Phase recovery and holographic image reconstruction using deep learning in neural networks

By Yair Rivenson, Yibo Zhang, Harun Gunaydin, Da Teng, Aydogan Ozcan. Light: Science & Applications, 2017

955 citations recorded by the source corpus.

Phase recovery from intensity-only measurements forms the heart of coherent imaging techniques and holography. In this study, we demonstrate that a neural network can learn to perform phase recovery and holographic image reconstruction after appropriate training. This deep learning-based approach provides an entirely new framework to conduct holographic imaging by rapidly eliminating twin-image and self-interference-related spatial artifacts. This neural network-based method is fast to compute and reconstructs phase and amplitude images of the objects using only one hologram, requiring fewer measurements in addition to being computationally faster. We validated this method by reconstructing the phase and amplitude images of various samples, including blood and Pap smears and tissue sections. These results highlight that challenging problems in imaging science can be overcome through machine learning, providing new avenues to design powerful computational imaging systems. A new method that uses neural-network-based deep learning could lead to faster and more accurate holographic image reconstruction and phase recovery. Optoelectronic sensors such as charge-coupled devices and complementary metal-oxide–semiconductor imagers are sensitive to intensity but are unable to directly detect the phase of light waves diffracted from an object. Additional information or measurement is thus needed to recover the missing phase information, which enables reconstructing the image of the sample. Now, Aydogan Ozcan and colleagues from the University of California, Los Angeles in the USA have designed a neural network that can perform phase recovery and holographic image reconstruction from a single intensity-only hologram. Using deep learning, they demonstrated the elimination of twin-image and self-interference-related spatial artifacts arising from missing phase information. The technique could significantly simplify the imaging hardware and speed up the image acquisition and reconstruction processes in various holographic and coherent imaging systems.

Subfields: Atomic and Molecular Physics, and Optics, Computer Vision and Pattern Recognition, Radiation

View Paper: https://www.semanticscholar.org/paper/d8df8e2c10fce631173aae694f017dd136226284

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Atomic and Molecular Physics, and Optics

In Computer Vision and Pattern Recognition

In Radiation

Citations in the corpus

On the use of deep learning for computational imaging (2020, methodology): “DNN reconstructions have been successfully demonstrated in, for example, holographic, 47,48 ptychographic, 49 and coherent modulation imaging (CMI).”
PhaseStain: the digital staining of label-free quantitative phase microscopy images using deep learning (2018, background): “…out-of-focus particles can be digitally removed by using a recently introduced deep learning-based hologram reconstruction method, which learns, through data, to attack or eliminate twin-image artifacts as well as the interference fringes resulting from out-of-focus or undesired objects 19,20 .”
Fringe pattern analysis using deep learning (2018, positive): “10,11 In this letter, we demonstrate experimentally for the first time, to our knowledge, that the use of a deep neural network can substantially enhance the accuracy of phase demodulation from a single fringe pattern.”
Deep Learning Techniques for Inverse Problems in Imaging (2020, background): “ing [45], holography [46], [47], ghost imaging [48], imaging through scattering media [49]–[51], and non-line-of-sight imaging [52], which is illustrated in Figure 5.”
Machine learning and applications in ultrafast photonics (2020, methodology): “In this latter work, the genetic algorithm was added to generate a wide range of nanoantenna designs to be fed into the neural network.”
Deep Learning Microscopy: Enhancing Resolution, Field-of-View and Depth-of-Field of Optical Microscopy Images Using Neural Networks (2017, methodology): “Recently, deep neural networks have also been used for optical phase recovery and holographic image reconstruction 14 .”
Neural holography with camera-in-the-loop training (2020, methodology): “For example, neural networks have been used in lensless holographic microscopy to help solve phase-retrieval problems [Rivenson et al. 2018; Sinha et al. 2017].”
Virtual histological staining of unlabelled tissue-autofluorescence images via deep learning (2018, methodology): “A more general CNN model can be learnt for multiple tissue/stain combinations by e.g., increasing the number of trained parameters in the model 22 , at the cost of a possible increase in the training and inference times.”