Signal-domain speed-of-sound correction for ring-array-based photoacoustic tomography

Abstract

Photoacoustic imaging combines the advantages of optical and acoustic imaging, making it a promising tool in biomedical imaging. Photoacoustic tomography (PAT) reconstructs images by solving the inverse problem from detected photoacoustic waves to initial pressure map. The heterogeneous speed of sound (SoS) distribution in biological tissue significantly affects image quality, as uncertain SoS variations can cause image distortions. Previously reported dual-speed-of-sound (dual-SoS) imaging methods effectively address these distortions by accounting for the SoS differences between tissues and the coupling medium. However, these methods require recalculating the distribution parameters of the SoS for each frame during dynamic imaging, which is highly time-consuming and poses a significant challenge for achieving real-time dynamic dual-SoS PAT imaging. To address this issue, we propose a signal-domain dual-SoS correction method for PAT image reconstruction. In this method, two distinct SoS regions are differentiated by recognizing the photoacoustic signal features of the imaging target's contours. The signals are then corrected based on the respective SoS values, enabling signal-domain-based dual-SoS dynamic real-time PAT imaging. The proposed method was validated through numerical simulations and in-vivo experiments of human finger. The results show that, compared to the single-SoS reconstruction method, the proposed approach produces higher-quality images, achieving the resolution error by near 12 times and a 30 % increase in contrast. Furthermore, the method enables dual-SoS dynamic real-time PAT reconstruction at 10 fps, which is 187.22 % faster than existing dual-SoS reconstruction methods, highlighting its feasibility for dynamic PAT imaging of heterogeneous tissues.

Keywords: Image reconstruction; Photoacoustic tomography; Speed correction; Speed of sound.

Fig. 1

Dual speed-of-sound (SoS) distribution in the ring-array PAT system for signal propagation and image reconstruction. (a) SoS distribution during PA signal detection. (b) SoS map configuration for PA image reconstruction. The dashed circle represents the ring-array ultrasound transducer, the red square indicates the imaging region, and the black and white backgrounds represent different SoS regions.

Fig. 2

Schematic illustration of acoustic wave propagation in different medium. (a) Ring-array PAT system configuration. (b) Typical waveform of a single PA signal in a dual-SoS medium. (c) Sinogram of all PA signals in a dual-SoS medium. (d) Original and SoS-corrected waveforms of a PA signal. The different background colors represent distinct regions with different speeds of sound.

Fig. 3

Workflow of dual-SoS PAT image reconstruction with signal-domain SoS correction. For each frame of PA signals, a signal-domain SoS map is first extracted. Based on the SoS values, signal-domain correction is then performed on the PA waveforms to compensate for dual-SoS propagation effects. Finally, the corrected signals are used for image reconstruction to obtain a PA image with accurate acoustic localization.

Fig. 4

Simulation experiment setup. (a) and (b) show the imaging phantom and the configuration of the ring-array transducer. (c) and (d) display the corresponding SoS distributions used in the simulation setups shown in (a) and (b), respectively. SoS: speed of sound.

Fig. 5

Setup of the PAT system for in vivo human finger imaging. The trigger synchronizes the DAQ system and the laser. The fiber bundle delivers laser pulses to illuminate the finger, while the UT array detects the generated PA waves. The signals are acquired by the DAQ system and transferred to the computer for PAT image reconstruction. DAQ: data acquisition; UT: ultrasound transducer.

Fig. 6

Simulation results of PA signals. (a) and (d) show the computed signal-domain SoS maps corresponding to the phantoms in Figs. 4(a) and 4(b). (b) and (e) present the raw PA signals without SoS correction, while (c) and (f) show the PA signals after applying signal-domain dual-SoS correction. The black and white backgrounds represent different SoS regions, and PA signal intensity is encoded by color.

Fig. 7

Simulation results of PAT image reconstruction with different SoS settings. (a)-i to (a)-iii and (b)-i to (b)-iii show reconstructed images using single SoS values of 1480 m/s, 1560 m/s, and 1520 m/s, respectively. (a)-iv and (b)-iv present the reconstructed images using the proposed signal-domain dual-SoS correction method.

Fig. 8

Pixel value profiles of single-SoS and dual-SoS reconstruction results. (a)–(d) correspond to the dotted lines indicated in Figs. 4(a) and 4(b). The black dashed line represents the ground truth from the simulation setup; the red dash-dot line indicates the dual-SoS reconstruction result; the blue solid line shows the single-SoS reconstruction result. GT: ground truth; D-SoS: dual speed of sound; S-SoS: single speed of sound.

Fig. 9

In-vivo human finger cross-sectional PAT imaging results. (a)–(d) show the reconstructed images using a single SoS of 1520 m/s. (e)–(h) present the corresponding images reconstructed with the PA signals which corrected by the proposed signal-domain dual-SoS method (1500 m/s for water and 1560 m/s for tissue). The red dashed line highlights the same vascular cross-sectional regions in both reconstruction methods for comparison.

Fig. 10

Pixel value profiles of single-SoS and dual-SoS in-vivo human finger PAT imaging results. (a)–(d) correspond to the dotted lines indicated in Fig. 9(a)–(d), comparing the reconstructed pixel intensities between single-SoS and signal-domain dual-SoS correction methods. The blue solid line represents the single-SoS reconstruction result, while the red dotted line represents the dual-SoS reconstruction result. SoS: speed of sound.

Fig. 11

Time consumption comparison between conventional dual-SoS PAT imaging and the proposed signal-domain dual-SoS correction method. (a) Detailed time consumption for each processing step. (b) Proportional breakdown of time consumption for delay-and-sum reconstruction using the proposed signal-domain dual-SoS correction method.