Neural Components Research Articles

Longitudinal development of muscle strength and relationship with motor unit activity and muscle morphological characteristics in youth athletes.

Neural and morphological adaptations determine gains of muscle strength. For youth athletes, the importance of morphological adaptation is typically highlighted based on the change in maturity status. However, the long-term development of neural components in youth athletes remains unclear. The present study investigated the longitudinal development of muscle strength, muscle thickness (MT), and motor unit firing activity of the knee extensor and their relationships in youth athletes. Seventy male youth soccer players (mean ± SD age = 16.3 ± 0.6years) performed neuromuscular, maximal voluntary isometric contraction (MVC), and submaximal ramp contraction (at 30 and 50% MVC) tests with knee extensors, two times with a 10-month measurement interval. High-density surface electromyography was recorded from the vastus lateralis and decomposed to identify each individual motor unit activity. MT was evaluated by the sum of the vastus lateralis and vastus intermedius thicknesses. Finally, sixty-four participants were employed to compare MVC and MT, and 26 participants were employed to analyze motor unit activity. MVC and MT were increased from pre to post (p < 0.05, 6.9 and 1.7% for MVC and MT, respectively). Y-intercept of the regression line between median firing rate vs. recruitment threshold was also increased (p < 0.05, 13.3%). Multiple regression analysis demonstrated that the gains of both MT and Y-intercept were explanatory variables for the gain of strength. These findings suggest that the neural adaptation could also make the important contribution to the strength gain for the youth athletes over a 10-month training period.

Longitudinal transcriptomic analysis of mouse sciatic nerve reveals pathways associated with age-related muscle pathology.

Sarcopenia, the age-associated decline in skeletal muscle mass and strength, has long been considered a disease of muscle only, but accumulating evidence suggests that sarcopenia could originate from the neural components controlling muscles. To identify early molecular changes in nerves that may drive sarcopenia initiation, we performed a longitudinal transcriptomic analysis of the sciatic nerve, which governs lower limb muscles, in aging mice. Sciatic nerve and gastrocnemius muscle were obtained from female C57BL/6JN mice aged 5, 18, 21 and 24months old (n=6 per age group). Sciatic nerve RNA was extracted and underwent RNA sequencing (RNA-seq). Differentially expressed genes (DEGs) were validated using quantitative reverse transcription PCR (qRT-PCR). Functional enrichment analysis of clusters of genes associated with patterns of gene expression across age groups (adjusted P-value<0.05, likelihood ratio test [LRT]) was performed. Pathological skeletal muscle aging was confirmed between 21 and 24months by a combination of molecular and pathological biomarkers. Myofiber denervation was confirmed with qRT-PCR of Chrnd, Chrng, Myog, Runx1 and Gadd45ɑ in gastrocnemius muscle. Changes in muscle mass, cross-sectional myofiber size and percentage of fibres with centralized nuclei were analysed in a separate cohort of mice from the same colony (n=4-6 per age group). We detected 51 significant DEGs in sciatic nerve of 18-month-old mice compared with 5-month-old mice (absolute value of fold change>2; false discovery rate [FDR]<0.05). Up-regulated DEGs included Dbp (log2 fold change [LFC]=2.63, FDR<0.001) and Lmod2 (LFC=7.52, FDR=0.001). Down-regulated DEGs included Cdh6 (LFC=-21.38, FDR<0.001) and Gbp1 (LFC=-21.78, FDR<0.001). We validated RNA-seq findings with qRT-PCR of various up- and down-regulated genes including Dbp and Cdh6. Up-regulated genes (FDR<0.1) were associated with the AMP-activated protein kinase signalling pathway (FDR=0.02) and circadian rhythm (FDR=0.02), whereas down-regulated DEGs were associated with biosynthesis and metabolic pathways (FDR<0.05). We identified seven significant clusters of genes (FDR<0.05, LRT) with similar expression patterns across groups. Functional enrichment analysis of these clusters revealed biological processes that may be implicated in age-related changes in skeletal muscles and/or sarcopenia initiation including extracellular matrix organization and an immune response (FDR<0.05). Gene expression changes in mouse peripheral nerve were detected prior to disturbances in myofiber innervation and sarcopenia onset. These early molecular changes we report shed a new light on biological processes that may be implicated in sarcopenia initiation and pathogenesis. Future studies are warranted to confirm the disease modifying and/or biomarker potential of the key changes we report here.

The multiple roles of GH in neural ageing and injury.

