Search bioRxiv⌕ Search

Biology subjects

Fei, P.

Publications and source records attributed to Fei, P..

27 records · Page 2Linked to original sources
bioRxiv2020

Bi-channel Image Registration and Deep-learning Segmentation (BIRDS) for efficient, versatile 3D mapping of mouse brain

We have developed an open-source software called BIRDS (bi-channel image registration and deep-learning segmentation) for the mapping and analysis of 3D microscopy data of mouse brain. BIRDS features a graphical user interface that is used to submit jobs, monitor their progress, and display results. It implements a full pipeline including image pre-processing, bi-channel registration, automatic annotation, creation of 3D digital frame, high-resolution visualization, and expandable quantitative analysis (via link with Imaris). The new bi-channel registration algorithm is adaptive to various types of whole brain data from different microscopy platforms and shows obviously improved registration accuracy. Also, the attraction of combing registration with neural network lies in that the registration procedure can readily provide training data for network, while the network can efficiently segment incomplete/defective brain data that are otherwise difficult for registration. Our software is thus optimized to enable either minute-timescale registration-based segmentation of cross-modality whole-brain datasets, or real-time inference-based image segmentation for various brain region of interests. Jobs can be easily implemented on Fiji plugin that can be adapted for most computing environments.

neuroscience↗
bioRxiv2020

Network-enabled efficient image restoration for 3D microscopy of turbid biological specimens

Though three-dimensional (3D) fluorescence microscopy has been an essential tool for modern life science research, the light scattering by biological specimens fundamentally prevents its more widespread applications in live imaging. We hereby report a deep-learning approach, termed ScatNet, that enables reversion of 3D fluorescence microscopy from high-resolution targets to low-quality, light-scattered measurements, thereby allowing restoration for a single blurred and light-scattered 3D image of deep tissue, with achieving improved resolution and signal-to-noise ratio. Our approach can computationally extend the imaging depth for current 3D fluorescence microscopes, without the addition of complicated optics. Combining ScatNet approach with cutting-edge light-sheet fluorescence microscopy, we demonstrate that the image restoration of cell nuclei in the deep layer of live Drosophila melanogaster embryos at single-cell resolution. Applying our approach to two-photon excitation microscopy, we could improve the signal and resolution of neurons in mouse brain beyond the photon ballistic region.

bioengineering↗
bioRxiv2019

Minutes-timescale 3D isotropic imaging of entire organs at subcellular resolution by content-aware compressed-sensing light-sheet microscopy

Instant 3D imaging of entire organs and organisms at cellular resolution is a recurring challenge in life science. Here we report on a computational light-sheet microscopy able to achieve minute-timescale mapping of entire macro-scale organs at high spatial resolution, thereby overcoming the throughput limit of current 3D microscopy implementations. Through combining a dual-side confocally-scanned Bessel light-sheet illumination which provides thinner-and-wider optical sectioning of deep tissues, with a content-aware compressed sensing (CACS) computation pipeline which further improves the contrast and resolution based on a single acquisition, our method yields 3D images with high, isotropic spatial resolution and rapid acquisition improved by two-orders of magnitude. We demonstrate the imaging of whole brain ([~]400 mm3), entire gastrocnemius and tibialis muscles ([~]200 mm3) of mouse at subcellular resolution (0.5-m isovoxel) and ultra-high throughput of 5[~]10 minutes per sample. Various system-level cellular analyses, such as mapping cell populations at different brain sub-regions, tracing long-distance projection neurons over the entire brain, and calculating neuromuscular junction occupancy across whole muscle, were also readily enabled by our method.

bioengineering↗
bioRxiv2019

Lossless and Contamination-Free Digital PCR

The realization of the vast potential of digital PCR (dPCR), to provide extremely accurate and sensitive measurements in the clinical setting, has thus far been hindered by challenges such as assay robustness and high costs. Existing popular dPCR platforms that target the clinic have not reached wide-spread adoption, due to problems with sample loss and risk of contamination during sample preparation, compartmentalization, and transfers; limitations of dynamic range and signal-to-noise in the result readout also restricts broad applications. Here we introduce a lossless and contamination-free dPCR technology termed CLEAR-dPCR, which addresses these challenges by completing the dPCR sample preparation, PCR reaction, and readout all in one tube. We achieve this by adjusting the refractive index of the aqueous PCR mix to make the emulsion optically transparent, and devised a light-sheet microscope to capture 3D images of the cleared emulsion for results readout. This approach demonstrates improved accuracy over existing dPCR platforms, and enables a greatly increased dynamic range to be comparable to that of real-time quantitative PCR (qPCR). CLEAR-dPCR is an easy to operate, sensitive and accurate dPCR platform that we envision will fulfill the potential of dPCR for routine use clinical diagnosis.

