Least-squares Fitting Research Articles

Tree position estimation from TLS data using hough transform and robust least-squares circle fitting

Forest management and planning require information regarding the current state of the forest. Remote sensing techniques allow to obtain geospatial data, also for the forestry sector. As one of the remote-sensed technologies datasets, Terrestrial Laser Scanning data is widely used to derive detailed information about tree and forest stand parameters. This article presents the combination of circular Hough transform, denoising procedure, and robust least-square circle fitting method to extract stem positions from Terrestrial Laser Scanning data. In the proposed approach, initial tree stems position was detected with circular Hough transform. Then, obtained results were denoised to exclude most non-tree trunk points and analyze three-dimensional data from laser scanning to find exact circular tree stems with a robust least-square circle fitting method. The developed algorithm is effective in obtaining the trees’ geodetic positions from laser scanning data. The results generated in this study can be used as basics for further automatic determination of tree characteristics, such as tree species, height, or crown range. In this study, 94.8% tree stems delineation was generated with a mean accuracy of 87.2%, 1.64 cm of root mean square error for stem position, and 1.15 cm for tree radius measured at ground level. The process conducted in this research can be used to detect other circle-shaped objects, such as lamps or power towers, for which obtaining dense Terrestrial Laser Scanning data is available. The detected positions of these objects can power the geographic information systems or thematic industry systems, where it is necessary to determine the geodetic object position results from legal regulations.

Simplified Algorithms for Adaptive Experiment Design in Parameter Estimation.

Measurements to estimate parameters of a model are commonplace in the physical sciences, where the traditional approach to automation is to use a sequence of preselected settings followed by least-squares fitting of a model function to the data. This measure-then-fit approach is simple and effective and entirely appropriate for many applications but when measurement resources are limited, efficiency becomes more important. To increase efficiency, Bayesian experiment design allows measurement settings to be chosen adaptively based on accumulated data and utility, the predicted improvement in results as a function of settings. However, the calculation of utility has been judged too impractical for most applications. In this paper, we introduce computational methods and simplified algorithms that accelerate utility calculations by over an order of magnitude, with only slight degradation in measurement efficiency. The methods eliminate utility calculation as a barrier to practical application of efficient adaptive measurement.

An unsupervised convolutional neural network method for estimation of intravoxel incoherent motion parameters

Objective. Intravoxel incoherent motion (IVIM) imaging obtained by fitting a biexponential model to multiple b-value diffusion-weighted magnetic resonance imaging (DW-MRI) has been shown to be a promising tool for different clinical applications. Recently, several deep neural network (DNN) methods were proposed to generate IVIM imaging. Approach. In this study, we proposed an unsupervised convolutional neural network (CNN) method for estimation of IVIM parameters. We used both simulated and real abdominal DW-MRI data to evaluate the performance of the proposed CNN-based method, and compared the results with those obtained from a non-linear least-squares fit (TRR, trust-region reflective algorithm) and a feed-forward backward-propagation DNN-based method. Main results. The simulation results showed that both the DNN- and CNN-based methods had lower coefficients of variation than the TRR method, but the CNN-based method provided more accurate parameter estimates. The results obtained from real DW-MRI data showed that the TRR method produced many biased IVIM parameter estimates that hit the upper and lower parameter bounds. In contrast, both the DNN- and CNN-based methods yielded less biased IVIM parameter estimates. Overall, the perfusion fraction and diffusion coefficient obtained from the DNN- and CNN-based methods were close to literature values. However, compared with the CNN-based method, both the TRR and DNN-based methods tended to yield increased pseudodiffusion coefficients (55%–180%). Significance. Our preliminary results suggest that it is feasible to estimate IVIM parameters using CNN.

Weighted least-squares fitting of circles with variance component estimation

Although the least-squares (LS) circle fit has been widely used, the weighted LS fitting of circle is not thoroughly investigated, in particular, when the prior weight information is only partly known. Based on the Gauss–Helmert model (GHM), we first investigate the invariance on translation and rotation. The results show that though the translational invariance holds, the rotation invariance is broken except for some specific weight structures. As the main finding of this paper, we develop the VCE theory directly adapting to the nonlinear GHM representation of the circle fitting problem where the weight information is not exactly known. In the simulated example and the real applications, we show that: (1) The conclusions about the invariance of translation and rotation are validated. (2) The estimated variance components can perfectly represent the uncertainty of different point groups or different coordinate components from the statistical perspective.

