This PDF file contains the front matter associated with SPIE Proceedings Volume 9720 including the Title Page, Copyright information, Table of Contents, Introduction, and Conference Committee listing.

Video recording of ultrafast phenomena using a detector array based on the CCD or CMOS technologies is fundamentally limited by the sensor’s on-chip storage and data transfer speed. To get around this problem, the most practical approach is to utilize a streak camera. However, the resultant image is normally one dimensional—only a line of the scene can be seen at a time. Acquiring a two-dimensional image thus requires mechanical scanning across the entire field of view. This requirement poses severe restrictions on the applicable scenes because the event itself must be repetitive. To overcome these limitations, we have developed a new computational ultrafast imaging method, referred to as compressed ultrafast photography (CUP), which can capture two-dimensional dynamic scenes at up to 100 billion frames per second. Based on the concept of compressed sensing, CUP works by encoding the input scene with a random binary pattern in the spatial domain, followed by shearing the resultant image in a streak camera with a fully-opened entrance slit. The image reconstruction is the solution of the inverse problem of above processes. Given sparsity in the spatiotemporal domain, the original event datacube can be reasonably estimated by employing a two-step iterative shrinkage/thresholding algorithm. To demonstrate CUP, we imaged light reflection, refraction, and racing in two different media (air and resin). Our technique, for the first time, enables video recording of photon propagation at a temporal resolution down to tens of picoseconds. Moreover, to further expand CUP’s functionality, we added a color separation unit to the system, thereby allowing simultaneous acquisition of a four-dimensional datacube (x,y,t,λ), where λ is wavelength, within a single camera snapshot.

The time resolution of charge modulation in CMOS image sensors has entered the sub-nano second regime and is still reducing toward tens of pico-second. The lateral electric field modulators (LEFM) invented at Shizuoka University has significantly contributed to the recent progress in the solid-state time-resolved imaging field. Based on the LEFM technology, we are developing ultra-high-speed CMOS image sensors whose frame rate or time resolution is determined only by the charge modulation speed. In this presentation, the concept, architecture, example of implementation, and demonstration of 200Mfps single-shot video capturing based on our scheme are shown.

Optical time-stretch imaging entails a stringent requirement of state-of-the-art high-speed data acquisition unit in order to preserve high image resolution at an ultrahigh frame rate — hampering the widespread application of such technology. We here propose a pixel super-resolution (pixel SR) technique tailored for time-stretch imaging that can relax the sampling rate requirement. It harnesses a concept of equivalent-time sampling, which effectively introduces sub-pixel shifts between frames. It involves no active opto-mechanical subpixel-shift control and any additional hardware. We present the system design rules and a proof-of-principle experiment which restores high-resolution images at a relaxed sampling rate of 5 GSa=s.

Optical frequency comb (OFC) has attracted attentions for optical frequency metrology in visible and infrared regions because the mode-resolved OFC spectrum can be used as a precise frequency ruler due to both characteristics of broadband radiation and narrow-line CW radiation. Furthermore, the absolute accuracy of all frequency modes in OFC is secured by phase-locking a repetition frequency f_rep and a carrier-envelope-offset frequency f_ceo to a frequency standard. However, application fields of OFC other than optical frequency metrology are still undeveloped. One interesting aspect of OFC except for the frequency ruler is optical carrier having a huge number of discrete frequency channels because OFC is composed of a series of frequency spikes regularly separated by f_rep in the broad spectral range. If a certain quantity to be measured is encoded on each comb mode by dimensional conversion, a huge number of data for the measured quantity can be obtained from a single mode-resolved spectrum of OFC. In this paper, we encode the confocal microscopic line-image of a sample on the mode-resolved OFC spectrum by the dimensional conversion between wavelength and 1D-space. The resulting image-encoded OFC spectrum is acquired by an optical spectrum analyzer or dual comb spectrometer. Finally, the line image of the sample is decoded from the spectral amplitude of the mode-resolved OFC spectrum. The combination of OFC with the dimensional conversion enables to establish both confocal modality and line-field imaging under the scan-less condition.