bioengineering↗
bioRxiv2018

Generative adversarial network (GAN) enabled on-chip contact microscopy

We demonstrate a deep learning based contact imaging on a CMOS chip to achieve [~]1 m spatial resolution over a large field of view of [~]24 mm2. By using regular LED illumination, we acquire the single lower-resolution image of the objects placed approximate to the sensor with unit fringe magnification. For the raw contact-mode lens-free image, the pixel size of the sensor chip limits the spatial resolution. We apply a generative and adversarial network (GAN), a type of deep learning algorithm, to circumvent this limitation and effectively recover much higher resolution image of the objects, permitting sub-micron spatial resolution to be achieved across the entire sensor chip active area, which is also equivalent to the imaging field-of-view (24 mm2) due to unit magnification. This GAN-contact imaging approach eliminates the need of either lens or multi-frame acquisition, being very handy and cost-effective. We demonstrate the success of this approach by imaging the proliferation dynamics of cells directly cultured on the chip.

biophysics↗
bioRxiv2018

Deep-learning light-sheet fluorescence microscopy for high-throughput, voxel-super-resolved imaging of biomedical specimens

Although three-dimensional (3D) fluorescence microscopy is an essential tool for life science research, the fundamentally-limited optical throughput, as reflected in the compromise between speed and resolution, so far prevents further movement towards faster, clearer, and higher-throughput applications. We herein report a dual-stage mutual-feedback deep-learning approach that allows gradual reversion of microscopy degradation from high-resolution targets to low-resolution images. Using a single blurred-and-pixelated 3D image as input, our trained network infers a 3D output with notably higher resolution and improved contrast. The performance is better than conventional 1-stage network approaches. It pushes the throughput limit of current 3D fluorescence microscopy in three ways: notably reducing the acquisition time for accurate mapping of large organs, breaking the diffraction limit for imaging subcellular events with faster lower-toxicity measurement, and improving temporal resolution for capturing instantaneous biological processes. Combining our network approach with light-sheet fluorescence microscopy, we demonstrate the imaging of vessels and neurons in the mouse brain at single-cell resolution and with a throughput of 6 minutes for a whole brain. We also image cell organelles beyond the diffraction limit at a 2-Hz volume rate, and map neuronal activities of freely-moving C. elegans at single-cell resolution and 30-Hz volume rate.

biophysics↗
bioRxiv2018

Deep learning light field microscopy for rapid four-dimensional imaging of behaving animals

Light-field microscopy has emerged as a technique of choice for high-speed volumetric imaging of fast biological processes. However, artefacts, non-uniform resolution, and a slow reconstruction speed have limited its full capabilities for in toto extraction of the dynamic spatiotemporal patterns in samples. Here, we combined a view-channel-depth (VCD) neural network with light-field microscopy to mitigate these limitations, yielding artefact-free three-dimensional image sequences with uniform spatial resolution and three-order-higher video-rate reconstruction throughput. We imaged neuronal activities across moving C. elegans and blood flow in a beating zebrafish heart at single-cell resolution with volume rates up to 200 Hz.

biophysics↗
bioRxiv2018

Sub-voxel light-sheet microscopy for high-resolution, high-throughput volumetric imaging of large biomedical specimens

A key challenge when imaging whole biomedical specimens is how to quickly obtain massive cellular information over a large field of view (FOV). Here, we report a sub-voxel light-sheet microscopy (SLSM) method enabling high-throughput volumetric imaging of mesoscale specimens at cellular-resolution. A non-axial, continuous scanning strategy is used to rapidly acquire a stack of large-FOV images with three-dimensional (3-D) nanoscale shifts encoded. Then by adopting a sub-voxel-resolving procedure, the SLSM method models these low-resolution, cross-correlated images in the spatial domain and iteratively recovers a 3-D image with improved resolution throughout the sample. This technique can surpass the optical limit of a conventional light-sheet microscope by more than three times, with high acquisition speeds of gigavoxels per minute. As demonstrated by quick reconstruction (minutes to hours) of various samples, e.g., 3-D cultured cells, an intact mouse heart, mouse brain, and live zebrafish embryo, the SLSM method presents a high-throughput way to circumvent the tradeoff between intoto mapping of large-scale tissue (>100 mm3) and isotropic imaging of single-cell (~1-m resolution). It also eliminates the need of complicated mechanical stitching or precisely modulated illumination, using a simple light-sheet setup and fast graphics-processing-unit (GPU)-based computation to achieve high-throughput, high-resolution 3-D microscopy, which could be tailored for a wide range of biomedical applications in pathology, histology, neuroscience, etc.

bioengineering↗