High dynamical range ptychography by applying Fresnel-based unsupervised neural network for arbitrary pixel size (FUNNAPS) propagation

Ptychography is a lensless phase retrieval technique widely used in material science, bioimaging, and semiconductor inspection. However, the practical resolution of ptychography is limited by two factors: the reconstructed pixel size determined by the experimental setup, and the constrained effective numerical aperture (NA) due to the confined well depth of the detector. To address these issues, we propose high dynamical range ptychography using a Fresnel-based unsupervised neural network for arbitrary pixel size (FUNNAPS) propagation. This new approach enables high effective NA with pixel size adjustment, demonstrating reconstruction close to the resolution limit.