Fast time optical variability in Be/X-ray binaries

Context. Classical Be stars, regardless of spectral subtype, display multi-periodic light modulations in the frequency range 0.1–12 c d−1 when observed with a high cadence and a long duration. This behavior is attributed to non-radial pulsations and/or the rotation of the Be star. A similar study on the optical counterparts to Be/X-ray binaries is yet to be carried out. Aims. The main goal of this work is to investigate the fast photometric variability of the optical counterparts to Be/X-ray binaries and compare the general patterns of such variability with the Galactic population of classical Be stars. Methods. The main core of our analysis is based on space-based observations performed by TESS. We analyzed 21 sources with TESS. High-cadence photometry with two ground-based telescopes was also performed for 15 sources. The TESS light curves were created from the full-frame images using the Lightkurve package. The ground-based light curves were obtained through differential photometry between the target and a number of non-variable stars in the same field of view. Standard Fourier analysis and least-squares fitting methods were employed in the frequency analysis. Results. All sources exhibit intra-night light variations with intensity variations of 0.01–0.06 mag in the ground-based observations and up to 5% in flux in TESS observations. This variability manifests itself as multi-periodic signals in the frequency range 0.2–12 c d−1. We find that the patterns of variability of the Be stars in Be/X-ray binaries agree with that of classical early-type Be stars in terms of the general shape of the periodograms. Based on the general shape and number of peaks in the periodograms, Be/X-ray binaries can be classified into different types. The most common case is the presence of groups of closely spaced frequencies (67%), followed by sources that exhibit isolated signals (18%). The remaining source type displays frequency spectra characterized by a mixed pattern of stochastic variability and high-frequency peaks. Conclusions. This study reveals that short-term optical photometric variability is a very common, if not ubiquitous, feature intrinsic to the Be optical companions in Be/X-ray binaries. This variability is mainly attributed to pulsations that originate in the stellar interior.

Avalanche Photodiode-Based γ-Photon Scintillation Detectors for Personalized Dosimetry in Targeted Radionuclide Therapy

<h3>Purpose/Objective(s)</h3> Targeted radionuclide therapy holds great promise as shown in neuroendocrine and prostate cancer, but highly variable biodistribution and tumor uptake requires personalized dosimetry for each patient. Optimal dosimetry requires the measurement of total integrated dose, demanding dose measurements over multiple half-lives of the radionuclide. The current state-of-art, SPECT, is often only available to a small subset of patients and is rarely done more than once. Therefore, a method for continuous radionuclide dosimetry is needed. Highly-sensitive γ-photon detectors placed on the body could solve this problem by monitoring γ decays indicative of the activity within tumors and organs at risk (OAR). We have developed a system capable of detecting γ-photons externally, in real time, with high sensitivity and dynamic range. Here we present our results for photon detection as well as a methodology to use these signals to monitor the activity in tumors and OAR in real time. <h3>Materials/Methods</h3> Y₂O₃ Eu-doped phosphor was compacted at the interface of a bare optical fiber to scintillate incoming γ-photons into 610 nm (red) light. The optical fiber was coupled to avalanche photodiodes (APDs) with a 60% peak detection efficiency at 610 nm and dead time of 50ns. The transistor-transistor logic signals from each APD are conveyed to an FPGA to compute counts per minute (CPM). The system was tested by measuring the emitted γ-photons from a titration of ¹⁷⁷Lu. Six vials were prepared by serially diluting ¹⁷⁷Lu to activities per mL ranging from 0.6∼566.2µCi/mL. The detector was placed on the outside of the vials, and the vials were incrementally changed. The CPM were recorded in real time. To determine the ability to reconstruct a SPECT image in tumors and OAR, behavioral simulations were conducted. %ID/g in these organs, paralleling the measured activity in the ¹⁷⁷Lu dilutions, was varied and the ability of the sensors to track these variations was evaluated. <h3>Results</h3> The mean CPM for the six vials ranged from 30.8 CPM (0.6µCi/mL) to 47,240 CPM (566.2µCi/mL) and was highly linear with activity. The least-squares regression fit between µCi/mL and CPM was CPM=82.28 ⨉ Concentration(µCi/mL) with an R² value of 0.9970. This range of activities is consistent with tumor and kidney uptake of an 800µCi ¹⁷⁷Lu injection in a mouse model. With this mapping from µCi/mL to CPM, the behavioral simulation demonstrated that the temporally varying trend of %ID/g of tumors and OAR could successfully be tracked in real time. It was shown that sensor placement can be optimized to predictably quantify and prevent excessive cross-talk between signals from different organs. <h3>Conclusion</h3> An APD-based γ-photon scintillation detector was developed and verified with a ¹⁷⁷Lu titration. We envision that the proposed work could be used to create a sensor network on the body capable of real time dosimetry of tumors and OAR.

