Phase Retrieval Based on Deep Learning in Grating Interferometer

Grating interferometry is a promising technique to obtain differential phase contrast images with illumination source of low intrinsic transverse coherence. However, retrieving the phase contrast image from the differential phase contrast image is difficult due to the accumulated noise and artifacts from the differential phase contrast image (DPCI) reconstruction. In this paper, we implemented a deep learning-based phase retrieval method to suppress these artifacts. Conventional deep learning based denoising requires noisy-clean image pair, but it is not feasible to obtain sufficient number of clean images for grating interferometry. In this paper, we apply a recently developed neural network called Noise2Noise (N2N) that uses noise-noise image pairs for training. We obtained many differential phase contrast images through combination of phase stepping images, and these were used as noise input/target pairs for N2N training. The application of the N2N network to simulated and measured DPCI showed that the phase contrast images were retrieved with strongly suppressed phase retrieval artifacts. These results can be used in grating interferometer applications which uses phase stepping method.

Isotropic quantitative differential phase contrast imaging techniques: a review

Optical phase shifts generated by the spatial variation of refractive index and thickness inside the transparent samples can be determined by intensity measurements through quantitative phase contrast imaging. In this review, we focus on isotropic quantitative differential phase-contrast microscopy(qDPC), which is a non-interferometric quantitative phase imaging technique and belongs to the class of deterministic phase retrieval from intensity. The qDPC is based on the principle of a weak object transfer function together with the first-order Born approximation in a partially coherent illumination system and wide-field detection, which offers multiple advantages. We review basic principles, imaging systems, and demonstrate examples of differential phase contrast (DPC) imaging for biomedical applications. In addition to the previous work, we present the latest results for isotropic phase contrast enhancements using a deep learning approach. We implemented a supervised learning approach with the U-Net model to reduce the number of measurements required for multi-axis measurements associated with the isotropic phase transfer function. We show that a well-designed and trained neural network provide a fast and efficient way to predict quantitative phase maps for live cells, which can help in determining morphological parameters. The prospects of deep learning in quantitative phase microscopy, particularly for isotropic quantitative phase estimation, are discussed.

Flexible STEM with Simultaneous Phase and Depth Contrast

Recent advances in scanning transmission electron microscopy (STEM) have rekindled interest in multi-channel detectors and prompted the exploration of unconventional scan patterns. These emerging needs are not yet addressed by standard commercial hardware. The system described here incorporates a flexible scan generator that enables exploration of low-acceleration scan patterns, while data are recorded by a scalable eight-channel array of nonmultiplexed analog-to-digital converters. System integration with SerialEM provides a flexible route for automated acquisition protocols including tomography. Using a solid-state quadrant detector with additional annular rings, we explore the generation and detection of various STEM contrast modes. Through-focus bright-field scans relate to phase contrast, similarly to wide-field TEM. More strikingly, comparing images acquired from different off-axis detector elements reveals lateral shifts dependent on defocus. Compensation of this parallax effect leads to decomposition of integrated differential phase contrast (iDPC) to separable contributions relating to projected electric potential and to defocus. Thus, a single scan provides both a computationally refocused phase contrast image and a second image in which the signed intensity, bright or dark, represents the degree of defocus.