Single-shot higher-order transport-of-intensity quantitative phase imaging using deep learning

Area of Science:

• Optics and Photonics
• Biomedical Imaging
• Computational Imaging

Background:

• Single-shot higher-order transport-of-intensity quantitative phase imaging (SHOT-QPI) offers a simple, scanless approach to phase imaging.
• Traditional SHOT-QPI methods suffer from low light-use efficiency and phase measurement inaccuracies due to hologram quantization errors.

Purpose of the Study:

• To develop an advanced SHOT-QPI method, termed Deep-SHOT, that leverages deep learning to improve phase measurement accuracy.
• To address the nonlinear relationship between distorted intensity distributions and phase in SHOT-QPI systems.

Main Methods:

• Implementation of a neural network within the Deep-SHOT framework to learn the mapping from distorted intensity profiles to ground truth phase distributions.
• Optimization of the neural network for specific optical system characteristics to correct for intrinsic intensity distortions.
• Utilizing multiple defocused intensity measurements to further enhance phase accuracy.

Main Results:

• Deep-SHOT demonstrates improved accuracy in phase measurements compared to conventional SHOT-QPI methods.
• The neural network effectively compensates for intensity distortions inherent in the optical system.
• Experimental validation confirms the efficacy of Deep-SHOT, even when dealing with nonlinearities.

Conclusions:

• Deep-SHOT represents a significant advancement in SHOT-QPI, offering higher accuracy and robustness.
• The integration of deep learning provides a powerful solution for overcoming limitations in quantitative phase imaging.
• This approach holds promise for various applications requiring precise phase measurements from biological or material samples.