Search

The object tracking alongside the image segmentation have recently become a particular significance in satellite and aerial imagery. The latest achievements in this field are closely related to the application of the deep-learning algorithms and, particularly, convolutional neural networks (CNNs). Supplemented by the sufficient amount of the training data, CNNs provide the advantageous performance in comparison to the classical methods based on Viola–Jones or support vector machines. However, the application of CNNs for the object detection on the aerial images faces several general issues that cause classification error. The first one is related to the limited camera shooting angle and spatial resolution. The second one arises from the restricted dataset for specific classes of objects that rarely appear in the captured data. This article represents a comparative study on the effectiveness of different deep neural networks for detection of the objects with similar patterns on the images within a limited amount of the pretrained datasets. It has been revealed that YOLO ver. 3 network enables better accuracy and faster analysis than region convolution neural network (R-CNN), Fast R-CNN, Faster R-CNN, and SSD architectures. This has been demonstrated on the example of “Stanford dataset,” “DOTA v-1.5,” and “xView 2018 Detection” datasets. The following metrics on the accuracy have been obtained for the YOLO ver. 3 network: 89.12 mAP (Stanford dataset), 80.20 mAP (DOTA v-1.5), and 78.29 (xView 2018) for testing; and 85.51 mAP (Stanford dataset), 79.28 (DOTA v-1.5), and 79.92 (xView 2018) on validation with the analysis speed of 26.82 frames/s.

A method is proposed for processing large data sets of fluorescence decay kinetic curves using data mining algorithms to determine the parameters of biophysical and optical processes in molecular systems. The idea of this method involves breaking the initial set of fluorescence decay curves into clusters in terms of some degree of similarity, finding medoids of the clusters, applying a dimensionality reduction method to the data and imaging the experimental data in two- and three-dimensional space, and analyzing the decay curves of the medoids using analytic or simulation models. The applicability of the method is examined for the example of analyzing sets of data representing systems of fluorophores. This method requires substantially less time and calculations of the analytic approximation functions, while the accuracy of the estimated parameters is higher than in the classical approach.

A biochemical arrayed sensor based on optical resonance technology and characterized by the dispersed positioning and properties of individual sensing units is discussed. Fabrication simplicity and reusability of the sensor as well as robustness of signal excitation and possibility for parallel probing are among the advantageous features for the implementation of optical resonance sensing for practical-oriented tasks. The arrayed sensor applicability, performance, and correspondence between the responses and analytical models are analyzed on the example of biochemical components’ monitoring. The efficient regeneration of sensing properties realized by cleaning of the reacting surface allows to perform multiple sensing tasks in a single experimental run. The sensor interrogation approach based on the dispersion of the sensing units’ spectral characteristics is discussed.

Different methods (correlation, logistic regression, median, and random forest methods) for the selection of informative characteristics of the intensity distribution of fluorescent nuclei on multichannel luminescent images of cancer cells are considered. The input data are three-channel RGB images. In total, 39 standard characteristics of distributions are studied, including 13 characteristics per each color channel. It is established that the use of six features permits one to achieve the same classification accuracy as the use of 39 features. Moreover, one can use only two features with an insignificant change in the classification accuracy (by 0.005). It is proposed to use the data of immunohistochemical analysis of biomarkers in breast cancer cells during the analysis of luminescent images when processing the results in oncocytology.

Protein dimerization plays a crucial role in the regulation of numerous biological processes. However, detecting protein dimers in a cellular environment is still a challenge. Here we present a methodology to measure the extent of dimerization of GFP-tagged proteins in living cells, using a combination of fluorescence correlation spectroscopy (FCS) and photon counting histogram (PCH) analysis of single-color fluorescence fluctuation data. We named this analysis method brightness and diffusion global analysis (BDGA) and adapted it for biological purposes. Using cell lysates containing different ratios of GFP and tandem-dimer GFP (diGFP), we show that the average brightness per particle is proportional to the fraction of dimer present. We further adapted this methodology for its application in living cells, and we were able to distinguish GFP, diGFP, as well as ligand-induced dimerization of FKBP12 (FK506 binding protein 12)-GFP. While other analysis methods have only sporadically been used to study dimerization in living cells and may be prone to errors, this paper provides a robust approach for the investigation of any cytosolic protein using single-color fluorescence fluctuation spectroscopy.

