Search

Introduction: Although a powerful biological imaging technique, fluorescence lifetime imaging microscopy (FLIM) faces challenges such as a slow acquisition rate, a low signal-to-noise ratio (SNR), and high cost and complexity. To address the fundamental problem of low SNR in FLIM images, we demonstrate how to use pre-trained convolutional neural networks (CNNs) to reduce noise in FLIM measurements.Methods: Our approach uses pre-learned models that have been previously validated on large datasets with different distributions than the training datasets, such as sample structures, noise distributions, and microscopy modalities in fluorescence microscopy, to eliminate the need to train a neural network from scratch or to acquire a large training dataset to denoise FLIM data. In addition, we are using the pre-trained networks in the inference stage, where the computation time is in milliseconds and accuracy is better than traditional denoising methods. To separate different fluorophores in lifetime images, the denoised images are then run through an unsupervised machine learning technique named “K-means clustering”.Results and Discussion: The results of the experiments carried out on in vivo mouse kidney tissue, Bovine pulmonary artery endothelial (BPAE) fixed cells that have been fluorescently labeled, and mouse kidney fixed samples that have been fluorescently labeled show that our demonstrated method can effectively remove noise from FLIM images and improve segmentation accuracy. Additionally, the performance of our method on out-of-distribution highly scattering in vivo plant samples shows that it can also improve SNR in challenging imaging conditions. Our proposed method provides a fast and accurate way to segment fluorescence lifetime images captured using any FLIM system. It is especially effective for separating fluorophores in noisy FLIM images, which is common in in vivo imaging where averaging is not applicable. Our approach significantly improves the identification of vital biologically relevant structures in biomedical imaging applications.

The insect predator–prey system mediates several feedback mechanisms which regulate species abundance and spatial distribution. However, the spatiotemporal dynamics of such discrete systems with the refuge effect remain elusive. In this study, we analyzed a discrete Holling type II model incorporating the refuge effect using theoretical calculations and numerical simulations, and selected moths with high and low growth rates as two exemplifications. The result indicates that only the flip bifurcation opens the routes to chaos, and the system undergoes four spatiotemporally behavioral patterns (from the frozen random pattern to the defect chaotic diffusion pattern, then the competition intermittency pattern, and finally to the fully developed turbulence pattern). Furthermore, as the refuge effect increases, moths with relatively slower growth rates tend to maintain stability at relatively low densities, whereas moths with relatively faster growth rates can induce chaos and unpredictability on the population. According to the theoretical guidance of this study, the refuge effect can be adjusted to control pest populations effectively, which provides a new theoretical perspective and is a feasible tool for protecting crops.

To ensure the security of information exchange in software-defined optical networks, quantum key distribution (QKD) based on quantum mechanics is introduced. However, the slow and valuable process of generating quantum key resources contradicts the high-speed data transmission requirements of optical networks. To address this issue, this paper proposes the CPSR-HQKDN scheme, which takes into account factors such as security requests, key demand, key residual, and key update rates for trusted and untrusted links. This approach improves resource utilization and service efficiency by optimizing the processing order of key requests. Moreover, the routing strategy dynamically adjusts based on the network resource environment, thereby increasing the success rate of key requests. Through simulation experiments comparing the performance of the CPSR-HQKDN routing scheme with existing schemes, it is observed that in high-concurrent scenarios, the CPSR-HQKDN routing scheme can improve the success rate of key requests by at least 5%.

Matcha and green tea catechins such as (−)-epicatechin (EC), (−)-epigallocatechin (EGC) and (−)-epigallocatechin gallate (EGCG) have long been studied for their antioxidant and health-promoting effects. Using specific fluorophores for H2S (AzMC) and polysulfides (SSP4) as well as IC-MS and UPLC-MS/MS-based techniques we here show that popular Japanese and Chinese green teas and select catechins all catalytically oxidize hydrogen sulfide (H2S) to polysulfides with the potency of EGC > EGCG >> EG. This reaction is accompanied by the formation of sulfite, thiosulfate and sulfate, consumes oxygen and is partially inhibited by the superoxide scavenger, tempol, and superoxide dismutase but not mannitol, trolox, DMPO, or the iron chelator, desferrioxamine. We propose that the reaction proceeds via a one-electron autoxidation process during which one of the OH-groups of the catechin B-ring is autooxidized to a semiquinone radical and oxygen is reduced to superoxide, either of which can then oxidize HS− to thiyl radicals (HS•) which react to form hydrogen persulfide (H2S2). H2S oxidation reduces the B-ring back to the hydroquinone for recycling while the superoxide is reduced to hydrogen peroxide (H2O2). Matcha and catechins also concentration-dependently and rapidly produce polysulfides in HEK293 cells with the potency order EGCG > EGC > EG, an EGCG threshold of ~300 nM, and an EC50 of ~3 μM, suggesting green tea also acts as powerful pro-oxidant in vivo. The resultant polysulfides formed are not only potent antioxidants, but elicit a cascade of secondary cytoprotective effects, and we propose that many of the health benefits of green tea are mediated through these reactions. Remarkably, all green tea leaves constitutively contain small amounts of H2S2.

This paper proposes a novel deep learning framework named bidirectional-convolutional long short term memory (Bi-CLSTM) network to automatically learn the spectral-spatial features from hyperspectral images (HSIs). In the network, the issue of spectral feature extraction is considered as a sequence learning problem, and a recurrent connection operator across the spectral domain is used to address it. Meanwhile, inspired from the widely used convolutional neural network (CNN), a convolution operator across the spatial domain is incorporated into the network to extract the spatial feature. In addition, to sufficiently capture the spectral information, a bidirectional recurrent connection is proposed. In the classification phase, the learned features are concatenated into a vector and fed to a Softmax classifier via a fully-connected operator. To validate the effectiveness of the proposed Bi-CLSTM framework, we compare it with six state-of-the-art methods, including the popular 3D-CNN model, on three widely used HSIs (i.e., Indian Pines, Pavia University, and Kennedy Space Center). The obtained results show that Bi-CLSTM can improve the classification performance by almost 1.5 % as compared to 3D-CNN.

We propose a novel feature, local histogram of figure/ground segmentations, for robust and efficient background subtraction (BGS) in dynamic scenes (e.g., waving trees, ripples in water, illumination changes, camera jitters, etc.). We represent each pixel as a local histogram of figure/ground segmentations, which aims at combining several candidate solutions that are produced by simple BGS algorithms to get a more reliable and robust feature for BGS. The background model of each pixel is constructed as a group of weighted adaptive local histograms of figure/ground segmentations, which describe the structure properties of the surrounding region. This is a natural fusion because multiple complementary BGS algorithms can be used to build background models for scenes. Moreover, the correlation of image variations at neighboring pixels is explicitly utilized to achieve robust detection performance since neighboring pixels tend to be similarly affected by environmental effects (e.g., dynamic scenes). Experimental results demonstrate the robustness and effectiveness of the proposed method by comparing with four representatives of the state of the art in BGS.