A5048675456- Neural Networks and Reservoir Computing
- Advanced Optical Imaging Technologies
- Optical Coherence Tomography Applications
- Random lasers and scattering media
- Advanced X-ray Imaging Techniques
- Digital Holography and Microscopy
- Cell Image Analysis Techniques
- Photonic and Optical Devices
- Optical Network Technologies
- Image Processing Techniques and Applications
- Advanced Fluorescence Microscopy Techniques
- Optical Wireless Communication Technologies
- Optical and Acousto-Optic Technologies
- Bacterial Identification and Susceptibility Testing
- Advanced Optical Sensing Technologies
- Quantum optics and atomic interactions
- X-ray Diffraction in Crystallography
- Optical measurement and interference techniques
- Microfluidic and Bio-sensing Technologies
- Nuclear Physics and Applications
- Advanced Electron Microscopy Techniques and Applications
- Laser-Matter Interactions and Applications
- X-ray Spectroscopy and Fluorescence Analysis
- Electron and X-Ray Spectroscopy Techniques
- Phase-change materials and chalcogenides
University of California, Los Angeles
2020-2025
California NanoSystems Institute
2021-2025
Samueli Institute
2023
Aselsan (Turkey)
2017-2019
Middle East Technical University
2018-2019
Abstract Image denoising, one of the essential inverse problems, targets to remove noise/artifacts from input images. In general, digital image denoising algorithms, executed on computers, present latency due several iterations implemented in, e.g., graphics processing units (GPUs). While deep learning-enabled methods can operate non-iteratively, they also introduce and impose a significant computational burden, leading increased power consumption. Here, we an analog diffractive denoiser...
High-resolution synthesis/projection of images over a large field-of-view (FOV) is hindered by the restricted space-bandwidth-product (SBP) wavefront modulators. We report deep learning-enabled diffractive display design that based on jointly-trained pair an electronic encoder and optical decoder to synthesize/project super-resolved using low-resolution The digital encoder, composed trained convolutional neural network (CNN), rapidly pre-processes high-resolution interest so their spatial...
Free-space optical communication becomes challenging when an occlusion blocks the light path. Here, we demonstrate a direct scheme, passing information around fully opaque, arbitrarily shaped that partially or entirely occludes transmitter's field-of-view. In this electronic neural network encoder and passive, all-optical diffractive network-based decoder are jointly trained using deep learning to transfer of interest opaque arbitrary shape. Following its training, encoder-decoder pair can...
Abstract Phase imaging is widely used in biomedical imaging, sensing, and material characterization, among other fields. However, direct of phase objects with subwavelength resolution remains a challenge. Here, we demonstrate amplitude based on all-optical diffractive encoding decoding. To resolve features an object, the imager uses thin, high-index solid-immersion layer to transmit high-frequency information object spatially-optimized encoder, which converts/encodes input into low-frequency...
Classical phase retrieval problem is the recovery of a constrained image from magnitude its Fourier transform. Although there are several well-known algorithms including hybrid input-output (HIO) method, reconstruction performance generally sensitive to initialization and measurement noise. Recently, deep neural networks (DNNs) have been shown provide state-of-the-art in solving inverse problems such as denoising, deconvolution, superresolution. In this work, we develop algorithm that...
Environmental factors such as temperature, nutrients, and pollutants affect the growth rates physical characteristics of microalgae populations. As algae play a vital role in marine ecosystems, monitoring is important to observe state an ecosystem. However, analyzing these populations using conventional light microscopy time-consuming requires experts both identify count algal cells, which turn considerably limits volume samples that can be measured each experiment. In this work we use...
Free-space optical information transfer through diffusive media is critical in many applications, such as biomedical devices and communication, but remains challenging due to random, unknown perturbations the path. We demonstrate an diffractive decoder with electronic encoding accurately of interest, corresponding to, e.g., any arbitrary input object or message, random phase diffusers along This hybrid electronic-optical model, trained using supervised learning, comprises a convolutional...
Gram staining has been a frequently used protocol in microbiology. It is vulnerable to artifacts due to, e.g., operator errors and chemical variations. Here, we introduce virtual of label-free bacteria using trained neural network that digitally transforms dark-field images unstained into their Gram-stained equivalents matching bright-field image contrast. After one-time training, the model processes an axial stack microscopy (never seen before) rapidly generate staining, bypassing several...
Wide-field interferometric microscopy is a highly sensitive, label-free, and low-cost biosensing imaging technique capable of visualizing individual biological nanoparticles such as viral pathogens exosomes. However, further resolution enhancement necessary to increase detection classification accuracy subdiffraction-limited nanoparticles. In this study, we propose deep-learning approach, based on coupled deep autoencoders, improve images L-shaped nanostructures. During training, our method...
Phase imaging is widely used in biomedical imaging, sensing, and material characterization, among other fields. However, direct of phase objects with subwavelength resolution remains a challenge. Here, we demonstrate amplitude based on all-optical diffractive encoding decoding. To resolve features an object, the imager uses thin, high-index solid-immersion layer to transmit high-frequency information object spatially-optimized encoder, which converts/encodes input into low-frequency spatial...
We report an optical diffractive decoder with electronic encoder network to facilitate the accurate transmission of information interest through unknown random phase diffusers along path. This hybrid electronic-optical model was trained via supervised learning, and comprises a convolutional neural network-based jointly-trained passive layers. After their joint-training using deep our can accurately transfer even in presence diffusers, generalizing new never seen before. experimentally...
We directly transfer optical information around arbitrarily-shaped, fully-opaque occlusions that partially or entirely block the line-of-sight between transmitter and receiver apertures. An electronic neural network (encoder) produces an encoded phase representation of to be transmitted. Despite being obstructed by opaque occlusion, this phase-encoded wave is decoded a diffractive at receiver. experimentally validated our framework in terahertz spectrum communicating images different using...
The integration of deep learning techniques with biophotonic setups has opened new horizons in bioimaging. A compelling trend this field involves deliberately compromising certain measurement metrics to engineer better bioimaging tools terms cost, speed, and form-factor, followed by compensating for the resulting defects through utilization models trained on a large amount ideal, superior or alternative data. This strategic approach found increasing popularity due its potential enhance...
Abstract Large‐scale and high‐dimensional permutation operations are important for various applications in, example, telecommunications encryption. Here, all‐optical diffractive computing is used to execute a set of between an input output field‐of‐view through layer rotations in optical network. In this reconfigurable multiplexed design , every has four orientations: . Each unique combination these layers represents distinct rotation state, tailored specific operation. Therefore, K ‐layer...
We report an image denoising analog processor composed of passive diffractive layers engineered through deep learning to filter out various types noise from input images, instantly projecting denoised images at the output field-of-view.
Gram staining has been one of the most frequently used protocols in microbiology for over a century, utilized across various fields, including diagnostics, food safety, and environmental monitoring. Its manual procedures make it vulnerable to errors artifacts due to, e.g., operator inexperience chemical variations. Here, we introduce virtual label-free bacteria using trained deep neural network that digitally transforms darkfield images unstained into their Gram-stained equivalents matching...
Unidirectional imagers form images of input objects only in one direction, e.g., from field-of-view (FOV) A to FOV B, while blocking the image formation reverse B A. Here, we report unidirectional imaging under spatially partially coherent light and demonstrate high-quality forward direction (A->B) with high power efficiency distorting backward (B->A) along low efficiency. Our reciprocal design features a set engineered linear diffractive layers that are statistically optimized for...