Acquisition-inspired motion artifact modeling and degradation-consistent diffusion restoration in photoacoustic microscopy
- 1Key Laboratory of Brain Health Intelligent Evaluation and Intervention (Beijing Institute of Technology), Ministry of Education, China; School of Medical Science and Engineering, Beijing Institute of Technology, China
- 2Key Laboratory of Brain Health Intelligent Evaluation and Intervention (Beijing Institute of Technology), Ministry of Education, China; Frontier Interdisciplinary Domain, Beijing Institute of Technology, Zhuhai, China
- 3Key Laboratory of Brain Health Intelligent Evaluation and Intervention (Beijing Institute of Technology), Ministry of Education, China; School of Computer Science and Technology, Beijing Institute of Technology, China
Abstract
Motion artifacts are a major source of degradation in photoacoustic microscopy (PAM), where sequential raster scanning converts target-probe motion into structured coordinate mismatch, causing line misregistration, vessel discontinuity, truncation, and local deformation. We propose an acquisition-inspired framework for PAM motion artifact modeling and degradation-guided diffusion restoration. Motion corruption is formulated as raster-scan coordinate mismatch and decomposed into inter-line misregistration and intra-line sampling distortion. Based on this formulation, we construct a controllable simulator that generates paired clean-corrupted PAM images with explicit degradation priors. A conditional diffusion restoration model then uses motion-related priors for guidance and enforces degradation consistency by applying the forward inter-line and intra-line degradation operators during training. To support real-data inference, a degradation prior estimator predicts row-wise displacement and artifact-region information from corrupted images. Experiments on synthetic mouse brain PAM data show superior restoration performance, while selected real examples provide preliminary qualitative observations on non-synthetic motion-corrupted acquisitions.
Related
- Acoustic coherence-aware oriented 3D photoacoustic microscopy super-resolution with noise-suppression
- Dual Branch Transformer with Physical Priors for PSF Acquisition to Enable PSF Guided Image Restoration against Co Phasing Errors in Optical Synthetic Aperture Interferometric Imaging
- Metalens-transparent ultrasound transducer module for extended depth photoacoustic microscopy
- Reference-guided defocus estimation enables physics-guided image restoration in deep ultraviolet wafer microscopy
- Acquisition-calibrated degradation synthesis and dual-domain reconstruction for accelerated spectral-domain OCT
BibTeX
@article{wang2026acquisitioninspired,
title = {{Acquisition-inspired motion artifact modeling and degradation-consistent diffusion restoration in photoacoustic microscopy}},
author = {Heqing Wang and Shuyan Zhang and Jingtan Li and Xinyi Lim and Naidi Sun and Bin Hu},
journal = {Photoacoustics},
year = {2026},
month = oct,
doi = {10.1016/j.pacs.2026.100867},
url = {https://doi.org/10.1016/j.pacs.2026.100867},
note = {CC BY},
}
RIS
TY - JOUR
TI - Acquisition-inspired motion artifact modeling and degradation-consistent diffusion restoration in photoacoustic microscopy
AU - Heqing Wang
AU - Shuyan Zhang
AU - Jingtan Li
AU - Xinyi Lim
AU - Naidi Sun
AU - Bin Hu
PY - 2026
DA - 2026/10/29
JO - Photoacoustics
DO - 10.1016/j.pacs.2026.100867
UR - https://doi.org/10.1016/j.pacs.2026.100867
AB - Motion artifacts are a major source of degradation in photoacoustic microscopy (PAM), where sequential raster scanning converts target-probe motion into structured coordinate mismatch, causing line misregistration, vessel discontinuity, truncation, and local deformation. We propose an acquisition-inspired framework for PAM motion artifact modeling and degradation-guided diffusion restoration. Motion corruption is formulated as raster-scan coordinate mismatch and decomposed into inter-line misregistration and intra-line sampling distortion. Based on this formulation, we construct a controllable simulator that generates paired clean-corrupted PAM images with explicit degradation priors. A conditional diffusion restoration model then uses motion-related priors for guidance and enforces degradation consistency by applying the forward inter-line and intra-line degradation operators during training. To support real-data inference, a degradation prior estimator predicts row-wise displacement and artifact-region information from corrupted images. Experiments on synthetic mouse brain PAM data show superior restoration performance, while selected real examples provide preliminary qualitative observations on non-synthetic motion-corrupted acquisitions.
SN - 2213-5979
ER -