High space-bandwidth in quantitative phase imaging using partially spatially coherent digital holographic microscopy and a deep neural network

Abstract

Quantitative phase microscopy (QPM) is a label-free technique that enables monitoring of morphological changes at the subcellular level. The performance of the QPM system in terms of spatial sensitivity and resolution depends on the coherence properties of the light source and the numerical aperture (NA) of objective lenses. Here, we propose high space-bandwidth quantitative phase imaging using partially spatially coherent digital holographic microscopy (PSC-DHM) assisted with a deep neural network. The PSC source synthesized to improve the spatial sensitivity of the reconstructed phase map from the interferometric images. Further, compatible generative adversarial network (GAN) is used and trained with paired low-resolution (LR) and high-resolution (HR) datasets acquired from the PSC-DHM system. The training of the network is performed on two different types of samples, i.e. mostly homogenous human red blood cells (RBC), and on highly heterogeneous macrophages. The performance is evaluated by predicting the HR images from the datasets captured with a low NA lens and compared with the actual HR phase images. An improvement of 9× in the space-bandwidth product is demonstrated for both RBC and macrophages datasets. We believe that the PSC-DHM + GAN approach would be applicable in single-shot label free tissue imaging, disease classification and other high-resolution tomography applications by utilizing the longitudinal spatial coherence properties of the light source.

High‐resolution single‐shot phase‐shifting interference microscopy using deep neural network for quantitative phase imaging of biological samples

“…There is a slight mismatch in SSIM between the phase map calculated from experimentally captured and network predicted interferograms. The mismatch in SSIM may occur due to phase-related artifacts during the data acquisition [8] such as spatial phase sensitivity, temporal phase sensitivity, and mismatch of equal phase-shift between the data frames.…”

“…The experimental and computational advancement in QPI is being widely adopted for extracting quantitative information of various industrial and biological specimens such as an optical waveguide, stem cells, human red blood cells (RBC), tissue sections, sperm samples, and among others [2][3][4][5]. In the past few decades, various newly developed QPI techniques have been implemented to improve the spatial resolution, spacebandwidth, temporal phase sensitivity, acquisition rate, and spatial phase sensitivity of the system [1,[6][7][8]. In QPI system, the spatial phase sensitivity and data acquisition rate are inversely related to each other.…”

“…offer single-shot data acquisition and phase reconstruction of the specimens but suffers with poor spatial phase sensitivity due to high spatial and temporal coherence. Singleshot phase reconstruction is possible in laser-based QPI by acquiring a high fringe density interferogram and then applying the standard Fourier transform (FT) approach [5,8,9]. Since the implementation of the FT algorithm requires high fringe density to separate the DC and twin images for noise-free phase recovery, so sufficiently large tilt angle is needed between the reference and object beams which essentially makes it off-axis interferometry [5].…”

“…In the past few years, a significant amount of work has been done in the development of various PSI techniques [20,21]. In PSI, a pre-calibrated piezoelectric transducer (PZT) is used and sufficient time is required to record precise phase-shifted multiple interferograms [8]. Although PSI has the advantage of utilizing the full resolution of the system, requirement of multiple frames is the key obstacle in PSI for many applications such as live-cell imaging and measurement with dynamic samples [22].…”

Zero-order-removal off-axis digital holographic reconstruction using different beam-ratio holographic recordings

“…11 The spatial-frequency bandwidth can also be improved using compression codecs or convolution neural networks. 12,13 In contrast, the spectrum filtering bandwidth in off-axis digital holographic reconstruction is also limited because of the existence of the zero-order spectrum. So, another major way to enlarge filtering bandwidth is achieved by eliminating or suppressing the zeroorder spectrum.…”