<p>Photoacoustic tomography (PAT) is a biomedical imaging modality. In PAT imaging, a short-pulsed laser beam irradiates biological tissues and the caused acoustic pressure signals are acquired by transducers around the tissue. In static PAT imaging, adequate transducers are positioned on a full circle surrounding the tissue. Such case requires more transducers and thus leads to the high cost. In dynamic PAT, fewer transducers are required and they are placed only on an arc around the tissue, which results in limited data acquisition in the collection at a time instant. Therefore, the arc of the transducers is rotated and acoustic data from multiple angles are collected at different time instants. On the other hand, a living tissue may exhibit slight movements, which means that data acquired from the same spatial location at different instants may actually correspond to different tissue positions. Dynamic photoacoustic tomography (PAT) reconstruction seeks to reconstruct a sequence of PAT images by using photoacoustic signals captured at corresponding time instants. In this paper, we construct an efficient nonconvex model for dynamic PAT reconstruction with simultaneous motion estimation. The nonconvex total variation with overlapping group sparsity regularization for each PAT image, and the detail-preserving regularization for velocity field between two consecutive PAT images that describes the tissue motion are employed in the proposed model. A proximal alternating reweighted minimization algorithm is designed to solve the proposed model and the convergence analysis of the proposed algorithm is established. Numerical results validate the effectiveness of the proposed method.</p>