Abstract:Photoacoustic (PA) imaging is a biomedical imaging technique that combines the principles of both optics and acoustics to visualize tissues and structures inside the body. It involves the generation of acoustic waves by the absorption of pulsed laser light in tissues, and the subsequent detection of these acoustic waves to form images. The acoustic-resolution photoacoustic microscopy (AR-PAM) is a simple but powerful photoacoustic imaging method. However, there are two major challenges for AR-PAM existed: the quickly degrading in spatial resolution in the out-of-focus region and the image reconstruction artefacts due to the limited-view issue. Those two limitations were caused by the strong scattering of optical-acoustic phenomenon. To enhance the image quality, we reported that the deep learning (DL) convolutional neural network (CNN) can not only significantly enhance the resolution of AR-PAM but also help correcting reconstruction artifacts. Simulated PA data are used to train the network to approximate the real physical structures from scanned images. Not only finished the simulation study, we also validated our method by both phantom and in vivo animal studies. Phantoms are used to verify the full width half maximum of object in out-of-focus area can be significantly enhanced, while that in the focal area are also improved. In the animal studies, a rat’s abdomen was also input into the network for prediction. The result indicates that in animal images, utilizing repeated the DL processing twice, three-dimensional structural information can gain the clarity. Our results demonstrated that DL could play an important role of the imaging enhancement and 3D structural defuzzification for AR-PAM system.

Y-Net: Hybrid deep learning image reconstruction for photoacoustic tomography in vivo

Y-Net: A Hybrid Deep Learning Reconstruction Framework for Photoacoustic Imaging in vivo

Quantitative Endoscopic Photoacoustic Tomography Using a Convolutional Neural Network.

Deep-Learning Image Reconstruction for Real-Time Photoacoustic System

Deep learning-based photoacoustic imaging of vascular network through thick porous media

AS-Net: Fast Photoacoustic Reconstruction With Multi-Feature Fusion From Sparse Data

HDN:Hybrid Deep-learning and Non-line-of-sight Reconstruction Framework for Photoacoustic Brain Imaging

Multikernel positional embedding convolutional neural network for photoacoustic reconstruction with sparse data

End-to-end Res-Unet based reconstruction algorithm for photoacoustic imaging

Reconstructing the Photoacoustic Image with High Quality using the Deep Neural Network Model

Domain Transform Network for Photoacoustic Tomography from Limited-view and Sparsely Sampled Data

Imaging Enhancement for Acoustic-Resolution Photoacoustic Microscopy by Deep Learning

UPAMNet: A unified network with deep knowledge priors for photoacoustic microscopy

Sparse-view Signal-domain Photoacoustic Tomography Reconstruction Method Based on Neural Representation

Ultra-Sparse Reconstruction for Photoacoustic Tomography: Sinogram Domain Prior-Guided Method Exploiting Enhanced Score-Based Diffusion Model

Improved Photoacoustic Imaging of Numerical Bone Model Based on Attention Block U-Net Deep Learning Network

FBP-Net for Direct Reconstruction of Dynamic PET Images.

A Jointed Feature Fusion Framework for Photoacoustic Reconstruction

Deep Learning for Photoacoustic Tomography from Sparse Data

An iterative gradient convolutional neural network and its application in endoscopic photoacoustic image formation from incomplete acoustic measurement

PA-NeRF, a neural radiance field model for 3D photoacoustic tomography reconstruction from limited Bscan data