object tracking is one of the most sophisticated and least researched tasks in computer vision, especially with respect to unmanned aerial vehicles. Primarily it caused by several challenges such as high distance to the tracking objects, variety in object sizes, camera motion, etc. This paper demonstrates the implementation of a tracking pipeline using the YOLOv4-based architecture network defined as YOLOv4eff with the doubled long-short time memory (LSTM) network for road-objects tracking. Our optimization method within YOLOv4eff addresses higher accuracy by improving the network architecture, using modified CSP techniques, Swish activation function, etc. Our object tracker used convolutional network as feature map extractor network of differential sequential frames, YOLOv4eff as detector and doubled long-short time memory (LSTM) network as tracking locations predictor. Our extensive experimental results on self-collected dataset at the height of 10–30 meters and performance comparison with other state-of-the-art tracking methods show that our LYOLOv4eff is more accurate and effectively applicable to objects tracking from a drone.

MotivationModern methods of whole transcriptome sequencing accurately recover nucleotide sequences of RNA molecules present in cells and allow for determining their quantitative abundances. The coding potential of such molecules can be estimated using open reading frames (ORF) finding algorithms, implemented in a number of software packages. However, these algorithms show somewhat limited accuracy, are intended for single-molecule analysis and do not allow selecting proper ORFs in the case of long mRNAs containing multiple ORF candidates.ResultsWe developed a computational approach, corresponding machine learning model and a package, dedicated to automatic identification of the ORFs in large sets of human mRNA molecules. It is based on vectorization of nucleotide sequences into features, followed by classification using a random forest. The predictive model was validated on sets of human mRNA molecules from the NCBI RefSeq and Ensembl databases and demonstrated almost 95% accuracy in detecting true ORFs. The developed methods and pre-trained classification model were implemented in a powerful ORFhunteR computational tool that performs an automatic identification of true ORFs among large set of human mRNA molecules.Availability and implementationThe developed open-source R package ORFhunteR is available for the community at GitHub repository (https://github.com/rfctbio-bsu/ORFhunteR), from Bioconductor (https://bioconductor.org/packages/devel/bioc/html/ORFhunteR.html) and as a web application (http://orfhunter.bsu.by).

A method is proposed for the complex analysis of fluctuations in the fluorescence intensity of molecular compounds, which allows determining the structural composition of protein oligomers. The idea of the method is to analyze the photon counting histograms of experimental measurements using principal component analysis to assess the presence of oligomeric compounds, and to perform hierarchical cluster analysis, to determine the data classes corresponding to various molecular compounds, followed by selecting cluster medoids to determine the oligomeric composition of protein complexes. The efficiency of the analysis algorithms developed within the framework of the proposed method was confirmed on simulated and experimental photon counting histograms of the measured fluorescence intensity fluctuations of monomeric and dimeric forms of green-fluorescent protein (GFP).

Pattern recognition is one of the most important tasks in aerospace image processing. Various methods based on convolutional neural networks attain state-of-the-art accuracy; however, their effectiveness on exact images is influenced by the chosen architecture and its training parameters.This work present methods based on convolutional neural networks for pattern recognition on the aerospace images. A possibility for objects segmentation into ten classes is demonstrated on example of the multispectral images from the World View 3 satellite. Four networks with different architectures were built, trained and optimized parametrically based on the auto-encoder neural networks. Segmentation results has been analyzed by means of three parameters: training Jacard Index, testing Jacard Index and weight numbers. The positive impact of the properly selected shearing augmentation on extension of a small marked dataset is discussed. The influence of the nonequilibrium classes on the segmentation accuracy and how to account this feature during training of deep neural networks is pointing out.

