Search for: All records

Artificial intelligence has recently been widely used in computational imaging. The deep neural network (DNN) improves the signal-to-noise ratio of the retrieved images, whose quality is otherwise corrupted due to the low sampling ratio or noisy environments. This work proposes a new computational imaging scheme based on the sequence transduction mechanism with the transformer network. The simulation database assists the network in achieving signal translation ability. The experimental single-pixel detector’s signal will be ‘translated’ into a 2D image in an end-to-end manner. High-quality images with no background noise can be retrieved at a sampling ratio as low as 2%. The illumination patterns can be either well-designed speckle patterns for sub-Nyquist imaging or random speckle patterns. Moreover, our method is robust to noise interference. This translation mechanism opens a new direction for DNN-assisted ghost imaging and can be used in various computational imaging scenarios.

Abstract In a viral pandemic, a few important tests are required for successful containment of the virus and reduction in severity of the infection. Among those tests, a test for the neutralizing ability of an antibody is crucial for assessment of population immunity gained through vaccination, and to test therapeutic value of antibodies made to counter the infections. Here, we report a sensitive technique to detect the relative neutralizing strength of various antibodies against the SARS-CoV-2 virus. We used bright, photostable, background-free, fluorescent upconversion nanoparticles conjugated with SARS-CoV-2 receptor binding domain as a phantom virion. A glass bottom plate coated with angiotensin-converting enzyme 2 (ACE-2) protein imitates the target cells. When no neutralizing IgG antibody was present in the sample, the particles would bind to the ACE-2 with high affinity. In contrast, a neutralizing antibody can prevent particle attachment to the ACE-2-coated substrate. A prototype system consisting of a custom-made confocal microscope was used to quantify particle attachment to the substrate. The sensitivity of this assay can reach 4.0 ng/ml and the dynamic range is from 1.0 ng/ml to 3.2  $$\upmu$$ μ g/ml. This is to be compared to 19 ng/ml sensitivity of commercially available kits.

Quantum entanglement has emerged as a great resource for spectroscopy and its importance in two-photon spectrum and microscopy has been demonstrated. Current studies focus on the two-photon absorption, whereas the Raman spectroscopy with quantum entanglement still remains elusive, with outstanding issues of temporal and spectral resolutions. Here we study the new capabilities provided by entangled photons in coherent Raman spectroscopy. An ultrafast frequency-resolved Raman spectroscopy with entangled photons is developed for condensed-phase molecules, to probe the electronic and vibrational coherences. Using quantum correlation between the photons, the signal shows the capability of both temporal and spectral resolutions not accessible by either classical pulses or the fields without entanglement. We develop a microscopic theory for this Raman spectroscopy, revealing the electronic coherence dynamics even at timescale of 50fs. This suggests new paradigms of optical signals and spectroscopy, with potential to push detection below standard quantum limit.

Optical imaging through scattering media has long been a challenge. Many approaches have been developed for focusing light or imaging objects through scattering media, but usually, they are either invasive, limited to stationary or slow-moving media, or require high-resolution cameras and complex algorithms to retrieve the images. By utilizing spatial–temporal encoded patterns (STEPs), we introduce a technique for the computation of imaging that overcomes these restrictions. With a single-pixel photodetector, we demonstrate non-invasive imaging through scattering media. This technique is insensitive to the motion of the media. Furthermore, we demonstrate that our image reconstruction algorithm is much more efficient than correlation-based algorithms for single-pixel imaging, which may allow fast imaging for applications with limited computing resources.