Abstract: The trade-off between resolution and the number of measurements, along with the time-consuming and memory-intensive demands of traditional Fourier ptychography (FP) systems, has long posed challenges for the efficient and widespread application of this highly scalable quantitative phase imaging (QPI) technique. While recent approaches aim to improve imaging speed and reduce data sampling, they often struggle to maintain a high resolution. This is particularly true for images of thick samples. Consequently, these methods typically fail to simultaneously achieve reduced data sampling and high-resolution measurements. In this study, a resolution and efficiency enhancement scheme is introduced that integrates Fourier light-field microscopy (FLFM), FP, and multi-slice neural network (MSNN) to overcome these limitations. This system significantly reduces the amount of image data required for rapid three-dimensional (3D) imaging while preserving lateral measurement accuracy by replacing traditional microscopes with FLFM and incorporating encoded multiplexed illumination and aperture synthesis from FP. The refocusing capability of FLFM combined with an MSNN improves the vertical resolution, enabling more effective imaging of thick samples. This FLFM-based FP imaging system achieves a six-fold increase in data acquisition speed and a two-fold resolution improvement in both the lateral and longitudinal directions, compared to those of traditional FP systems. These improvements enable faster and more precise acquisition of structural information, particularly for thick samples. This study offers a practical and cost-effective solution to the limitations of traditional FP systems by providing a superior 3D measurement approach.

The full text can be downloaded at https://doi.org/10.1007/s40436-025-00586-5

Key words: Resolution enhancement, Fourier light-field microscopy (FLFM), Fourier ptychography (FP), Coded multiplexed illumination, Multi-slice neural network (MSNN)