Digital holography is a technique of 3D measurement of object. The technique uses an image sensor to record the interference fringe image containing the complex amplitude of object, and numerically reconstructs the complex amplitude by computer. Parallel phase-shifting digital holography is capable of accurate 3D measurement of dynamic object. This is because this technique can reconstruct the complex amplitude of object, on which the undesired images are not superimposed, form a single hologram. The undesired images are the non-diffraction wave and the conjugate image which are associated with holography. In parallel phase-shifting digital holography, a hologram, whose phase of the reference wave is spatially and periodically shifted every other pixel, is recorded to obtain complex amplitude of object by single-shot exposure. The recorded hologram is decomposed into multiple holograms required for phase-shifting digital holography. The complex amplitude of the object is free from the undesired images is reconstructed from the multiple holograms. To validate parallel phase-shifting digital holography, a high-speed parallel phase-shifting digital holography system was constructed. The system consists of a Mach-Zehnder interferometer, a continuous-wave laser, and a high-speed polarization imaging camera. Phase motion picture of dynamic air flow sprayed from a nozzle was recorded at 180,000 frames per second (FPS) have been recorded by the system. Also phase motion picture of dynamic air induced by discharge between two electrodes has been recorded at 1,000,000 FPS, when high voltage was applied between the electrodes.

The single-shot compressed ultrafast photography (CUP) camera is the fastest receive-only camera in the world. In this work, we introduce an external CCD camera and a space- and intensity-constrained (SIC) reconstruction algorithm to improve the image quality of CUP. The CCD camera takes a time-unsheared image of the dynamic scene. Unlike the previously used unconstrained algorithm, the proposed algorithm incorporates both spatial and intensity constraints, based on the additional prior information provided by the external CCD camera. First, a spatial mask is extracted from the time-unsheared image to define the zone of action. Second, an intensity threshold constraint is determined based on the similarity between the temporally projected image of the reconstructed datacube and the time-unsheared image taken by the external CCD. Both simulation and experimental studies showed that the SIC reconstruction improves the spatial resolution, contrast, and general quality of the reconstructed image.

We propose single-shot multiwavelength digital holography using a monochromatic image sensor and dual reference arms. Multiple wavelength information is multiplexed on the monochromatic image sensor plane in the space domain and is separated in the spatial frequency domain by utilizing the difference between the spatial frequencies of interference fringes at respective wavelengths. The recordable spatial bandwidth that is utilized for object waves is extended by using dual reference arms in comparison with that using a single reference arm. Both the three-dimensional and three-wavelength information of an object were recorded and reconstructed without the crosstalk between object waves with multiple wavelengths.

Phase-space measurements enable characterization of second-order spatial coherence properties and can be used for digital aberration removal or 3D position reconstruction. Previous methods use a scanning aperture to measure the phase space spectrogram, which is slow and light inefficient, while also attenuating information about higher-order correlations. We demonstrate a significant improvement of speed and light throughput by incorporating multiplexing techniques into our phase-space imaging system. The scheme implements 2D coded aperture patterning in the Fourier (pupil) plane of a microscope using a Spatial Light Modulator (SLM), while capturing multiple intensity images in real space. We compare various multiplexing schemes to scanning apertures and show that our phase-space reconstructions are accurate for experimental data with biological samples containing many 3D fluorophores.

We demonstrate the use of phase-space (e.g. light field) imaging for 3D localization of multiple point sources inside scattering material. The effect of scattering is to spread angular (spatial frequency) information, which can be modeled by a multi-slice forward model in phase space. We propose a sparsity-constrained atomic norm reconstruction algorithm in order to further constrain the problem. By using 4D measurements for 3D reconstruction, the dimensionality mismatch provides significant robustness to multiple scattering effects, with either static or dynamic diffusers.

A crucial part of the drug discovery process involves imaging the response of thousands of cell cultures to candidate drugs. Quantitative parameters from these “high content screens”, such as protein expression and cell morphology, are extracted from fluorescence and brightfield micrographs. Due to the sheer number of cells that need to imaged for adequate statistics, the imaging time itself is a major bottleneck. Automated microscopes image small fields-of-view (FOVs) serially, which are then stitched together to form gigapixel-scale mosaics. We have developed a microscopy architecture that reduces mechanical overhead of traditional large field-of-view by parallelizing the image capture process. Instead of a single objective lens imaging FOVs one by one, we employ a microlens array for continuous photon capture, resulting in a 3-fold throughput increase. In this contribution, we present the design and imaging results of this microscopy architecture in three different contrast modes: multichannel fluorescence, hyperspectral fluorescence and brightfield.

