Non‐Convex Low‐Rank Minimization‐Based Regularization Method for Quantitative Photoacoustic Tomography Reconstruction

ABSTRACT Photoacoustic tomography (PAT) is a new biomedical imaging modality that has been applied in biomedical imaging and clinical practice to help for diagnosing and monitoring diseases. In the PAT imaging, a short‐pulsed laser beam irradiates the biological tissue and causes acoustic waves in the tissue due to the photoacoustic effect. PAT reconstruction aims to obtain the initial acoustic pressure that contains the structural information of the tissue. Although the reconstructed PAT image is essential for knowing about the variety of the tissues, the optical absorption coefficient and diffusion coefficient of the tissue got by further quantitative photoacoustic tomography (QPAT) reconstruction are necessary for the accurate diagnosis. Therefore, QPAT reconstruction is an indispensable step in the clinical application of PAT. In this paper, the QPAT reconstruction is considered. We propose a non‐convex low‐rank minimization based regularization model for the QPAT reconstruction. A majorized alternating iterative algorithm is developed to solve the proposed model, and the convergence is demonstrated. Experimental results show that the optical absorption coefficient and diffusion coefficient of the tissue reconstructed by the proposed method are more accurate than those obtained by the existing methods for the QPAT reconstruction.