Deconvolution microscopy has been extensively used to improve the resolution\nof the widefield fluorescent microscopy. Conventional approaches, which usually\nrequire the point spread function (PSF) measurement or blind estimation, are\nhowever computationally expensive. Recently, CNN based approaches have been\nexplored as a fast and high performance alternative. In this paper, we present\na novel unsupervised deep neural network for blind deconvolution based on cycle\nconsistency and PSF modeling layers. In contrast to the recent CNN approaches\nfor similar problem, the explicit PSF modeling layers improve the robustness of\nthe algorithm. Experimental results confirm the efficacy of the algorithm.\n