Serendipter is a device that distinguishes and selects very rare particles and cells from huge amount of population. We are currently designing and constructing information processing system for a Serendipter. The information processing system for Serendipter is a kind of sensor-fusion system but with much more difficulties: To fulfill these requirements, we adopt All IP based architecture: All IP-Ethernet based data processing system consists of (1) sensor/detector directly output data as IP-Ethernet packet stream, (2) single Ethernet/TCP/IP streams by a L2 100Gbps Ethernet switch, (3) An FPGA board with 100Gbps Ethernet I/F connected to the switch and a Xeon based server. Circuits in the FPGA include 100Gbps Ethernet MAC, buffers and preprocessing, and real-time Deep learning circuits using multi-layer neural networks. Proposed All-IP architecture solves existing problem to construct large-scale sensor-fusion systems.

High-speed imaging is an indispensable technique, particularly for identifying or analyzing fast-moving objects. The serial time-encoded amplified microscopy (STEAM) technique was proposed to enable us to capture images with a frame rate 1,000 times faster than using conventional methods such as CCD (charge-coupled device) cameras. The application of this high-speed STEAM imaging technique to a real-time system, such as flow cytometry for a cell-sorting system, requires successively processing a large number of captured images with high throughput in real time. We are now developing a high-speed flow cytometer system including a STEAM camera. In this paper, we describe our approach to processing these large amounts of image data in real time. We use an analog-to-digital converter that has up to 7.0G samples/s and 8-bit resolution for capturing the output voltage signal that involves grayscale images from the STEAM camera. Therefore the direct data output from the STEAM camera generates 7.0G byte/s continuously. We provided a field-programmable gate array (FPGA) device as a digital signal pre-processor for image reconstruction and finding objects in a microfluidic channel with high data rates in real time. We also utilized graphics processing unit (GPU) devices for accelerating the calculation speed of identification of the reconstructed images. We built our prototype system, which including a STEAM camera, a FPGA device and a GPU device, and evaluated its performance in real-time identification of small particles (beads), as virtual biological cells, owing through a microfluidic channel.

We discuss the implementation of wavelength-swept coherent Raman scattering (CRS) microscopy for the rapid acquisition of hyperspectral datacubes. We highlight two multivariate analysis approaches for efficiently generating spectroscopic maps from the acquired data: principal component analysis (PCA), which is a popular method for extracting information from multidimensional datasets, and vertex component analysis (VCA), which has previously been successfully used for the analysis of spontaneous Raman microscopy data. Through several biomedical imaging examples, we discuss the advantages and disadvantages of these approaches for CRS microscopy.

Microbes, especially microalgae, have recently been of great interest for developing novel biofuels, drugs, and biomaterials. Imaging-based screening of live cells can provide high selectivity and is attractive for efficient bio-production from microalgae. Although conventional cellular screening techniques use cell labeling, labeling of microbes is still under development and can interfere with their cellular functions. Furthermore, since live microbes move and change their shapes rapidly, a high-speed imaging technique is required to suppress motion artifacts. Stimulated Raman scattering (SRS) microscopy allows for label-free and high-speed spectral imaging, which helps us visualize chemical components inside biological cells and tissues. Here we demonstrate high-speed SRS imaging, with temporal resolution of 0.14 seconds, of intracellular distributions of lipid, polysaccharide, and chlorophyll concentrations in rapidly moving Euglena gracilis, a unicellular phytoflagellate. Furthermore, we show that our method allows us to analyze the amount of chemical components inside each living cell. Our results indicate that SRS imaging may be applied to label-free screening of living microbes based on chemical information.

A CMOS image sensor using high-speed lock-in pixels for stimulated Raman scattering (SRS) spectroscopy is presented in this paper. The effective SRS signal from the stimulated emission of SRS mechanism is very small in contrast to the offset of a probing laser source, which is in the ratio of 10^-4 to 10^-5. In order to extract this signal, the common offset component is removed, and the small difference component is sampled using switched-capacitor integrator with a fully differential amplifier. The sampling is performed over many integration cycles to achieve appropriate amplification. The lock-in pixels utilizes high-speed lateral electric field charge modulator (LEFM) to demodulate the SRS signal which is modulated at high-frequency of 20MHz. A prototype chip is implemented using 0.11μm CMOS image sensor technology.