Bayesian Inference on Hamiltonian Selections for Mössbauer Spectroscopy

M\"ossbauer spectroscopy, which provides knowledge related to electronic states in materials, has been applied to various fields such as condensed matter physics and material sciences. In conventional spectral analyses based on least-square fitting, hyperfine interactions in materials have been determined from the shape of observed spectra. In conventional spectral analyses, it is difficult to discuss the validity of the hyperfine interactions and the estimated values. We propose a spectral analysis method based on Bayesian inference for the selection of hyperfine interactions and the estimation of M\"ossbauer parameters. An appropriate Hamiltonian has been selected by comparing Bayesian free energy among possible Hamiltonians. We have estimated the M\"ossbauer parameters and evaluated their estimated values by calculating the posterior distribution of each M\"ossbauer parameter with confidence intervals. We have also discussed the accuracy of the spectral analyses to elucidate the noise intensity dependence of numerical experiments.

Climatology and process-oriented analysis of the Adriatic sea level extremes

The strongest episodes of extremely high sea levels in the Mediterranean are regularly observed in the Adriatic Sea, where they can cause substantial damage and loss of human lives. In this study, episodes of positive and negative sea level extremes were extracted from hourly series measured at six tide gauge stations located along the Adriatic coast (Venice, Trieste, Rovinj, Bakar, Split, Dubrovnik) between 1956 and 2019/2020. The time series were first checked for spurious data and then decomposed using tidal analysis, least-squares fitting and filtering procedures into (1) trend; (2) seasonal; (3) tide, (4) longer than 100 d (> 100 d), (5) 10–100 d, (6) 6 h-10 d, and (7) < 6 h components. These components correspond to sea level oscillations dominantly (but not exclusively) forced by (1) climate and isostatic change; (2) seasonal changes in thermohaline properties and circulation patterns, (3) tidal forcing, (4) quasi-stationary atmospheric and ocean circulation and climate variability patterns, (5) planetary atmospheric waves, (6) synoptic, and (7) mesoscale atmospheric processes.Significant differences exist between (1) the northern and middle/southern Adriatic extremes and (2) positive and negative extremes. The heights and return levels of positive (negative) extremes are 50–100% higher (lower) in the northern than in the middle/southern Adriatic. The northern Adriatic positive sea level extremes dominantly occur due to the superposition of the 6 h-10 d component and tide (contributing jointly to ∼70% of the total extreme height), whereas the middle/southern Adriatic positive extremes mostly occur due to the superposition of the 10–100 d component, 6 h-10 d component, and tide (each contributing ∼25% on average). The negative sea level extremes are explained as a combination of the 10–100 d component and tide: in the northern Adriatic tide provides the largest contribution (∼60%), while in the middle/southern Adriatic, the impacts of the two processes are similar (each contributing an average of ∼30%). Over the entire Adriatic, the < 6 h and seasonal components contribute the least to both positive and negative extremes. Sea level trends at all stations are positive; however, the observed sea level rise did not contribute significantly to the total height of extremes. Extreme episodes tend to occur simultaneously over larger parts of the coast and are often clustered within a few days. Both positive and negative extremes have a strong decadal variability, whereas trends of their number, duration and intensity point to shortening of negative extremes and prolonging and strengthening of positive extremes.

