Several adaptive optics techniques have been developed to tackle optical aberrations in recent years. However, the methods often rely on complex systems or iterative algorithms. Our study proposes an innovative approach through a single acquisition of the probed sample using a detector array. Our solution is gentle to the sample and applicable to any laser scanning microscope equipped with a detector array. We pre-process the raw dataset – a set of simultaneous images of the same specimen – with a phase correlation algorithm and we use the results as input of a pre-trained CNN, enabling real-time aberration estimation with high accuracy of the first 11 Zernike coefficients.