Over the last 20 years, optical coherence tomography (OCT) has become a valuable diagnostic tool in ophthalmology with several 10,000 devices sold today. Other applications, like intravascular OCT in cardiology and gastro-intestinal imaging will follow. OCT provides 3-dimensional image data with microscopic resolution of biological tissue in vivo. In most applications, off-line processing of the acquired OCT-data is sufficient. However, for OCT applications like OCT aided surgical microscopes, for functional OCT imaging of tissue after a stimulus, or for interactive endoscopy an OCT engine capable of acquiring, processing and displaying large and high quality 3D OCT data sets at video rate is highly desired. We developed such a prototype OCT engine and demonstrate live OCT with 25 volumes per second at a size of 320x320x320 pixels. The computer processing load of more than 1.5 TFLOPS was handled by a GTX 690 graphics processing unit with more than 3000 stream processors operating in parallel. In the talk, we will describe the optics and electronics hardware as well as the software of the system in detail and analyze current limitations. The talk also focuses on new OCT applications, where such a system improves diagnosis and monitoring of medical procedures. The additional acquisition of hyperspectral stimulated Raman signals with the system will be discussed.

Optical coherence tomography (OCT) is a non-invasive imaging technique that measures reflectance from within biological tissues. Current higher-NA optical coherence microscopy (OCM) technologies with near cellular resolution have limitations on volumetric imaging capabilities due to the trade-offs between resolution vs. depth-of-field and sensitivity to aberrations. Such trade-offs can be addressed using computational adaptive optics (CAO), which corrects aberration computationally for all depths based on the complex optical field measured by OCT. However, due to the large size of datasets plus the computational complexity of CAO and OCT algorithms, it is a challenge to achieve high-resolution 3D-OCM reconstructions at speeds suitable for clinical and research OCM imaging. In recent years, real-time OCT reconstruction incorporating both dispersion and defocus correction has been achieved through parallel computing on graphics processing units (GPUs). We add to these methods by implementing depth-dependent aberration correction for volumetric OCM using plane-by-plane phase deconvolution. Following both defocus and aberration correction, our reconstruction algorithm achieved depth-independent transverse resolution of 2.8 um, equal to the diffraction-limited focal plane resolution. We have translated the CAO algorithm to a CUDA code implementation and tested the speed of the software in real-time using two GPUs - NVIDIA Quadro K600 and Geforce TITAN Z. For a data volume containing 4096×256×256 voxels, our system’s processing speed can keep up with the 60 kHz acquisition rate of the line-scan camera, and takes 1.09 seconds to simultaneously update the CAO correction for 3 en face planes at user-selectable depths.

Light-sheet microscopy, also named selective-plane illumination microscopy, enables optical sectioning with minimal light delivered to the sample. Therefore, it allows one to gather volumetric datasets of developing embryos and other light-sensitive samples over extended times. We have configured a light-sheet microscope that, unlike most previous designs, can image samples in formats compatible with high-content imaging. Our microscope can be used with multi-well plates or with microfluidic devices. In designing our optical system to accommodate these types of sample holders we encounter large optical aberrations. We counter these aberrations with both static optical components in the imaging path and with adaptive optics. Potential applications of this microscope include studying the development of a large number of embryos in parallel and over long times with subcellular resolution and doing high-throughput screens on organisms or cells where volumetric data is necessary.

We developed isotropic multiview (IsoView) light-sheet microscopy in order to image fast cellular dynamics, such as cell movements in an entire developing embryo or neuronal activity throughput an entire brain or nervous system, with high resolution in all dimensions, high imaging speeds, good physical coverage and low photo-damage. To achieve high temporal resolution and high spatial resolution at the same time, IsoView microscopy rapidly images large specimens via simultaneous light-sheet illumination and fluorescence detection along four orthogonal directions. In a post-processing step, these four views are then combined by means of high-throughput multiview deconvolution to yield images with a system resolution of ≤ 450 nm in all three dimensions. Using IsoView microscopy, we performed whole-animal functional imaging of Drosophila embryos and larvae at a spatial resolution of 1.1-2.5 μm and at a temporal resolution of 2 Hz for up to 9 hours. We also performed whole-brain functional imaging in larval zebrafish and multicolor imaging of fast cellular dynamics across entire, gastrulating Drosophila embryos with isotropic, sub-cellular resolution. Compared with conventional (spatially anisotropic) light-sheet microscopy, IsoView microscopy improves spatial resolution at least sevenfold and decreases resolution anisotropy at least threefold. Compared with existing high-resolution light-sheet techniques, such as lattice lightsheet microscopy or diSPIM, IsoView microscopy effectively doubles the penetration depth and provides subsecond temporal resolution for specimens 400-fold larger than could previously be imaged.

