Paper
25 January 2024 Generative model for limited-view photoacoustic tomography reconstruction
Kangjun Guo, Zhiyuan Zheng, Guijun Wang, Xianlin Song
Author Affiliations +
Abstract
Limited-view photoacoustic tomography images will have a lot of artifacts and information loss when using conventional photoacoustic tomography image reconstruction algorithms. To solve this problem, this paper proposed a limited-view photoacoustic tomography reconstruction method based on a generative model. The network is trained through the noise-perturbed method and can learn a kind of scoring functions (gradients of logarithmic probability density functions) of the training dataset. The trained network has the capability to generate samples that conform to the distribution of the training dataset. Blood vessels simulation data were used to evaluate the performance of the proposed method. Experimental results on simulated blood vessels show that, compared with traditional reconstruction methods, the proposed method can effectively remove artifacts and improve image quality with measured data collected from 90°, 120°, and 180°.
(2024) Published by SPIE. Downloading of the abstract is permitted for personal use only.
Kangjun Guo, Zhiyuan Zheng, Guijun Wang, and Xianlin Song "Generative model for limited-view photoacoustic tomography reconstruction", Proc. SPIE 12972, International Academic Conference on Optics and Photonics (IACOP 2023), 129720A (25 January 2024); https://doi.org/10.1117/12.3022605
Advertisement
Advertisement
RIGHTS & PERMISSIONS
Get copyright permission  Get copyright permission on Copyright Marketplace
KEYWORDS
Image restoration

Photoacoustic spectroscopy

Education and training

Photoacoustic tomography

Blood vessels

Data modeling

Reconstruction algorithms

Back to Top