. 119251P (Proceedings of SPIE - The International Society for Optical Engineering; Vol. 11925).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Deep neural networks enabled isotropic quantitative differential phase contrast microscopy

AU - Li, An Cin

AU - Lin, Yu Hsiang

AU - Huang, Hsuan Ming

AU - Luo, Yuan

N1 - Publisher Copyright: © COPYRIGHT SPIE. Downloading of the abstract is permitted for personal use only.

PY - 2021

Y1 - 2021

N2 - Isotropic quantitative differential phase contrast (iDPC) microscopy based on pupil engineering has made significant improvement in reconstructing phase image of weak phase objects. In previous researches, the pupil designs have been investigated for enhancing the data acquisition efficiency. To further improve the phase retrieval procedure in iDPC, we adapt deep neural networks to achieve isotropic phase distribution from half-pupil based quantitative differential phase contrast (qDPC) microscopy. In this study, we utilized U-net model for mapping from 1-axis phase reconstruction to 12- axis one. The results show that the deep neural network we proposed achieved expecting performance. The final testing loss value of our model after 1000 epochs of training achieved 6.7e-5 after normalized. The peak signal to noise ratio improvement is from 26dB to 30dB.

AB - Isotropic quantitative differential phase contrast (iDPC) microscopy based on pupil engineering has made significant improvement in reconstructing phase image of weak phase objects. In previous researches, the pupil designs have been investigated for enhancing the data acquisition efficiency. To further improve the phase retrieval procedure in iDPC, we adapt deep neural networks to achieve isotropic phase distribution from half-pupil based quantitative differential phase contrast (qDPC) microscopy. In this study, we utilized U-net model for mapping from 1-axis phase reconstruction to 12- axis one. The results show that the deep neural network we proposed achieved expecting performance. The final testing loss value of our model after 1000 epochs of training achieved 6.7e-5 after normalized. The peak signal to noise ratio improvement is from 26dB to 30dB.

KW - Deep neural network

KW - Isotropic quantitative differential phase contrast microscopy

KW - Patch-based

KW - Phase retrieval

UR - https://www.scopus.com/pages/publications/85120459515

U2 - 10.1117/12.2615974

DO - 10.1117/12.2615974

M3 - 会议稿件

AN - SCOPUS:85120459515

T3 - Proceedings of SPIE - The International Society for Optical Engineering

BT - Biomedical Imaging and Sensing Conference 2021

A2 - Yatagai, Toyohiko

A2 - Aizu, Yoshihisa

A2 - Matoba, Osamu

A2 - Awatsuji, Yasuhiro

A2 - Luo, Yuan

PB - SPIE

T2 - Biomedical Imaging and Sensing Conference 2021

Y2 - 20 April 2021 through 22 April 2021

ER -

Deep neural networks enabled isotropic quantitative differential phase contrast microscopy

Abstract

Publication series

Conference

Bibliographical note

Keywords

Access to Document

Other files and links

Fingerprint

Cite this