Spectral-distortion-suppressed deep learning for fiber sensor photoacoustic microscopy

Liying Zhu; Xiaoxuan Zhong; Xuanhao Zhang; Huan Cheng; Long Jin; Yizhi Liang; Lidai Wang

doi:10.1016/j.pacs.2025.100774

Spectral-distortion-suppressed deep learning for fiber sensor photoacoustic microscopy

Liying Zhu (Co-first Author)
, Xiaoxuan Zhong (Co-first Author)
, Xuanhao Zhang
, Huan Cheng
, Long Jin
, Yizhi Liang^*
, Lidai Wang^*

^*Corresponding author for this work

Research output: Journal Publications and Reviews › RGC 21 - Publication in refereed journal › peer-review

6 Downloads (CityUHK Scholars)

Abstract

Fiber laser sensors offer significant advantages for photoacoustic microscopy (PAM), including compact size, electromagnetic immunity, and suitability for fast scanning systems. However, its signal-to-noise ratio (SNR) may rapidly degrade when the field of view (FOV) is enlarged. This compromised SNR adversely affects the accuracy of blood oxygen saturation (sO₂) derived from noisy photoacoustic signals. To address this problem, a two-stage deep learning framework for fiber laser sensor-based PAM is proposed. The first stage reduces the 3D data to 2D image and suppresses the noises. The second stage integrates the dual-wavelengths images and suppresses the spectral distortion, so that the accuracy of sO₂ can be preserved. The network performance is validated using imaging datasets acquired with a conventional high-SNR photoacoustic microscopy system. Results demonstrate that this approach does not only denoise images acquired with the unfocused fiber laser sensor, but also maintains high fidelity in sO₂ calculation, addressing a key challenge in fast functional PAM. © 2025 The Authors.

Original language	English
Article number	100774
Number of pages	11
Journal	Photoacoustics
Volume	46
Online published	10 Oct 2025
DOIs	https://doi.org/10.1016/j.pacs.2025.100774
Publication status	Published - Dec 2025

Funding

This work was partially supported by the Research Grants Council of the Hong Kong Special Administrative Region (11104922), the National Natural Science Foundation of China (NSFC) (61805102, 62322506, 62275104, 62135006), the China Postdoctoral Science Foundation (2025M770849), National Key Research and Development Program of China (2023YFF0715302).

Research Keywords

Deep learning
Fiber sensor
Image denoising
Photoacoustic microscopy

Publisher's Copyright Statement

This full text is made available under CC-BY-NC-ND 4.0. https://creativecommons.org/licenses/by-nc-nd/4.0/

RGC Funding Information

RGC-funded

Access Output

10.1016/j.pacs.2025.100774

350609064.pdfFinal Published version, 3.72 MBLicence: CC BY-NC-ND

Other Access Options

Check CityUHK Library

Permanent Link

Right click to copy link

Cite this

@article{04936598cad647fb84fd719dfbda55da,

title = "Spectral-distortion-suppressed deep learning for fiber sensor photoacoustic microscopy",

abstract = "Fiber laser sensors offer significant advantages for photoacoustic microscopy (PAM), including compact size, electromagnetic immunity, and suitability for fast scanning systems. However, its signal-to-noise ratio (SNR) may rapidly degrade when the field of view (FOV) is enlarged. This compromised SNR adversely affects the accuracy of blood oxygen saturation (sO2) derived from noisy photoacoustic signals. To address this problem, a two-stage deep learning framework for fiber laser sensor-based PAM is proposed. The first stage reduces the 3D data to 2D image and suppresses the noises. The second stage integrates the dual-wavelengths images and suppresses the spectral distortion, so that the accuracy of sO2 can be preserved. The network performance is validated using imaging datasets acquired with a conventional high-SNR photoacoustic microscopy system. Results demonstrate that this approach does not only denoise images acquired with the unfocused fiber laser sensor, but also maintains high fidelity in sO2 calculation, addressing a key challenge in fast functional PAM. {\textcopyright} 2025 The Authors.",

keywords = "Deep learning, Fiber sensor, Image denoising, Photoacoustic microscopy",

author = "Liying Zhu and Xiaoxuan Zhong and Xuanhao Zhang and Huan Cheng and Long Jin and Yizhi Liang and Lidai Wang",

year = "2025",

month = dec,

doi = "10.1016/j.pacs.2025.100774",

language = "English",

volume = "46",

journal = "Photoacoustics",

issn = "2213-5979",

publisher = "Elsevier GmbH",

}

TY - JOUR

T1 - Spectral-distortion-suppressed deep learning for fiber sensor photoacoustic microscopy

