Computational Adaptive Optics for Fluorescence Microscopy via Sparse Blind Deconvolution

Abstract Fluorescence microscopy is an indispensable tool for investigating cellular and tissue‐level biology, yet its performance is often limited by optical diffraction, aberrations, and noise, resulting in suboptimal imaging quality. Traditional adaptive optics (AO) methods typically rely on additional hardware, such as wavefront sensors, to measure and correct system aberrations, which can be both complex and costly. Here, a computational adaptive optics technique based on sparse blind deconvolution (CAO‐SBD) is introduced, which uses a single blurred image to estimate aberrations and perform image deblurring. By incorporating sparse priors of fluorescent specimens with Zernike polynomial‐based aberration modeling, CAO‐SBD allows for the simultaneous reconstruction of both the aberrated point spread function (PSF) and the sample information, eliminating the need for precise PSF calibration. This method outperforms traditional Richardson‐Lucy deconvolution by enhancing robustness to noise and stabilizing the deconvolution process through adaptive PSF correction. Experimental results on bovine pulmonary artery endothelial cells demonstrate that CAO‐SBD significantly enhances image resolution and contrast across both wide‐field and confocal fluorescence microscopic systems, positioning CAO‐SBD as a powerful tool for high‐resolution biological imaging with broad applications.