Browse
Submit Data
Databases
API
Help

Dataset Information

32 Views

0 Connections

0 Citations

0 Reanalyses

0 Downloads

Omics score: 0

Far-field super-resolution ghost imaging with a deep neural network constraint.

ABSTRACT: Ghost imaging (GI) facilitates image acquisition under low-light conditions by single-pixel measurements and thus has great potential in applications in various fields ranging from biomedical imaging to remote sensing. However, GI usually requires a large amount of single-pixel samplings in order to reconstruct a high-resolution image, imposing a practical limit for its applications. Here we propose a far-field super-resolution GI technique that incorporates the physical model for GI image formation into a deep neural network. The resulting hybrid neural network does not need to pre-train on any dataset, and allows the reconstruction of a far-field image with the resolution beyond the diffraction limit. Furthermore, the physical model imposes a constraint to the network output, making it effectively interpretable. We experimentally demonstrate the proposed GI technique by imaging a flying drone, and show that it outperforms some other widespread GI techniques in terms of both spatial resolution and sampling ratio. We believe that this study provides a new framework for GI, and paves a way for its practical applications.

SUBMITTER: Wang F

PROVIDER: S-EPMC8720314 | biostudies-literature |

REPOSITORIES: biostudies-literature

ACCESS DATA

Json Xml

Similar Datasets

High-resolution far-field ghost imaging via sparsity constraint.

Project description:Ghost imaging (GI) is a method to nonlocally image an object with a single-pixel detector. However, the speckle's transverse size at the object plane limits the system's imaging resolution for conventional GI linear reconstruction algorithm. By combining the sparsity constraint of imaging object with ghost imaging method, we demonstrate experimentally that ghost imaging via sparsity constraint (GISC) can dramatically enhance the imaging resolution even using the random measurements far below the Nyquist limit. The image reconstruction algorithm of GISC is based on compressive sensing. Factors affecting the reconstruction quality of high-resolution GISC, such as the receiving system's numerical aperture and the object's sparse representation basis, are also investigated experimentally. This high-resolution imaging technique will have great applications in the microscopy and remote-sensing areas.

| S-EPMC4365410 | biostudies-other

Far-field super-resolution imaging of resonant multiples.

Project description:We demonstrate for the first time that seismic resonant multiples, usually considered as noise, can be used for super-resolution imaging in the far-field region of sources and receivers. Tests with both synthetic data and field data show that resonant multiples can image reflector boundaries with resolutions more than twice the classical resolution limit. Resolution increases with the order of the resonant multiples. This procedure has important applications in earthquake and exploration seismology, radar, sonar, LIDAR (light detection and ranging), and ultrasound imaging, where the multiples can be used to make high-resolution images.

| S-EPMC4928974 | biostudies-literature

Novel approach for label free super-resolution imaging in far field.

Project description:Progress in the emerging areas of science and technology, such as bio- and nano-technologies, depends on development of corresponding techniques for imaging and probing the structures with high resolution. Recently, the far field diffraction resolution limit in the optical range has been circumvented and different methods of super-resolution optical microscopy have been developed. The importance of this breakthrough achievement has been recognized by Nobel Prize for Chemistry in 2014. However, the fluorescence based super-resolution techniques only function with fluorescent molecules (most of which are toxic and can destroy or lead to artificial results in living biological objects) and suffer from photobleaching. Here we show a new way to break the diffraction resolution limit, which is based on nano-sensitivity to internal structure. Instead of conventional image formation as 2D intensity distribution, in our approach images are formed as a result of comparison of the axial spatial frequency profiles, reconstructed for each image point. The proposed approach dramatically increases the lateral resolution even in presence of noise and allows objects to be imaged in their natural state, without any labels.

| S-EPMC4558609 | biostudies-literature

Three-dimensional ghost imaging lidar via sparsity constraint.

Project description:Three-dimensional (3D) remote imaging attracts increasing attentions in capturing a target's characteristics. Although great progress for 3D remote imaging has been made with methods such as scanning imaging lidar and pulsed floodlight-illumination imaging lidar, either the detection range or application mode are limited by present methods. Ghost imaging via sparsity constraint (GISC), enables the reconstruction of a two-dimensional N-pixel image from much fewer than N measurements. By GISC technique and the depth information of targets captured with time-resolved measurements, we report a 3D GISC lidar system and experimentally show that a 3D scene at about 1.0 km range can be stably reconstructed with global measurements even below the Nyquist limit. Compared with existing 3D optical imaging methods, 3D GISC has the capability of both high efficiency in information extraction and high sensitivity in detection. This approach can be generalized in nonvisible wavebands and applied to other 3D imaging areas.

| S-EPMC4868975 | biostudies-literature

Deep learning-based point-scanning super-resolution imaging.

Project description:Point-scanning imaging systems are among the most widely used tools for high-resolution cellular and tissue imaging, benefiting from arbitrarily defined pixel sizes. The resolution, speed, sample preservation and signal-to-noise ratio (SNR) of point-scanning systems are difficult to optimize simultaneously. We show these limitations can be mitigated via the use of deep learning-based supersampling of undersampled images acquired on a point-scanning system, which we term point-scanning super-resolution (PSSR) imaging. We designed a 'crappifier' that computationally degrades high SNR, high-pixel resolution ground truth images to simulate low SNR, low-resolution counterparts for training PSSR models that can restore real-world undersampled images. For high spatiotemporal resolution fluorescence time-lapse data, we developed a 'multi-frame' PSSR approach that uses information in adjacent frames to improve model predictions. PSSR facilitates point-scanning image acquisition with otherwise unattainable resolution, speed and sensitivity. All the training data, models and code for PSSR are publicly available at 3DEM.org.

| S-EPMC8035334 | biostudies-literature

Mouse brain MR super-resolution using a deep learning network trained with optical imaging data.

