Search

We present a method to generate multiple virtual stains on an image of label-free tissue using a single deep neural network, which is fed with the autofluorescence images of the unlabeled tissue alongside a user-defined digital-staining matrix. Users can indicate which stain to apply on each pixel by editing the digital-staining matrix and blend multiple virtual stains, creating entirely new stain combinations.

We report in vivo virtual histology of skin without a biopsy, where deep learning is used to virtually stain tissue and generate hematoxylin and eosin (H&E)-like microscopic images of skin using a reflectance confocal microscope.

We present deep learning-based virtual immunohistochemical (IHC) HER2 staining of label-free breast tissue sections, matching the standard IHC HER2 staining performed by histotechnologists. © 2022 The Author(s)

Diffractive deep neural networks composed of spatially-engineered transmissive layers were designed and experimental demonstrated to see through unknown, random phase diffusers, instantaneously reconstructing images of unknown objects without the need for a digital computer.

Deep learning has been revolutionizing information processing in many fields of science and engineering owing to the massively growing amounts of data and the advances in deep neural network architectures. As these neural networks are expanding their capabilities toward achieving state-of-the-art solutions for demanding statistical inference tasks in various applications, there appears to be a global need for low-power, scalable, and fast computing hardware beyond what existing electronic systems can offer. Optical computing might potentially address some of these needs with its inherent parallelism, power efficiency, and high speed. Recent advances in optical materials, fabrication, and optimization techniques have significantly enriched the design capabilities in optics and photonics, leading to various successful demonstrations of guided-wave and free-space computing hardware for accelerating machine learning tasks using light. In addition to statistical inference and computing, deep learning has also fundamentally affected the field of inverse optical/photonic design. The approximation power of deep neural networks has been utilized to develop optics/photonics systems with unique capabilities, all the way from nanoantenna design to end-to-end optimization of computational imaging and sensing systems. In this review, we attempt to provide a broad overview of the current state of this emerging symbiotic relationship between deep learning and optics/photonics.

We present and experimentally demonstrate a deep learning-driven machine-vision framework that trains diffractive surfaces to encode the spatial information objects into the output power spectrum for all-optical image classification using a single-pixel spectroscopic detector.

We report ensemble learning of diffractive optical neural networks that result in substantial improvements in the all-optical image classification accuracy of the resulting models.

We demonstrate diffractive optical networks that are trained with deep learning to engineer input terahertz pulses into desired temporal waveforms using passive diffractive surfaces that control the spectral phase and amplitude of the output pulse.

We report all-optical object classification systems that are based on class-specific design of diffractive neural networks followed by a differential detection scheme. The blind inference accuracies achieved through this framework are significantly enhanced.