Deep learning-based quantitative phase microscopy

Quantitative phase microscopy (QPM) is a powerful tool for label-free and noninvasive imaging of transparent specimens. In this paper, we propose a novel QPM approach that utilizes deep learning to reconstruct accurately the phase image of transparent specimens from a defocus bright-field image. A U-net based model is used to learn the mapping relation from the defocus intensity image to the phase distribution of a sample. Both the off-axis hologram and defocused bright-field image are recorded in pair for thousands of virtual samples generated by using a spatial light modulator. After the network is trained with the above data set, the network can fast and accurately reconstruct the phase information through a defocus bright-field intensity image. We envisage that this method will be widely applied in life science and industrial detection.

Automated summary

In this paper, a novel approach for quantitative phase microscopy (QPM) using deep learning is proposed to accurately reconstruct the phase image of transparent specimens from a defocus bright-field image. A U-net based model is trained to learn the mapping relation between the defocus intensity image and the phase distribution of the sample. After training with a large data set of off-axis holograms and defocused bright-field images, the network can quickly and accurately reconstruct the phase information from a defocus bright-field intensity image. This method is expected to have wide applications in life science and industrial detection.