Abstract: Backprojection imaging with dense arrays serves as a vital tool for investigating earthquake source physics, characterizing seismic energy release patterns, and constraining fault structures. This technique generates timely, robust rupture images by tracking the spatiotemporal migration of seismic radiation through phase-aligned coherent wavefields recorded by dense networks, with the advantage of least prior constraints. With the deployment of dense seismic networks (Hi-net array (Japan), USArray (USA), Australia array (AU), etc) around the globe, this method has been widely applied to earthquakes in diverse tectonic settings, including subduction zones (e.g. 2011 Tohoku earthquake), continental interiors (e.g. 2021 Maduo earthquake), and transform boundaries (e.g. 2023 Kahramanmaraş earthquake doublet). These studies provide a unique perspective for unraveling the complex and multi-scale rupture processes of large earthquakes. Furthermore, they reveal a ubiquitous frequency-dependent radiation style among subduction earthquakes, shedding light on seismic radiation patterns in similar tectonic environments.This paper reviews recent methodological developments and applications of this technique. Methodological advances primarily focus on four key areas: the formulation of the waveform contribution term, the correction of seismic phase travel times, the application of array waveform processing techniques, and the design of weighting schemes. Different formulations of the waveform contribution term serve to amplify the signals of interest. For instance, a function based on waveform cross-correlation can enhance the resolution of neighboring weak subevents, which are often obscured by their low amplitudes in raw waveform data. The correction of travel times is necessary due to the difference between the assumed 1D velocity model and the real 3D velocity structure. Various array waveform processing techniques can further improve the spatiotemporal resolution of backprojection results. A weighting term scales the contribution of each station, thereby mitigating biases from uneven station distribution or anomalous waveform amplitudes.By synthesizing studies on earthquake ruptures across diverse tectonic settings, we highlight its critical contributions to resolve coseismic rupture process and seismic radiation characteristics. For example, the backprojection imaging results for the 2012 Sumatra earthquake indicate a complex rupture path, with the rupture propagating on different fault planes within an orthogonal conjugate fault system. Back-projection results for subduction-zone earthquakes, such as the 2011 Mw 9.0 Tohoku event, reveal a ubiquitous frequency-dependent radiation pattern related to depth-varying frictions along subduction plates.Finally, potential future directions are discussed based on the method’s unique advantages. Integrating machine learning with the backprojection method allows for the extraction of more subtle features from seismic waveforms, advancing the detection and location of small earthquakes. Furthermore, multi-scale rupture behaviors of earthquakes in diverse tectonic settings are to be investigated using backprojection results across a broad frequency range.