The maximum entropy method (MEM) was used for the analysis of polarized fluorescence decays of enhanced green fluorescent protein (EGFP) in buffered water/glycerol mixtures, obtained with time-correlated single-photon counting (Visser et al 2016 Methods Appl. Fluoresc. 4 035002). To this end, we used a general-purpose software module of MEM that was earlier developed to analyze (complex) laser photolysis kinetics of ligand rebinding reactions in oxygen binding proteins. We demonstrate that the MEM software provides reliable results and is easy to use for the analysis of both total fluorescence decay and fluorescence anisotropy decay of aqueous solutions of EGFP. The rotational correlation times of EGFP in water/glycerol mixtures, obtained by MEM as maxima of the correlation-time distributions, are identical to the single correlation times determined by global analysis of parallel and perpendicular polarized decay components. The MEM software is also able to determine homo-FRET in another dimeric GFP, for which the transfer correlation time is an order of magnitude shorter than the rotational correlation time. One important advantage utilizing MEM analysis is that no initial guesses of parameters are required, since MEM is able to select the least correlated solution from the feasible set of solutions.

One of the challenges of fluorescence fluctuation fpectroscopy (FFS) is an adequate approximation of a brightness profile. The key feature of fluorescence intensity distribution analysis (FIDA) is a polynomial approximation of a brightness profile. A broad range of brightness profile shapes can be well described by this approximation. A different approach consisting of the introduction of additional fitting parameters, defined as a relative difference between integrals of the actual brightness profile and its Gaussian approximation, is used in photon counting histogram (PCH) analysis. It is sufficient to introduce only one additional fitting parameter (first-order correction) to get an adequate fit to the experimental data in many practical applications. In the current study, we apply these approaches to the theory of time integrated fluorescence cumulants analysis. We demonstrate that developed corrections improve results of FFS analysis applied to simulated and experimental data. The use of different brightness profile approximations and normalizations in PCH and FIDA leads to different estimates of brightness and the number of molecules, even though they represent the same physical quantities. Based on the developed theory, we derive equations that relate brightness and the number of molecules in PCH and FIDA.

A linear algorithm for analysis of the dependences of phase shift on modulation frequency measured by the frequency-domain (phase-modulation) method is discussed. The analysis is made in terms of a model consisting of a sum of exponents adapted for use in the frequency domain. The algorithm is based on linearization of the model with respect to coefficients that depend on the estimated parameters followed by application of a linear weighted least squares method. The accuracy of the resulting estimates depends on the method for determining the weighting factors that are involved. Comparative testing of the operational quality of the algorithm is carried out for several methods of calculating the weighting factors and the reasons affecting their quality are studied. A way of calculating the weighting factors is proposed that enhances the efficiency of the algorithm.

Fluorescence correlation spectroscopy (FCS) and photon-counting histogram (PCH) analysis use the same experimental fluorescence intensity fluctuations, but each analytical method focuses on a different property of the signal. The time-dependent decay of the correlation of fluorescence fluctuations is measured in FCS yielding, for instance, molecular diffusion coefficients. The amplitude distribution of these fluctuations is calculated by PCH analysis yielding information about the molecular brightness of fluorescent species. Analysis of both FCS and PCH results in the molecular concentration of the sample. Using a previously described global analysis procedure we report here precise, simultaneous measurements of diffusion constants and brightness values from single fluorescence fluctuation traces of green-fluorescent protein (GFP, S65T) in the cytoplasm of Dictyostelium cells. The use of a polynomial profile in PCH analysis, describing the detected three-dimensional shape of the confocal volume, enabled us to obtain well fitting results for GFP in cells. We could visualize the polynomial profile and show its deviation from a Gaussian profile.

