Spatiotemporal Coherence Weighting for In Vivo Cardiac Photoacoustic Image Beamformation

Abstract

Photoacoustic (PA) image reconstruction generally utilizes delay-and-sum (DAS) beamforming of received acoustic waves from tissue irradiated with optical illumination. However, nonadaptive DAS reconstructed cardiac PA images exhibit temporally varying noise which causes reduced myocardial PA signal specificity, making image interpretation difficult. Adaptive beamforming algorithms such as minimum variance (MV) with coherence factor (CF) weighting have been previously reported to improve the DAS image quality. In this article, we report on an adaptive beamforming algorithm by extending CF weighting to the temporal domain for preclinical cardiac PA imaging (PAI). The proposed spatiotemporal coherence factor (STCF) considers multiple temporally adjacent image acquisition events during beamforming and cancels out signals with low spatial coherence and temporal coherence, resulting in higher background noise cancellation while preserving the main features of interest (myocardial wall) in the resultant PA images. STCF has been validated using the numerical simulations and in vivo ECG and respiratory-signal-gated cardiac PAI in healthy murine hearts. The numerical simulation results demonstrate that STCF weighting outperforms DAS and MV beamforming with and without CF weighting under different levels of inherent contrast, acoustic attenuation, optical scattering, and signal-to-noise (SNR) of channel data. Performance improvement is attributed to higher sidelobe reduction (at least 5 dB) and SNR improvement (at least 10 dB). Improved myocardial signal specificity and higher signal rejection in the left ventricular chamber and acoustic gel region are observed with STCF in cardiac PAI.

Fig. 1.

Schematic diagram with key steps of the hybrid photoacoustic imaging simulation. Optical fluence distribution estimated using MCMatlab is used to generate the initial pressure distribution for k-Wave acoustic simulation. Finally, a beamforming algorithm is utilized to reconstruct PA images from the received channel data.

Fig. 2.

Beamformed images of simulated point targets in a high contrast background. (a) DAS, (b) DAS-CF, (c) DAS-STCF, (d) MV, (e) MV-CF and (f) MV-STCF. Display dynamic range is 65 dB. Green and red (dotted) rectangles denote signal and noise ROIs, respectively. STCF weighted images had the lowest level of background signal or noise.

Fig. 3.

Lateral profiles of PSF of at depth of (a) 8 mm and (b) 20 mm for low contrast, (c) 8 mm and (d) 20 mm for high contrast background. Profiles around center of the point targets are zoomed in and displayed in the insets. Both CF and STCF weighting had similar FWHM values with improvement over DAS and MV beamformer alone.

Fig. 4.

Variation of SNR for the simulated point targets with contrast variation at a depth of (a) 8 mm and (b) 20 mm, respectively. STCF weighting had higher SNR values attributed to better background signal suppression.

Fig. 5.

Beamformed images of simulated point targets under acoustic absorption with power law exponent, y = 1.5. (a) DAS, (b) DAS-CF, (c) DAS-STCF, (d) MV, (e) MV-CF and (f) MV-STCF. Display dynamic range is 65 dB.

Fig. 6.

Lateral profiles of PSF of at depth of (a) 8 mm and (b) 20 mm for acoustic absorption simulation with y = 1.5. Impact of acoustic attenuation and resultant depth dependent blurring effect is apparent in 20 mm results.

Fig. 7.

Variation of SNR with power law absorption exponent at a depth of (a) 8 mm and (b) 20 mm, respectively.

Fig. 8.

Beamformed images of simulated point targets under optical absorption and scattering (μ_s = 15cm⁻¹). (a) DAS, (b) DAS-CF, (c) DAS-STCF, (d) MV, (e) MV-CF and (f) MV-STCF. Display dynamic range is 65 dB.

Fig. 9.

Variation of SNR with background scattering (μ_s) at a depth of (a) 8 mm and (b) 20 mm, respectively.

Fig. 10.

Variation of DAS-STCF beamformer point target discernibility with background scattering of (a) μ_s = 10cm⁻¹, (b) μ_s = 15cm⁻¹ and (c) μ_s = 112cm⁻¹ respectively. Display dynamic range is 65 dB.

Fig. 11.

Variation of SNR with channel data SNR at a depth of (a) 8 mm and (b) 20 mm, respectively.

Fig. 12.

Variation of STCF beamformer SNR with ensemble length under (a) no acoustic and optical absorption, (b) acoustic absorption and (c) optical scattering, respectively.

Fig. 13.

Cardiac PAI M-mode image reconstructed using (a) DAS, (b) DAS-CF, (c) DAS-STCF, (d) MV, (e) MV-CF and (f) MV-STCF. Display dynamic range is 65 dB.

Fig. 14.

In vivo cardiac photoacoustic images at ED. (a) US B-mode, (b) System PA image, (c) DAS, (d) DAS-CF, (e) DAS -STCF, (f) MV, (g) MV-CF and (h) MV-STCF. Arrows in Fig. 14 (b) indicate signals impeding contrast between myocardium and surrounding muscle. ROI definitions in Fig. 14 (c): Green = LV chamber blood, black = myocardial wall, blue = muscle and white = noise. STCF weighting better suppressed signals from gel region and LV chamber.

Fig. 15.

In vivo cardiac photoacoustic images at ES. (a) US B-mode, (b) System PA image, (c) DAS, (d) DAS-CF, (e) DAS -STCF, (f) MV, (g) MV-CF and (h) MV-STCF. Arrows in Fig. 15 (b) indicate signals impeding contrast between myocardium and surrounding muscle. ROI definitions in Fig. 15 (c): Green = LV chamber blood, black = myocardial wall, blue = muscle and white = noise. STCF weighting better suppressed signals from gel region and LV chamber.

Fig. 16.

(a) In vivo SNR comparison. (b) and (d) show CR and gCNR comparison between myocardial wall and muscle, respectively. (c) and (e) show CR and gCNR comparison between myocardial wall and LV chamber blood, respectively.

Fig. 17.

(a) SNR variation with ensemble length (K). (b) and (d) show CR and gCNR variation between myocardial wall and muscle, respectively. (c) and (e) show CR and gCNR variation between myocardial wall and LV chamber blood, respectively.