Fast phase retrieval in off-axis digital holographic microscopy through deep learning.

Area of Science:

• Optics and Photonics
• Computational Imaging
• Microscopy

Background:

• Traditional digital holographic imaging algorithms are time-consuming due to multiple iterations for focused image reconstruction.
• Phase retrieval in digital holography presents challenges, including focusing and phase compensation.

Purpose of the Study:

• To develop a novel, fast digital focus method for microscopic samples.
• To address the limitations of traditional algorithms in terms of speed and phase compensation.

Main Methods:

• Implementation of a U-type convolutional neural network (U-net) for phase recovery.
• Simulation of defocused microscopic images using generated datasets to train and validate the U-net.
• Integration of the U-net method into a real-time off-axis digital holographic microscope.

Main Results:

• The U-net successfully restores the original phase of microscopic samples with high accuracy.
• The proposed method effectively performs phase compensation simultaneously with focusing.
• Significant breakthroughs in imaging speed were achieved when applied to a real-time microscope.

Conclusions:

• The U-net based approach offers a substantial improvement in the speed of digital holographic microscopy.
• This method provides an efficient solution for fast digital focus and phase compensation in microscopic imaging.