Holographic Image Reconstruction with Phase Recovery and Autofocusing Using Recurrent Neural Networks

Digital holography is one of the most widely used label-free microscopy\ntechniques in biomedical imaging. Recovery of the missing phase information of\na hologram is an important step in holographic image reconstruction. Here we\ndemonstrate a convolutional recurrent neural network (RNN) based phase recovery\napproach that uses multiple holograms, captured at different sample-to-sensor\ndistances to rapidly reconstruct the phase and amplitude information of a\nsample, while also performing autofocusing through the same network. We\ndemonstrated the success of this deep learning-enabled holography method by\nimaging microscopic features of human tissue samples and Papanicolaou (Pap)\nsmears. These results constitute the first demonstration of the use of\nrecurrent neural networks for holographic imaging and phase recovery, and\ncompared with existing methods, the presented approach improves the\nreconstructed image quality, while also increasing the depth-of-field and\ninference speed.\n