Here, we present an in vitro assay based on fluorescence correlation spectroscopy (FCS), which allows investigation of the kinetic behaviour of human Dicer. The assay is based on the different mobilities of substrate and product. The change of substrate mobility was independent of the choice of the fluorescence label, allowing exclusion of non-specific photophysical artefacts. Dicer and RNase III cleavage led to different product diffusion times. Single-stranded RNA did not change its mobility after cleavage by both double-strand-specific RNases. In agreement with the literature, the RNase activity of Dicer could be inhibited by substituting Ca2+ for Mg2+. In a defined system of two diffusion species of similar label and mobility differences, such as substrate and product, the linearity of the assay could be proven. An FCS-based enzyme assay is proposed, which allows monitoring of Dicer activity with high specificity in vitro.

We present a simulation model for a stream of photon counts that can be detected in fluorescence fluctuation spectroscopy systems from diffusing molecules undergoing spontaneous transitions between radiative and nonradiative states. Using our model, we can obtain various characteristics of a random stream of photon counts which allows us to test fluorescence fluctuation spectroscopy methods and to predict the results of full-scale experiments.

Green-fluorescent protein (GFP) is the origin of the green bioluminescence color exhibited by several marine hydrozoans and anthozoans. The mechanism is believed to be Förster resonance energy transfer (FRET) within a luciferase-GFP or photoprotein-GFP complex. As the effect is found in vitro at micromolar concentrations, for FRET to occur this complex must have an affinity in the micromolar range. We present here a fluorescence dynamics investigation of the recombinant bioluminescence proteins from the jellyfish Clytia gregaria, the photoprotein clytin in its Ca(2+)-discharged form that is highly fluorescent (λ(max) = 506 nm) and its GFP (cgreGFP; λ(max) = 500 nm). Ca(2+)-discharged clytin shows a predominant fluorescence lifetime of 5.7 ns, which is assigned to the final emitting state of the bioluminescence reaction product, coelenteramide anion, and a fluorescence anisotropy decay or rotational correlation time of 12 ns (20 °C), consistent with tight binding and rotation with the whole protein. A 34 ns correlation time combined with a translational diffusion constant and molecular brightness from fluorescence fluctuation spectroscopy all confirm that cgreGFP is an obligate dimer down to nanomolar concentrations. Within the dimer, the two chromophores have a coupled excited-state transition yielding fluorescence depolarization via FRET with a transfer correlation time of 0.5 ns. The 34 ns time of cgreGFP showed no change upon addition of a 1000-fold excess of Ca(2+)-discharged clytin, indicating no stable complexation below 0.2 mM. It is proposed that any bioluminescence FRET complex with micromolar affinity must be one formed transiently by the cgreGFP dimer with a short-lived (millisecond) intermediate in the clytin reaction pathway.

Excitation energy migration between dyes embedded in hexagonal crystals of cylinder morphology is an attractive phenomenon for the construction of photonic anten- nae (1, 2). Detailed knowledge of the zeolite structure, the organization and the spectro- scopic properties of the dyes and the nature and strength of the host-guest interactions is required to optimize energy migration (EnM). Whether a dye-zeolite antenna efficiently transports excitation energy is mainly determined by the mechanism and rate of energy transfer (EnT) between the dyes embedded in the zeolite channels. The decay of the lumi- nescence of these dyes, which reveals information about the EnT, is measured indirectly using Multi-Frequency Phase Fluorimetry (MFPF). Subsequently a fit of the measured data to a multiexponential decay is performed. A new, user-friendly Windows software has been developed which performs this fit, allowing full control over all fit parameters and providing useful information about the quality of the fit. The software also contains a database capable of storing all the relevant information related to the experiment. It is therefore a highly optimized analysis and managing tool for our experimental method which has greatly improved the efficiency of data analysis. The software is highly flexible and can easily be adapted to other experimental methods.