AU - Zhu, Liying

AU - Zhong, Xiaoxuan

AU - Zhang, Xuanhao

AU - Cheng, Huan

AU - Jin, Long

AU - Liang, Yizhi

AU - Wang, Lidai

PY - 2025/12

Y1 - 2025/12

N2 - Fiber laser sensors offer significant advantages for photoacoustic microscopy (PAM), including compact size, electromagnetic immunity, and suitability for fast scanning systems. However, its signal-to-noise ratio (SNR) may rapidly degrade when the field of view (FOV) is enlarged. This compromised SNR adversely affects the accuracy of blood oxygen saturation (sO2) derived from noisy photoacoustic signals. To address this problem, a two-stage deep learning framework for fiber laser sensor-based PAM is proposed. The first stage reduces the 3D data to 2D image and suppresses the noises. The second stage integrates the dual-wavelengths images and suppresses the spectral distortion, so that the accuracy of sO2 can be preserved. The network performance is validated using imaging datasets acquired with a conventional high-SNR photoacoustic microscopy system. Results demonstrate that this approach does not only denoise images acquired with the unfocused fiber laser sensor, but also maintains high fidelity in sO2 calculation, addressing a key challenge in fast functional PAM. © 2025 The Authors.

AB - Fiber laser sensors offer significant advantages for photoacoustic microscopy (PAM), including compact size, electromagnetic immunity, and suitability for fast scanning systems. However, its signal-to-noise ratio (SNR) may rapidly degrade when the field of view (FOV) is enlarged. This compromised SNR adversely affects the accuracy of blood oxygen saturation (sO2) derived from noisy photoacoustic signals. To address this problem, a two-stage deep learning framework for fiber laser sensor-based PAM is proposed. The first stage reduces the 3D data to 2D image and suppresses the noises. The second stage integrates the dual-wavelengths images and suppresses the spectral distortion, so that the accuracy of sO2 can be preserved. The network performance is validated using imaging datasets acquired with a conventional high-SNR photoacoustic microscopy system. Results demonstrate that this approach does not only denoise images acquired with the unfocused fiber laser sensor, but also maintains high fidelity in sO2 calculation, addressing a key challenge in fast functional PAM. © 2025 The Authors.

KW - Deep learning

KW - Fiber sensor

KW - Image denoising

KW - Photoacoustic microscopy

UR - https://www.scopus.com/pages/publications/105018308104

UR - https://www.scopus.com/record/pubmetrics.uri?eid=2-s2.0-105018308104&origin=recordpage

U2 - 10.1016/j.pacs.2025.100774

DO - 10.1016/j.pacs.2025.100774

M3 - RGC 21 - Publication in refereed journal

C2 - 41142359

SN - 2213-5979

VL - 46

JO - Photoacoustics

JF - Photoacoustics

M1 - 100774

ER -

Spectral-distortion-suppressed deep learning for fiber sensor photoacoustic microscopy

Abstract

Funding

Research Keywords

Publisher's Copyright Statement

RGC Funding Information

Access Output

Other Access Options

Permanent Link

Other Files and Links

Fingerprint

GRF: Wireless Headmount Photoacoustic Tomography of the Brain in Free-moving Animals

Cite this

Spectral-distortion-suppressed deep learning for fiber sensor photoacoustic microscopy

Abstract

Funding

Research Keywords

Publisher's Copyright Statement

RGC Funding Information

Access Output

Other Access Options

Permanent Link

Other Files and Links

Fingerprint

Projects

GRF: Wireless Headmount Photoacoustic Tomography of the Brain in Free-moving Animals

Cite this