Deep learning enable real-time photoacoustic tomography system via single data acquisition channel

Abstract

Photoacoustic tomography (PAT) is an emerging technology for biomedical imaging that combines the superiorities of high optical contrast and acoustic penetration. In the PAT system, more photoacoustic (PA) signals are preferred to be detected to reconstruct PA image with higher fidelity. However, more PA signals’ detection leads to more time consumption for single-channel scanning based PAT system, or higher cost of data acquisition (DAQ) module for array-based PAT system. To address this issue, we proposed a real-time PAT system only using single DAQ channel, and a deep learning method for PA signal recover and image reconstruction. We superimpose 30 channels’ signals together, shrinking to 4 channels (120/30=4). Furthermore, a four-to-one delay-line module is designed to combine this 4 channels’ data into one channel DAQ. In order to reconstruct the image from four superimposed 30-channels’ PA signals, we train a dedicated deep learning model to reconstruct final PA image.