A Learning-based Framework for Spatial Impulse Response Compensation in 3D Photoacoustic Computed Tomography.

Abstract

Photoacoustic computed tomography (PACT) is an emerging imaging modality that combines the molecular contrast of optical absorption with the high spatial resolution of ultrasound detection. Utilizing ultrasound transducers with larger surface areas can improve detection sensitivity. However, when analytic reconstruction methods that neglect the spatial impulse responses (SIRs) of the transducer are employed, the spatial resolution of the reconstructed images will be compromised. Although optimization-based reconstruction methods can explicitly account for SIR effects, their computational cost is generally high, particularly in three-dimensional (3D) applications. To enable accurate but rapid 3D PACT image reconstruction, this study presents a framework for establishing a learned data-domain SIR compensation method that maps SIR-corrupted PACT measurement data to compensated data that would have been recorded by idealized point-like transducers. Subsequently, the compensated data can be used with a computationally efficient reconstruction method that neglects SIR effects. Two variants of the learned compensation model are investigated: a purely data-driven modeland a specifically designed, physics-inspired model, referred to as Deconv-Net, with training data generated by a fast, analytical procedure. The framework is rigorously validated in virtual imaging studies, demonstrating resolution improvement and robustness to noise variations, object complexity, and sound speed heterogeneity. When applied to in-vivo breast imaging data, the proposed models revealed fine structures that had been obscured by SIR-induced artifacts and achieved at least 30% reduction in relative squared error compared with the baseline. To our knowledge, this is the first demonstration of learned SIR compensation in 3D PACT imaging.