Fluorescence live cell microscopy is central to the analysis of inter- and intracellular signaling. However, analysis of highly dynamic, local processes requires high temporal and spatial resolution imaging, which is intrinsically linked to a low signal-to-noise ratio. To improve image quality after data acquisition, computational techniques, referred to as deconvolution, are being developed. Here, we discuss recent approaches in the areas of variational and deep learning image deconvolution.