Fast and Stable Approximation of Analytic Functions from Equispaced Samples via Polynomial Frames

We consider approximating analytic functions on the interval [-1,1] from their values at a set of m+1 equispaced nodes. A result of Platte, Trefethen & Kuijlaars states that fast and stable approximation from equispaced samples is generally impossible. In particular, any method that converges exponentially fast must also be exponentially ill-conditioned. We prove a positive counterpart to this ‘impossibility’ theorem. Our ‘possibility’ theorem shows that there is a well-conditioned method that provides exponential decay of the error down to a finite, but user-controlled tolerance epsilon > 0, which in practice can be chosen close to machine epsilon. The method is known as polynomial frame approximation or polynomial extensions. It uses algebraic polynomials of degree n on an extended interval [-gamma ,gamma ], gamma > 1, to construct an approximation on [-1,1] via a SVD-regularized least-squares fit. A key step in the proof of our main theorem is a new result on the maximal behaviour of a polynomial of degree n on [-1,1] that is simultaneously bounded by one at a set of m+1 equispaced nodes in [-1,1] and 1/epsilon on the extended interval [-gamma ,gamma ]. We show that linear oversampling, i.e. m = c n log (1/epsilon ) / sqrt{gamma ^2-1}, is sufficient for uniform boundedness of any such polynomial on [-1,1]. This result aside, we also prove an extended impossibility theorem, which shows that such a possibility theorem (and consequently the method of polynomial frame approximation) is essentially optimal.

Evaluation of Tubular SOFC’s Performance at Elevated Pressure for Highly-Efficient Clean Power Generation

Growing global energy needs and demand over imminent reductions of harmful greenhouse gas emissions, call for development of highly efficient, clean and reliable power generation from both renewable and fossil resources. A hybrid solid oxide fuel cell gas turbine (SOFC-GT) system is a promising candidate for fulfilling such need of a low emission, highly efficient power generation. This system can be applied to stationary, transportation, aviation, and maritime power generation applications. This study focuses on understanding operational characteristics of the pressurized SOFC as a key component for the successful design and construction of a highly efficient SOFC-GT hybrid system.Pressurization of an SOFC can cause improvement on electrode kinetics, stack lifetime and round-trip efficiency by cell integration with downstream components such as gas turbines and catalytic reactors (for electrolysis) [1-4]. However, literature data from experimental studies on the electrochemical behavior of the SOFC as a function of operating pressure is relatively scarce. This is presumably due to the complexity involved in the construction and operation of pressurized system. To address such research challenge, a pressurized SOFC test rig as shown in Fig. 1 has been constructed and commissioned at Tennessee Technological University (TTU) to allow tubular SOFCs to be characterized under a maximum operating pressure of 60 psia (~4 bar) and maximum operating temperature of 825oC. A detailed system set-up shown in Fig. 2. The power generation capacity for the test rig can be scaled up to 2.4 kW. For this study, performance of single anode supported SOFC tubes with H2 and CH4 fuels were examined by varying pressure from 1 bara to 4 bara and temperature from 675oC to 775oC, with fuel utilization held constant at 50% and an oxygen-to-carbon ratio (O/C) of 0.6 for the methane operation. The testing cells were gratuitously supplied by Special Power Sources (SPS) LLC, USA, with cathode/electrolyte outside-layers supported on the anode inner-layer as a 14”-long tube. Silver ink and wires are used for cell current collection from active surface area of ⁓100 cm2 . The connection fixture can be seen in Fig. 1. V-I curves of tubular SOFCs in Fig. 3 measured from step-wise electronic load profile indicate total performance gain of 13% as H2 pressure increases from 1 bara to 4 bara, with highest peak power density obtained at 0.61 Wcm-2. As comparison Fig.4 and 5 demonstrate pressurization suppressed gas diffusion limit for tubular SOFC at higher current density in CH4 fuel, leading to 44 % increase of peak power density to 0.51 Wcm-2 at 4 bara.Electrochemical impedance spectroscopy (EIS) studies were performed using a Solartron 1260/PAR 263 electrochemical testing system in 4-point connection measurement to understand effects of pressurization on the SOFC performance. SOFC impedance spectra as Nyquist and Bode plots in Fig.6 (b) and (d) indicate the tubular SOFC performance is mainly limited by anode mass transports, which are greatly improved by pressurization. From non-linear, least-square cell impedance fitting and previous findings on impedance of similar larger cells, an equivalent-circuit model (ECM) in Fig. 6(a) has been identified to best represent impedance responses from the most rate-limiting processes of tubular SOFCs. The Warburg component (Ws) in the ECM with resonance frequency around 10Hz typically corresponds to gas diffusion in porous electrode of tubular SOFC [3-5], while paralleled R-CPE component at <1Hz resonance frequency can be attributed to gas-conversion impedance (GCI) due to gas phases concentration gradient along fuel-flow channel resulting from reactant conversion [5,6]. Experimentally observed changes of such impedances with fuel compositions, flow rates and tubular SOFC configuration are generally in agreement with theoretical assumptions and other reports [4-6]. Overall, results of this work demonstrate pressurization can effectively boost anode-supported SOFCs’ performance by overcoming fuel-side mass-transport limit associated with practical cell configuration and testing geometry. The experimental performance data collected can then be used to fit key constants and estimate parameters for numerical modeling of the SOFC system performance at elevated pressures.References can be seen in the figure document. Figure 1

