Multi-frame pixel intensity consistency-based artifact reduction for photoacoustic computed tomography

Photoacoustic computed tomography (PACT) is an advanced imaging modality that combines the strengths of ultrasound and optical imaging, showing great promise for preclinical or clinical applications. However, image quality in PACT is often compromised by artifacts, which are typically unavoidable during reconstruction. To address this issue, we propose a novel artifact reduction algorithm that achieves dynamic artifact reduction via pixel intensity consistency. This method is based on the key physical observation that artifacts change with variations in the relative positions between true anatomical structures and the transducer array, whereas the reconstructed intensities of true features remain largely consistent. By exploiting the differential sensitivity of artifacts and true features to target motion, it effectively distinguishes artifact regions from true features, leading to robust artifact reduction. Compared with other single-frame-based algorithms, numerical simulation results verify that our method achieves the highest SSIM and PSNR values relative to the ground truth. Furthermore, in a 600-frame RF dataset, the proposed approach exhibits an approximately 100-fold enhancement in average per-frame processing speed compared with the conventional state-of-the-art artifact reduction method, thereby emerging as a promising candidate for enabling real-time artifact reduction in PACT.