AUTHOR=Shahid Husnain , Khalid Adnan , Liu Xin , Irfan Muhammad , Ta Dean 

TITLE=A Deep Learning Approach for the Photoacoustic Tomography Recovery From Undersampled Measurements

JOURNAL=Frontiers in Neuroscience

VOLUME=Volume 15 - 2021

YEAR=2021

URL=https://www.frontiersin.org/journals/neuroscience/articles/10.3389/fnins.2021.598693

DOI=10.3389/fnins.2021.598693

ISSN=1662-453X

ABSTRACT=Photoacoustic tomography (PAT) is a promising imaging modality, which is helpful for bio-medical study. However, challenges remain in fast PAT imaging. To address the problem, recently, the methods based on compressed sensing (CS) have been proposed, which allows low computational cost and high resolution for implementing PAT. Nevertheless, the imaging results of the sparsity-based methods strictly rely on sparsity and incoherence conditions. Hence, it is challenging for ensuring that the experimental acquired photoacoustic data meet the prerequisite conditions of CS. To overcome the limitations, in this work, a deep-learning-based PAT (Deep-PAT) method is introduced. By using U-NET, Deep-PAT not only able to reconstruct PAT imaging from less number of measurements without taking the prerequisite conditions of CS into consideration but also removes the under-sampled artifacts effectively. The experimental results demonstrate that Deep-PAT is capable to reconstruct the high-quality photoacoustic image by just using 5% of the original measurement data. In addition, compared to the sparsity-based method, it can be seen through statistical analysis that quality is significantly improved 30% (approximately), having SSIM=0.97 and PSNR=27.53dB, by the proposed Deep-PAT method.