Defects localization using the data fusion of laser Doppler and image correlation vibration measurements

This work proposes a defects localization method based on the combination of laser Doppler vibrometry (LDV) and Digital image correlation (DIC). DIC and LDV are the dominant non-contact vibration measurement methods with a matured theory and high stability. DIC gives a high spatial resolution with a measurement accuracy of several microns, whereas the Multipoint LDV has a limited spatial resolution with high-precision results for the sparse points up to nanometer level. However, the vibration-based defect localization (VBDL) necessitates certain measurement methods with both the high spatial resolution and measurement accuracy. Unfortunately, the accuracy of DIC is not sufficient to detect the data anomalies caused by the structural defects. Nevertheless, LDV cannot achieve a high-precision defect localization owing to the limited spatial resolution. In this study, a nonlinear weighted least-squares fitting has been proposed to improve the measurement accuracy and reliability. A Chebyshev polynomial has been employed as the basis function in the fitting process. The effective signal-to-noise ratio (SNR) and the correlation coefficient are employed as the weight basis of the LDV points and the DIC subsets, respectively. Based on the simulation analysis, the determination of the optimal polynomial order at multiple noise levels and the influence of the LDV points distribution are discussed. The experiment proves that the data fusion method can significantly improve spatial resolution and measurement accuracy. From the data fusion results, the reference-free defects localization in cantilever beams and plates based on residuals have been presented.

Quantification of motion characteristics of vertically ascending bubbles in NaCl solution via image processing

Abstract In this study, the motion characteristics of bubbles in a vertical rectangular glass groove were investigated using image-processing techniques. A new metric (bubble offset) to evaluate the level of solution disturbance was proposed. The motion characteristics, such as the trajectory, bubble diameter, rise velocity, and coalescence, during the bubble rise were studied using different NaCl concentrations and airflow rates. The functional relationship of the bubble diameter and velocity with height was obtained by least-squares fitting (correlation coefficient R2 ≥ 95%), and fits the power-law distribution. The results showed that the bubble motion characteristics were positively correlated with the airflow rate and negatively correlated with the NaCl concentration in the solution. The solution disturbance was closely related to the bubble offset, and the size of the bubble offset can be used to characterize the degree of the solution disturbance. The disturbance of the solution affects the bubble trajectory, rising velocity and coalescence. The bubble offset changed periodically with height, and its functional relationship was obtained. The accuracy of the functional relationship was verified using experimental data. The study of bubble motion characteristics is crucial for understanding phenomena such as heat and mass transfer and multiphase cooling.

Personalized Driver Braking Behavior Modeling in the Car-Following Scenario: An Importance-Weight-Based Transfer Learning Approach

Accurately recognizing braking intensity levels (BIL) of drivers is important for guaranteeing the safety and avoiding traffic accidents in intelligent transportation systems. In this article, an instance-level transfer learning framework is proposed to recognize BIL for a new driver with insufficient driving data by combining the Gaussian mixture model (GMM) and the importance-weighted least-squares probabilistic classifier (IWLSPC). By considering the statistic distribution, GMM is applied to cluster the data of braking behaviors into three levels with different intensities. With the density ratio calculated by unconstrained least-squares importance fitting, the least-squares probabilistic classifier is modified as IWLSPC to transfer the knowledge from one driver to another and recognize BIL for a new driver with insufficient driving data. Comparative experiments with nontransfer methods indicate that the proposed framework obtains a higher accuracy in recognizing BIL in the car-following scenario, especially when sufficient data are not available.