Fluorescence imaging using radiofrequency-tagged emission (FIRE) is an emerging technique that enables higher imaging speed (namely, temporal resolution) in fluorescence microscopy compared to conventional fluorescence imaging techniques such as confocal microscopy and wide-field microscopy. It works based on the principle that it uses multiple intensity-modulated fields in an interferometric setup as excitation fields and applies frequency-division multiplexing to fluorescence signals. Unfortunately, despite its high potential, FIRE has limited imaging speed due to two practical limitations: signal bandwidth and signal detection efficiency. The signal bandwidth is limited by that of an acousto-optic deflector (AOD) employed in the setup, which is typically 100-200 MHz for the spectral range of fluorescence excitation (400-600 nm). The signal detection efficiency is limited by poor spatial mode-matching between two interfering fields to produce a modulated excitation field. Here we present a method to overcome these limitations and thus to achieve higher imaging speed than the prior version of FIRE. Our method achieves an increase in signal bandwidth by a factor of two and nearly optimal mode matching, which enables the imaging speed limited by the lifetime of the target fluorophore rather than the imaging system itself. The higher bandwidth and better signal detection efficiency work synergistically because higher bandwidth requires higher signal levels to avoid the contribution of shot noise and amplifier noise to the fluorescence signal. Due to its unprecedentedly high-speed performance, our method has a wide variety of applications in cancer detection, drug discovery, and regenerative medicine.

We present a light sheet confocal microscope (LSCM) capable of high-resolution imaging of cell suspensions in a microfluidic environment. In lieu of conventional pressure-driven flow or mechanical translation of the samples, we have employed a novel method of fluid transport, redox-magnetohydrodynamics (redox-MHD). This method achieves fluid motion by inducing a small current into the suspension in the presence of a magnetic field via electrodes patterned onto a silicon chip. This on-chip transportation requires no moving parts, and is coupled to the remainder of the imaging system. The microscopy system comprises a 450 nm diode 20 mW laser coupled to a single mode fiber and a cylindrical lens that converges the light sheet into the back aperture of a 10x, 0.3 NA objective lens in an epi-illumination configuration. The emission pathway contains a 150 mm tube lens that focuses the light onto the linear sensor at the conjugate image plane. The linear sensor (ELiiXA+ 8k/4k) has three lateral binning modes which enables variable detection aperture widths between 5, 10, or 20 μm, which can be used to vary axial resolution. We have demonstrated redox-MHD-enabled light sheet microscopy in suspension of fluorescent polystyrene beads. This approach has potential as a high-throughput image cytometer with myriad cellular diagnostic applications.

Infrared spectroscopy is a highly attractive read-out technology for compositional analysis of biomedical specimens because of its unique combination of high molecular sensitivity without the need for exogenous labels. Traditional techniques such as FTIR and Raman have suffered from comparatively low speed and sensitivity however recent innovations are challenging this situation. Direct mid-IR spectroscopy is being speeded up by innovations such as MEMS-based FTIR instruments with very high mirror speeds and supercontinuum sources producing very high sample irradiation levels. Here we explore another possible method – external cavity quantum cascade lasers (EC-QCL’s) with high cavity tuning speeds (mid-IR swept lasers). Swept lasers have been heavily developed in the near-infrared where they are used for non-destructive low-coherence imaging (OCT). We adapt these concepts in two ways. Firstly by combining mid-IR quantum cascade gain chips with external cavity designs adapted from OCT we achieve spectral acquisition rates approaching 1 kHz and demonstrate potential to reach 100 kHz. Secondly we show that mid-IR swept lasers share a fundamental sensitivity advantage with near-IR OCT swept lasers. This makes them potentially able to achieve the same spectral SNR as an FTIR instrument in a time x N shorter (N being the number of spectral points) under otherwise matched conditions. This effect is demonstrated using measurements of a PDMS sample. The combination of potentially very high spectral acquisition rates, fundamental SNR advantage and the use of low-cost detector systems could make mid-IR swept lasers a powerful technology for high-throughput biomedical spectroscopy.