An approach for the automated whispering gallery mode (WGM) signal decomposition and its parameter estimation is discussed. The algorithm is based on the peak picking and can be applied for the preprocessing of the raw signal acquired from the multiplied WGM-based biosensing chips. Quantitative estimations representing physically meaningful parameters of the external disturbing factors on the WGM spectral shape are the output values. Derived parameters can be directly applied to the further deep qualitative and quantitative interpretations of the sensed disturbing factors. The algorithm is tested on both simulated and experimental data taken from the bovine serum albumin biosensing task. The proposed solution is expected to be a useful contribution to the preprocessing phase of the complete data analysis engine and is expected to push the WGM technology toward the real-live sensing nanobiophotonics.

In this chapter, we describe the global analysis approach for processing time-resolved fluorescence spectroscopy data of molecules in the condensed phase. Combining simultaneous analysis of data measured under different experimental conditions (spatial coordinates, temperature, concentration, emission wavelength, excitation intensity, etc.) with the fitting strategy, enabling parameter linkage and thus decreasing the total amount of estimated variables, makes global analysis more robust and more consistent compared to a sequential fit of single experimental data. We consider the main stages of the global analysis approach and provide some details that are important for its practical implementation. The application of the global approach to the analysis of time-resolved fluorescence anisotropy is demonstrated on experimental data of (enhanced) green fluorescent protein in aqueous solution.

In fluorescence correlation spectroscopy (FCS) and photon counting histogram (PCH) analysis, the same experimental fluorescence intensity fluctuations are used, but each analytical method focuses on a different property of the signal. The time-dependent decay of the correlation of fluorescence fluctuations is measured in FCS yielding molecular diffusion coefficients and triplet-state parameters such as fraction and decay time. The amplitude distribution of these fluctuations is calculated by PCH analysis yielding the molecular brightness. Both FCS and PCH give information about the molecular concentration. Here we describe a global analysis protocol that simultaneously recovers relevant and common parameters in model functions of FCS and PCH from a single fluorescence fluctuation trace. Application of a global analysis approach allows increasing the information content available from a single measurement that results in more accurate values of molecular diffusion coefficients and triplet-state parameters and also in robust, time-independent estimates of molecular brightness and number of molecules.

In modern fluorescence fluctuation spectroscopy, the autocorrelation function and photon counting distribution are two widely used statistical characteristics of the measured fluctuating fluorescence intensity signal. Applying special analysis methods such as fluorescence correlation spectroscopy (FCS) and photon counting histogram (PCH) to these properties, it is possible to recover values of different parameters of fluorescent molecules such as the concentration, diffusion coefficient, molecular brightness, and kinetic rate constants. The development of new analysis methods is senseless without testing their validity, accuracy, and robustness. The most appropriate check of a method is its application to experimental data. However, sometimes it is more convenient and easier to verify a method on simulated data. Simulation is also useful for better understanding the processes that were modeled during the development of analysis methods. Here, we present two simulation models providing an autocorrelation function and photon counting distribution of a sequence of photon arrival times detected in fluorescence fluctuation spectroscopy.

In fluorescence correlation spectroscopy (FCS) and photon counting histogram (PCH) analysis, the same experimental fluorescence intensity fluctuations are used, but each analytical method focuses on a different property of the signal. The time-dependent decay of the correlation of fluorescence fluctuations is measured in FCS yielding molecular diffusion coefficients and triplet-state parameters such as fraction and decay time. The amplitude distribution of these fluctuations is calculated by PCH analysis yielding the molecular brightness. Both FCS and PCH give information about the molecular concentration. Here we describe a global analysis protocol that simultaneously recovers relevant and common parameters in model functions of FCS and PCH from a single fluorescence fluctuation trace. Application of a global analysis approach allows increasing the information content available from a single measurement that results in more accurate values of molecular diffusion coefficients and triplet-state parameters and also in robust, time-independent estimates of molecular brightness and number of molecules.