A Ring Gyroscope With On-Chip Capacitive Stress Compensation

We present long-term stress compensation results for a 3.2mm diameter ring gyroscope integrated with 16 capacitive stress sensors for the first time in this work. A bridge-type capacitive sensor is preferred due to its compact size and temperature insensitivity for on-chip stress measurements. The ring design enables a high level of integration and stress sensor-gyroscope output correlation. We first demonstrate the stress sensor operation on a stress test-bed. The drift test for sixteen days at mismatched mode and the drift test for eight days at matched mode in room temperature reveal that the stress compensation can eliminate the gyroscope drift. The stability of the stress compensated gyroscope output can reach 0.008°/h in mismatched mode and 0.003°/h in matched mode at an averaging time of two days with no signs of long-term drift. High gyroscope stability is achieved with a partial least-squares fitting algorithm; however, we believe that stress and gyroscope output relation might be linear time-variant with possible nonlinear and hysteresis effects. Analysis of the drive and sense mode frequencies shows that only temperature cannot explain the frequency variations, and the inclusion of stress can comprehensively describe the frequency changes. [2022-0108]

Nusselt number for steady periodically developed heat transfer in micro- and mini-channels with arrays of offset strip fins subject to a uniform heat flux

In this work, the Nusselt number is examined for periodically developed heat transfer in micro- and mini-channels with arrays of offset strip fins, subject to a constant heat flux. The Nusselt number is defined on the basis of a heat transfer coefficient which represents the spatially constant macro-scale temperature difference between the fluid and solid during conjugate heat transfer. Its values are determined numerically on a single unit cell of the array for Reynolds numbers between 1 and 600 and fin height-to-length ratios below 1. Two combinations of the Prandtl number and the thermal conductivity ratio are selected, corresponding to air and water. It is shown that the Nusselt number correlations from the literature mainly apply to air in the transitional flow regime in larger conventional channels if the wall temperature remains uniform. As a result, they do not correctly capture the observed trends for the Nusselt number in micro- and mini-channels subject to a constant heat flux. Therefore, new Nusselt number correlations, obtained through a least-squares fitting of 2282 numerical simulations, are presented for air and water. The suitability of these correlations is assessed via the Bayesian approach for parameter estimation and model validation. The correlations respect the observed asymptotic trends and limits of the Nusselt number for all the geometrical parameters of the offset strip fins. In addition, they predict a linear dependence of the Nusselt number on the Reynolds number, in good agreement with the data from this work. Nevertheless, a detailed analysis reveals a more complex scaling of the Nusselt number with the Reynolds number, closely related to the underlying flow regimes, particularly the weak and strong inertia regimes. Finally, through 62 additional simulations, the influence of the material properties on the Nusselt number is illustrated and compared to the available literature.

Modelling and prediction of GNSS time series using GBDT, LSTM and SVM machine learning approaches

Global navigation satellite system (GNSS) site coordinate time series provides essential data for geodynamic and geophysical studies, realisation of a regional or global geodetic reference frames, and crustal deformation research. The coordinate time series has been conventionally modelled by least squares (LS) fitting with harmonic functions, alongside many other analysis methods. As a key limitation, the traditional modelling approaches simply use the functions of time variable, despite good knowledge of various underlying physical mechanisms responsible for the site displacements. This paper examines the use of machine learning (ML) models to reflect the effects or residential effects of physical variables related to Sun and the Moon ephemerides, polar motion, temperature, atmospheric pressure, and hydrology on the site displacements. To form the ML problem, these variables are constructed as the input vector of each ML training sample, while the vertical displacement of a GNSS site is regarded as the output value. In the evaluation experiments, three ML approaches, namely the gradient boosting decision tree (GBDT) approach, long short-term memory (LSTM) approach, and support vector machine (SVM) approach, are introduced and evaluated with the time series datasets collected from 9 GNSS sites over the period of 13 years. The results indicate that all three approaches achieve similar fitting precision in the range of 3–5 mm in the vertical displacement component, which is an improvement in over 30% with respect to the traditional LS fitting precision in the range of 4–7 mm. The prediction of the vertical time series with the three ML approaches shows the precision in the range of 4–7 mm over the future 24- month period. The results also indicate the relative importance of different physical features causing the displacements of each site. Overall, ML approaches demonstrate better performance and effectiveness in modelling and prediction of GNSS time series, thus impacting maintenance of geodetic reference frames, geodynamics, geophysics, and crustal deformation analysis.