Fiber optical parametric amplifier (FOPA) has gained its popularity in the telecommunication systems at the 1.5-um window for its gain, bandwidth etc. Unfortunately, its practical application at the bio-favorable window, i.e. 1.0 um, still requires substantial efforts. Thus, here we report a versatile all-fiber optical parametric amplifier for life-science (OPALS) at 1.0 um as an add-on module for optical imaging system. The parametric gain fiber (photonic-crystal fiber (PCF), 110 m in length) is specially designed to reduce the longitudinal dispersion fluctuation, which yields a superior figure of merit, i.e. a total insertion loss of ~2.5 dB and a nonlinear coefficient of 34 /(W•km). Our OPALS delivers a superior performance in terms of gain (~158,000), bandwidth (>100 nm) and gain flatness (< 3-dB ripple). Experimentally, we show that: 1) a wavelength-varying quasi-monochrome pump achieves a 52-dB gain and 160-nm bandwidth, but at the expense of a larger gain-spectrum ripple, i.e. a bell-shaped; 2) the birefringence of the parametric gain medium, i.e. PCF in this case, can be utilized to improve the gain-spectrum flatness of OPALS by 10.5 dB, meanwhile a 100-nm bandwidth can be guaranteed; 3) the gain-spectrum flatness of OPALS can be further flattened by using a high-speed wavelength-sweeping pump, which exhibits a 110-nm flat gain spectrum with ripple less than 3 dB. Finally, we employ this versatile all-fiber OPALS as an add-on module to enhance the sensitivity of a spectrally-encoded microscope by 47 dB over an ultra-wide spectral range.

Particle analysis is an effective method in analytical chemistry for sizing and counting microparticles such as emulsions, colloids, and biological cells. However, conventional methods for particle analysis, which fall into two extreme categories, have severe limitations. Sieving and Coulter counting are capable of analyzing particles with high throughput, but due to their lack of detailed information such as morphological and chemical characteristics, they can only provide statistical results with low specificity. On the other hand, CCD or CMOS image sensors can be used to analyze individual microparticles with high content, but due to their slow charge download, the frame rate (hence, the throughput) is significantly limited. Here by integrating a time-stretch optical microscope with a three-color fluorescent analyzer on top of an inertial-focusing microfluidic device, we demonstrate an optofluidic particle analyzer with a sub-micrometer spatial resolution down to 780 nm and a high throughput of 10,000 particles/s. In addition to its morphological specificity, the particle analyzer provides chemical specificity to identify chemical expressions of particles via fluorescence detection. Our results indicate that we can identify different species of microparticles with high specificity without sacrificing throughput. Our method holds promise for high-precision statistical particle analysis in chemical industry and pharmaceutics.

High-speed continuous imaging systems are constrained by analog-to-digital conversion, storage, and transmission. However, real video signals of objects such as microscopic cells and particles require only a few percent or less of the full video bandwidth for high fidelity representation by modern compression algorithms. Compressed Sensing (CS) is a recent influential paradigm in signal processing that builds real-time compression into the acquisition step by computing inner products between the signal of interest and known random waveforms and then applying a nonlinear reconstruction algorithm. Here, we extend the continuous high-rate photonically-enabled compressed sensing (CHiRP-CS) framework to acquire motion contrast video of microscopic flowing objects. We employ chirp processing in optical fiber and high-speed electro-optic modulation to produce ultrashort pulses each with a unique pseudorandom binary sequence (PRBS) spectral pattern with 325 features per pulse at the full laser repetition rate (90 MHz). These PRBS-patterned pulses serve as random structured illumination inside a one-dimensional (1D) spatial disperser. By multiplexing the PRBS patterns with a user-defined repetition period, the difference signal y_i=phi_i (x_i - x_{i-tau}) can be computed optically with balanced detection, where x is the image signal, phi_i is the PRBS pattern, and tau is the repetition period of the patterns. Two-dimensional (2D) image reconstruction via iterative alternating minimization to find the best locally-sparse representation yields an image of the edges in the flow direction, corresponding to the spatial and temporal 1D derivative. This provides both a favorable representation for image segmentation and a sparser representation for many objects that can improve image compression.

