Zero-Shot Artifact2Artifact: Self-incentive artifact removal for photoacoustic imaging

Abstract

Three-dimensional (3D) photoacoustic imaging (PAI) with detector arrays has shown superior imaging capabilities in biomedical applications. However, the quality of 3D PAI is often degraded due to reconstruction artifacts caused by sparse detectors. Existing iterative or deep learning-based methods are either time-consuming or require large training datasets, limiting their practical application. Here, we propose Zero-Shot Artifact2Artifact (ZS-A2A), a zero-shot self-supervised artifact removal method based on a super-lightweight network, which leverages the fact that patterns of artifacts are more sensitive to sensor data loss. By randomly dropping acquired PA data, it spontaneously generates subset data to reconstruct images, which in turn stimulates the network to learn the artifact patterns in reconstruction results, thus enabling zero-shot artifact removal. This approach requires neither training data nor prior knowledge of the artifacts, making it suitable for artifact removal for arbitrary detector array configurations. We validated ZS-A2A in both simulation study and $invivo$ animal experiments. Results demonstrate that ZS-A2A achieves high performance compared to existing zero-shot methods.