Deep learning-assisted wavefront correction with sparse data for holographic tomography

•The proposed method is capable of accurately correcting the aberration wavefront which carries the sample information.•A sparse data fitting algorithm is used to obtain the Zernike coefficients of the whole scanning range with the collected sparse data for the same correction performance as that of the original plain algorithm but at a much faster speed.•This study shows that the whole wavefront with sparse scanning can be predicted and corrected with the assistance of deep learning, and this method will have numerous applications in holographic tomography.