Isotropic quantitative differential phase contrast microscopy using deep neural networks
Journal
Proceedings of SPIE - The International Society for Optical Engineering
Journal Volume
11521
Pages
1152110
Date Issued
2020
Author(s)
Abstract
Isotropic quantitative differential phase contrast (iDPC) microscopy based on pupil engineering has made significant improvement in reconstructing phase image of weak phase objects. To further enhance acquisition speed for phase recovery in iDPC, we adapt deep neural networks to achieve isotropic phase retrieval from half-pupil based quantitative differential phase contrast (qDPC) microscopy. We proposed to utilize U-net model for transforming phase distribution from 2-axis reconstruction to 6-axis one. The results show that deep neural network we proposed works as well as we expected. The final loss value of our model after 500 epochs of training can achieve about 5.7e-5 after normalized.
Publisher
SPIE
Type
conference paper