Project description:IntroductionThe resolution of magnetic resonance imaging is often limited at the millimeter level due to its inherent signal-to-noise disadvantage compared to other imaging modalities. Super-resolution (SR) of MRI data aims to enhance its resolution and diagnostic value. While deep learning-based SR has shown potential, its applications in MRI remain limited, especially for preclinical MRI, where large high-resolution MRI datasets for training are often lacking.MethodsIn this study, we first used high-resolution mouse brain auto-fluorescence (AF) data acquired using serial two-photon tomography (STPT) to examine the performance of deep learning-based SR for mouse brain images.ResultsWe found that the best SR performance was obtained when the resolutions of training and target data were matched. We then applied the network trained using AF data to MRI data of the mouse brain, and found that the performance of the SR network depended on the tissue contrast presented in the MRI data. Using transfer learning and a limited set of high-resolution mouse brain MRI data, we were able to fine-tune the initial network trained using AF to enhance the resolution of MRI data.DiscussionOur results suggest that deep learning SR networks trained using high-resolution data of a different modality can be applied to MRI data after transfer learning.

| S-EPMC10365285 | biostudies-literature

Far-field Imaging of Non-fluorescent Species with Sub-diffraction Resolution.

Project description:Super-resolution optical microscopy is opening a new window to unveil the unseen details on the nanoscopic scale. Current far-field super-resolution techniques rely on fluorescence as the read-out1-5. Here, we demonstrate a scheme for breaking the diffraction limit in far-field imaging of non-fluorescent species by using spatially controlled saturation of electronic absorption. Our method is based on a pump-probe process where a modulated pump field perturbs the charge-carrier density in a sample, thus modulating the transmission of a probe field. A doughnut shape laser beam is then added to transiently saturate the electronic transition in the periphery of the focal volume, thus the induced modulation in the sequential probe pulse only occurs at the focal center. By raster scanning the three collinearly aligned beams, high-speed sub-diffraction-limited imaging of graphite nano-platelets was performed. This technique potentially enables super-resolution imaging of nano-materials and non-fluorescent chromophores, which may remain out of reach for fluorescence-based methods.

| S-EPMC3891596 | biostudies-literature

Wide-Field Super-Resolution Optical Fluctuation Imaging through Dynamic Near-Field Speckle Illumination.

Project description:Stochastic optical fluctuation imaging (SOFI) generates super-resolution fluorescence images by emphasizing the positions of fluorescent emitters via statistical analysis of their on-and-off blinking dynamics. In SOFI with speckle illumination (S-SOFI), the diffraction-limited grain size of the far-field speckles prevents independent blinking of closely located emitters, becoming a hurdle to realize the full super-resolution granted by SOFI processing. Here, we present a surface-sensitive super-resolution technique exploiting dynamic near-field speckle illumination to bring forth the full super-resolving power of SOFI without blinking fluorophores. With our near-field S-SOFI technique, up to 2.8- and 2.3-fold enhancements in lateral spatial resolution are demonstrated with computational and experimental fluorescent test targets labeled with conventional fluorophores, respectively. Fluorescent beads separated by 175 nm are also super-resolved by near-field speckles of 150 nm grain size, promising sub-100 nm resolution with speckle patterns of much smaller grain size.

| S-EPMC8949730 | biostudies-literature

An optical super-microscope for far-field, real-time imaging beyond the diffraction limit.

Project description:Optical microscopy suffers from a fundamental resolution limitation arising from the diffractive nature of light. While current solutions to sub-diffraction optical microscopy involve combinations of near-field, non-linear and fine scanning operations, we hereby propose and demonstrate the optical super-microscope (OSM) - a superoscillation-based linear imaging system with far-field working and observation distances - which can image an object in real-time and with sub-diffraction resolution. With our proof-of-principle prototype we report a point spread function with a spot size clearly reduced from the diffraction limit, and demonstrate corresponding improvements in two-point resolution experiments. Harnessing a new understanding of superoscillations, based on antenna array theory, our OSM achieves far-field, sub-diffraction optical imaging of an object without the need for fine scanning, data post-processing or object pre-treatment. Hence the OSM can be used in a wide variety of imaging applications beyond the diffraction limit, including real-time imaging of moving objects.

| S-EPMC3634104 | biostudies-literature

Ghost Imaging Based on Deep Learning.

Project description:Even though ghost imaging (GI), an unconventional imaging method, has received increased attention by researchers during the last decades, imaging speed is still not satisfactory. Once the data-acquisition method and the system parameters are determined, only the processing method has the potential to accelerate image-processing significantly. However, both the basic correlation method and the compressed sensing algorithm, which are often used for ghost imaging, have their own problems. To overcome these challenges, a novel deep learning ghost imaging method is proposed in this paper. We modified the convolutional neural network that is commonly used in deep learning to fit the characteristics of ghost imaging. This modified network can be referred to as ghost imaging convolutional neural network. Our simulations and experiments confirm that, using this new method, a target image can be obtained faster and more accurate at low sampling rate compared with conventional GI method.

| S-EPMC5915577 | biostudies-literature

OmicsDI is part of the ELIXIR infrastructure

OmicsDI is an Elixir interoperability service. Learn more ›

Tweets

OmicsDI Databases

PRIDE
PeptideAtlas
MassIVE
JPOST Repository
Physiome Model Repository

EGA
EVA
ENA
LINCS
PAXDB
Cell Collective

MetaboLights
Metabolomics Workbench
MetabolomeExpress
GNPS
BioModels
FAIRDOMHub

ArrayExpress
dbGaP
ExpressionAtlas
GEO
NODE

Information

Databases
Help
API
Contact us
Code on GitHub
Terms of use
Submit Data