Based on the interferometric or holographic approaches, recent QPM techniques provide quantitative-phase information, e.g cell volume, dry mass and optical scattering properties for label-free cellular physical phenotyping. These approaches generally rely on iterative phase-retrieval algorithms to obtain quantitative-phase information, which are computationally intensive. Moreover, current QPM techniques can only offer limited image acquisition rate by using CMOS/CCD image sensors, these two limitations hinder QPM for high-throughput quantitative image-based single-cell analysis in real-time. To this end, we demonstrate an interferometry-free quantitative phase microscopy developed on a new generation of time-stretch microscopy, asymmetric-detection time-stretch optical microscopy (ATOM), which is coined quantitative ATOM (Q-ATOM) - featuring an unprecedented cell measurement throughput together with the assorted intrinsic optical phenotypes (e.g. angular light scattering profile) and the derived physical properties of the cells (e.g. cell size, dry mass density etc.). Based on a similar concept to Schlieren imaging, Q-ATOM retrieves quantitative-phase information through multiple off-axis light-beam detection at a line-scan rate of <10 MHz - a speed unachievable by any existing QPM techniques. Phase retrieval in Q-ATOM relies on a non-iterative method, significantly reducing the computational complexity of the technique. It is a particularly important feature which facilitates real-time continuous label-free single-cell analysis in Q-ATOM. With the use of a non-interferometric configuration, we demonstrate ultrafast Q-ATOM of mouse chondrocytes and hypertrophic chondrocytes in ultrafast microfluidic flow with sub-cellular resolution at an imaging throughput equivalent to ~100,000 cells/sec without image blur. This technique shows a great potential for ultrahigh throughput label-free image-based single-cell biophysical phentotyping.

Flow cytometry analyzes multiple physical characteristics of a large population of single cells as cells flow in a fluid stream through an excitation light beam. Flow cytometers measure fluorescence and light scattering from which information about the biological and physical properties of individual cells are obtained. Although flow cytometers have massive statistical power due to their single cell resolution and high throughput, they produce no information about cell morphology or spatial resolution offered by microscopy, which is a much wanted feature missing in almost all flow cytometers. In this paper, we invent a method of spatial-temporal transformation to provide flow cytometers with cell imaging capabilities. The method uses mathematical algorithms and a specially designed spatial filter as the only hardware needed to give flow cytometers imaging capabilities. Instead of CCDs or any megapixel cameras found in any imaging systems, we obtain high quality image of fast moving cells in a flow cytometer using photomultiplier tube (PMT) detectors, thus obtaining high throughput in manners fully compatible with existing cytometers. In fact our approach can be applied to retrofit traditional flow cytometers to become imaging flow cytometers at a minimum cost. To prove the concept, we demonstrate cell imaging for cells travelling at a velocity of 0.2 m/s in a microfluidic channel, corresponding to a throughput of approximately 1,000 cells per second.

Comprehensive quantification of phytoplankton abundance, sizes and other parameters, e.g. biomasses, has been an important, yet daunting task in aquatic sciences and biofuel research. It is primarily because of the lack of effective tool to image and thus accurately profile individual microalgae in a large population. The phytoplankton species are highly diversified and heterogeneous in terms of their sizes and the richness in morphological complexity. This fact makes time-stretch imaging, a new ultrafast real-time optical imaging technology, particularly suitable for ultralarge-scale taxonomic classification of phytoplankton together with quantitative image recognition and analysis. We here demonstrate quantitative imaging flow cytometry of single phytoplankton based on quantitative asymmetric-detection time-stretch optical microscopy (Q-ATOM) – a new time-stretch imaging modality for label-free quantitative phase imaging without interferometric implementations. Sharing the similar concept of Schlieren imaging, Q-ATOM accesses multiple phase-gradient contrasts of each single phytoplankton, from which the quantitative phase profile is computed. We employ such system to capture, at an imaging line-scan rate of 11.6 MHz, high-resolution images of two phytoplankton populations (scenedesmus and chlamydomonas) in ultrafast microfluidic flow (3 m/s). We further perform quantitative taxonomic screening analysis enabled by this technique. More importantly, the system can also generate quantitative phase images of single phytoplankton. This is especially useful for label-free quantification of biomasses (e.g. lipid droplets) of the particular species of interest – an important task adopted in biofuel applications. Combining machine learning for automated classification, Q-ATOM could be an attractive platform for continuous and real-time ultralarge-scale single-phytoplankton analysis.