In modern fluorescence fluctuation spectroscopy, the autocorrelation function and photon counting distribution are two widely used statistical characteristics of the measured fluctuating fluorescence intensity signal. Applying special analysis methods such as fluorescence correlation spectroscopy (FCS) and photon counting histogram (PCH) to these properties, it is possible to recover values of different parameters of fluorescent molecules such as the concentration, diffusion coefficient, molecular brightness, and kinetic rate constants. The development of new analysis methods is senseless without testing their validity, accuracy, and robustness. The most appropriate check of a method is its application to experimental data. However, sometimes it is more convenient and easier to verify a method on simulated data. Simulation is also useful for better understanding the processes that were modeled during the development of analysis methods. Here, we present two simulation models providing an autocorrelation function and photon counting distribution of a sequence of photon arrival times detected in fluorescence fluctuation spectroscopy.

In fluorescence correlation spectroscopy (FCS) and photon counting histogram (PCH) analysis the same experimental fluorescence intensity fluctuations are used, but each analytical method focuses on a different property of the signal. The time-dependent decay of the correlation of fluorescence fluctuations is measured in FCS yielding, for instance, molecular diffusion coefficients. The amplitude distribution of these fluctuations is calculated by PCH yielding the molecular brightness. Both FCS and PCH give information about the molecular concentration. Here we describe a global analysis protocol that simultaneously recovers relevant and common parameters in model functions of FCS and PCH from a single fluorescence fluctuation trace. The global analysis approach is described and tested with experimental fluorescence fluctuation data of enhanced green-fluorescent protein (eGFP) and dimeric eGFP (two eGFP molecules connected by a six amino acid long linker) in aqueous buffer. Brightness values and diffusion constants are recovered with good precision elucidating novel excited-state and motional properties of both proteins.

Fluorescence correlation spectroscopy (FCS) provides a versatile tool to investigate molecular interaction under native conditions, approximating infinite dilution. One precondition for its application is a sufficient difference between the molecular weights of the fluorescence-labelled unbound and bound ligand. In previous studies, an 8-fold difference in molecular weights or correspondingly a 1.6-fold difference in diffusion coefficients was required to accurately distinguish between two diffusion species by FCS. In the presented work, the hybridization of two complementary equally sized RNA single strands was investigated at an excellent signal-to-noise ratio enabled by the highly photostable fluorophore Atto647N. The fractions of ssRNA and dsRNA were quantified by applying multicomponent model analysis of single autocorrelation functions and globally fitting several autocorrelation functions. By introducing a priori knowledge into the fitting procedure, 1.3- to 1.4-fold differences in diffusion coefficients of single- and double-stranded RNA of 26, 41, and 54 nucleotides could be accurately resolved. Global fits of autocorrelation functions of all titration steps enabled a highly accurate quantification of diffusion species fractions and mobilities. At a high signal-to-noise ratio, the median of individually fitted autocorrelation functions allowed a robust representation of heterogeneous data. These findings point out the possibility of studying molecular interaction of equally sized molecules based on their diffusional behavior, which significantly broadens the application spectrum of FCS.

The growing number of applications of Fluorescence Intensity Distribution Analysis (FIDA) demands for new approaches in data processing, aiming at increased speed and robustness. Iterative algorithms of parameter estimation, although proven to be universal and accurate, require some initial guesses (IG) of the unknown parameters. An essential component of any data processing technology, IG become especially important in case of FIDA, since even with apparently reasonable, and physically admissible but randomly chosen IG, the iterative procedure may converge to situations where the FIDA model cannot be evaluated correctly. In the present work we introduce an approach for IG generation in FIDA experiments based on the method of moments. IG are generated for the sample parameters: brightness, concentration, and for the parameters related to experimental set-up: background, observation volume profile. A number of analytical simplifications were introduced in order to increase the accuracy and robustness of the numerical algorithms. The performance of the developed method has been tested on number of simulations and experimental data. Iterative fitting with generated IG proved to be more robust and at least five times faster than with an arbitrarily chosen IG. Applicability of the proposed method for quick estimation of brightness and concentrations is discussed.