Bayesian neural networks enabled snapshot polarized light scattering spectroscopy with uncertainty quantification

Polarized light-scattering spectroscopy (PLSS) is a promising noninvasive early cancer detection technique via inversing the nuclear size of epithelial cells from collected single scattered light. Conventionally, the nuclear size is inferred by using model-driven least square fitting (LSF) methods, which is time-consuming and challenging for real-time cancer diagnosis. Although data-driven deep-learning-based techniques can be employed to speed up the inverse process, the labeled data are usually limited and the reliability quantification of prediction is lacking. Herein, by synthesizing a large quantity of single scattered spectra with physical priori, a Bayesian-deep-learning (BDL)-based PLSS framework is presented for the first time as far as we know, and is expected to provide model uncertainty of network prediction for the reliability quantification. Further, to explore the applicability of the BDL uncertain estimation, the neural networks for early cancer diagnosis are designed as a convolutional neural network (CNN) classification problem and a fully connected neural network (FNN) regression problem, respectively. They are verified using the single scattered spectra of polystyrene microsphere tissue models and human colorectal lesion tissue obtained by our proposed snapshot PLSS system. The results show that both BDL networks obtain similar nuclear sizes to the ground truth and their inferring speeds are far faster than LSF methods. Further, it is interesting to note that only the CNN classification network can provide effective uncertainty for reliability quantification. The work presents a new paradigm for the real-time diagnosis of early cancer and provides a reference for the design of snapshot PLSS endoscopy.

In vivo magnetic resonance P-Spectral Analysis With Neural Networks: 31P-SPAWNN.

We have introduced an artificial intelligence framework, 31P-SPAWNN, in order to fully analyze phosphorus-31 ( P) magnetic resonance spectra. The flexibility and speed of the technique rival traditional least-square fitting methods, with the performance of the two approaches, are compared in this work. Convolutional neural network architectures have been proposed for the analysis and quantification of P-spectroscopy. The generation of training and test data using a fully parameterized model is presented herein. In vivo unlocalized free induction decay and three-dimensional P-magnetic resonance spectroscopy imaging data were acquired from healthy volunteers before being quantified using either 31P-SPAWNN or traditional least-square fitting techniques. The presented experiment has demonstrated both the reliability and accuracy of 31P-SPAWNN for estimating metabolite concentrations and spectral parameters. Simulated test data showed improved quantification using 31P-SPAWNN compared with LCModel. In vivo data analysis revealed higher accuracy at low signal-to-noise ratio using 31P-SPAWNN, yet with equivalent precision. Processing time using 31P-SPAWNN can be further shortened up to two orders of magnitude. The accuracy, reliability, and computational speed of the method open new perspectives for integrating these applications in a clinical setting.

Underground transient electromagnetic real-time imaging system for coal mine water disasters

Due to advantageous sensitivity to low resistance and noncomplicated operation, transient electromagnetic methods are the most widely used geophysical methods for advanced detection of coal-mining-disaster water bodies. To realize real-time imaging, we developed a wireless antenna attitude angle measuring instrument based on an electronic compass, which measures antenna attitude angle real-time. According to attitude angle information sent through WiFi to the host of transient electromagnetic real-time imager instruments, current transmission, signal acquisition, and antenna rotation angle information are achieved for real-time imaging. We studied the direct algorithm of the apparent resistivity of the whole-space transient electromagnetic field in overlapping loop mode based on double-logarithmic-transform piecewise least-squares fitting, which not only ensures accuracy but also accelerates calculation. On-site detection was carried out in a coal mine, and the detection results are generally consistent with the drilling results.

Fitting second-order cone constraints to microbial growth data

Second-order cone programming is a highly tractable convex optimization class. In this paper, we fit general second-order cone constraints to data. This is of use when one must solve large-scale, nonlinear optimization problems, but modeling is either impractical or does not lead to second-order cone or otherwise tractable constraints. Our motivating application is biochemical process optimization, in which we seek to fit second-order cone constraints to microbial growth data. The fitting problem is nonconvex. We solve it using the concave–convex procedure, which takes the form of a sequence of second-order cone programs. We validate our approach on simulated and experimental microbial growth data, and compare its performance with conventional nonlinear least-squares fitting.