Compressive sampling (CS) is an emerging field that provides a new framework for image reconstruction and has potentially powerful implications for the design of optical imaging devices. Single-pixel camera, as a representative example of CS, enables the use of exotic detectors and can operate efficiently across a much broader spectral range than conventional silicon-based cameras. Recently, time-stretch CS imaging system is proposed to overcome the speed limitation of the conventional single-pixel camera. In the proposed system, as ultra-short optical pulses are used for active illumination, the performance of the imaging system is affected by the detection bandwidth. In this paper, we experimentally analyze the bandwidth limitation in the CS-based time-stretch imaging system. Various detector bandwidths are introduced in the system and the mean square error (MSE) is calculated to evaluate the quality of reconstructed images. The results show that the decreasing detection bandwidth leads to serious energy spread of the pulses, where the MSE increases rapidly and system performance is degraded severely.

We present a study on the characteristics of the AMD method. We have demonstrated that the photon economy of the AMD method is not degraded for longer lifetimes even when the applied integration window size is increased. By an extension of MCS, the photon economy with respect to different designs of the Gaussian low-pass filter (GLPF) used in the AMD setup was also studied. When a GLPF with the highest cutoff frequency of 100 MHz is applied, the most effective photon economy performance is achieved for lifetimes of 1, 3.2, 5, and 8 ns.

Optical coherence tomography (OCT) signal can provide microscopic characterization of biological tissue and assist clinical decision making in real-time. However, raw OCT data is noisy and complicated. It is challenging to extract information that is directly related to the pathological status of tissue through visual inspection on huge volume of OCT signal streaming from the high speed OCT engine. Therefore, it is critical to discover concise, comprehensible information from massive OCT data through novel strategies for signal analysis. In this study, we perform Shannon entropy analysis on OCT signal for automatic tissue characterization, which can be applied in intraoperative tumor margin delineation for surgical excision of cancer. The principle of this technique is based on the fact that normal tissue is usually more structured with higher entropy value, compared to pathological tissue such as cancer tissue. In this study, we develop high-speed software based on graphic processing units (GPU) for real-time entropy analysis of OCT signal.

Confocal laser microscope (CLM) has been widely used in the fields of the non-contact surface topography, biomedical imaging, and other applications, because of two-dimensional (2D) or three-dimensional (3D) imaging capability with the confocal effect and the stray light elimination. Although the conventional CLM has acquired the 2D image by mechanical scanning of the focused beam spot, further reduction of image acquisition time and the robustness to various disturbances are strongly required. To this end, it is essential to omit mechanical scanning for the image acquisition. In this article, we developed the scan-less, full-field CLM by combination of the line-focused CLM with the wavelength/1D-space conversion. This combination enables us to form the 2D focal array of a 2D rainbow beam on a sample and to encode the 2D image information of a sample on the 2D rainbow beam. The image-encoded 2D rainbow beam was decoded as a spectral line image by a multi-channel spectrometer equipped with a CMOS camera without the need for the mechanical scanning. The confocal full-field image was acquired during 0.23 ms with the lateral resolution of 26.3μm and 4.9μm for the horizontal and vertical directions, respectively, and the depth resolution of 34.9μm. We further applied this scan-less, full-field CLM for biomedical imaging of a sliced specimen and non-contact surface topography of an industry products. These demonstrations highlight a high potential of the proposed scan-less, full-field CLM.

A wavelength-swept fiber optical parametric oscillator (FOPO) based on dispersion tuning technology at wavelength around 1 μm is demonstrated. A continuous wave single-longitudinal-mode ytterbium doped fiber laser with a line-width of 0.05 nm is modulated through a LiNbO3 Mach-Zehnder modulator to be a pulsed source with variable repetition rate. The pulsed source is amplified with a two-stage ytterbium doped fiber amplifier (YDFA) to a mediate power and a high power YDFA to peak power higher than 40 W. And a homemade 50-m photonic crystal fiber (PCF) which provides the optical parametric gain is pumped by the pulsed source. The optical modulator is driven by a frequency-swept electrical clock signal with frequency ranges from 107.24 MHz to 107.31 MHz. Thus the FOPO generates a wavelength-swept light source with a range of 80 nm centered at 1065.10 nm. Through careful customizing the sweeping rate of the driving clock signal, the sweeping rate of the parametric oscillator can be up to 10 kHz, which is limited by currently used electrical sweeping source. The generated pulses train are with pulse width of about 110 ps. For the electrical scan is used instead of the traditional mechanical scanning method in conventional wavelength-swept sources, it performs better stability under prolonged operation. The wavelength-swept FOPO is potential to be applied in OCT systems for its good stability and advantaged wavelength band.