Advanced age is typically associated with a decrease in cognitive function including impairment in the formation and retention of new memories. The hippocampus is critical for learning and memory, especially spatial learning, and is particularly affected by ageing. With advanced age, multiple neural components can be detrimentally affected including a reduction in the number of neural stem and precursor cells, a decrease in the formation of adult born neurons (neurogenesis), and deficits in neural circuitry, all of which ultimately contribute to impaired cognitive function. Importantly, physical exercise has been shown to ameliorate many of these impairments and is able to improve learning and memory. Relevantly, growth hormone (GH) is an important protein hormone that decreases with ageing and increases following physical exercise. Originally described due to its role in longitudinal growth, GH has now been identified to play several additional key roles, especially in relation to the brain. Indeed, the regular decrease in GH levels following puberty is one of the most well documented components of neuroendocrine ageing. Growth hormone deficiency (GHD) has been described to have adverse effects on brain function, which can be ameliorated via GH replacement therapy. Physical exercise has been shown to increase circulating GH levels. Furthermore, we recently demonstrated the increase in exercise-mediated GH is critical for improved cognitive function in the aged mouse. Here we examine the multiple roles that GH plays, particularly in the aged brain and following trauma, irradiation and stroke, and how increasing GH levels can ameliorate deficits in cognition.

Neural mechanisms for emotional contagion and spontaneous mimicry of live facial expressions.

Viewing a live facial expression typically elicits a similar expression by the observer (facial mimicry) that is associated with a concordant emotional experience (emotional contagion). The model of embodied emotion proposes that emotional contagion and facial mimicry are functionally linked although the neural underpinnings are not known. To address this knowledge gap, we employed a live two-person paradigm (n = 20 dyads) using functional near-infrared spectroscopy during live emotive face-processing while also measuring eye-tracking, facial classifications and ratings of emotion. One dyadic partner, 'Movie Watcher', was instructed to emote natural facial expressions while viewing evocative short movie clips. The other dyadic partner, 'Face Watcher', viewed the Movie Watcher's face. Task and rest blocks were implemented by timed epochs of clear and opaque glass that separated partners. Dyadic roles were alternated during the experiment. Mean cross-partner correlations of facial expressions (r = 0.36 ± 0.11 s.e.m.) and mean cross-partner affect ratings (r = 0.67 ± 0.04) were consistent with facial mimicry and emotional contagion, respectively. Neural correlates of emotional contagion based on covariates of partner affect ratings included angular and supramarginal gyri, whereas neural correlates of the live facial action units included motor cortex and ventral face-processing areas. Findings suggest distinct neural components for facial mimicry and emotional contagion. This article is part of a discussion meeting issue 'Face2face: advancing the science of social interaction'.

Validity, Intra-Rater Reliability and Normative Data of the Neuroflexor™ Device to Measure Spasticity of the Ankle Plantar Flexors after Stroke.

Quantification of lower limb spasticity after stroke and the differentiation of neural from passive muscle resistance remain key clinical challenges. The aim of this study was to validate the novel NeuroFlexor foot module, to assess the intrarater reliability of measurements and to identify normative cut-off values. Fifteen patients with chronic stroke with clinical history of spasticity and 18 healthy subjects were examined with the NeuroFlexor foot module at controlled velocities. Elastic, viscous and neural components of passive dorsiflexion resistance were quantified (in Newton, N). The neural component, reflecting stretch reflex mediated resistance, was validated against electromyography activity. A test-retest design with a 2-way random effects model permitted study of intra-rater reliability. Finally, data from 73 healthy subjects were used to establish cutoff values according to mean + 3 standard deviations and receiver operating characteristic curve analysis. The neural component was higher in stroke patients, increased with stretch velocity and correlated with electromyography amplitude. Reliability was high for the neural component (intraclass correlation coefficient model 2.1 (ICC2,1) ≥ 0.903) and good for the elastic component (ICC2,1 ≥ 0.898). Cutoff values were identified, and all patients with neural component above the limit presented pathological electromyography amplitude (area under the curve (AUC) = 1.00, sensitivity = 100%, specificity = 100%). The NeuroFlexor may offer a clinically feasible and non-invasive way to objectively quantify lower limb spasticity.

Simulation of memristive crossbar arrays for seizure detection and prediction using parallel Convolutional Neural Networks

<h2>Abstract</h2> For epileptic seizure detection and prediction, to address the computational bottleneck of the von Neumann architecture, we develop an in-memory memristive crossbar-based accelerator simulator. The simulator software is composed of a Python-based neural network training component and a MATLAB-based memristive crossbar array component. The software provides a baseline network for developing deep learning-based signal processing tasks, as well as a platform to investigate the impact of weight mapping schemes and device and peripheral circuitry non-idealities.

Disentangling the Skeins of Brain.

Some have argued that the brain is so complex that it cannot be understood using current reductive approaches. Drawing on examples from decision neuroscience, we instead contend that combining new neuroscientific techniques with reductive approaches that consider central brain components in time and space has generated significant progress over the past 2 decades. This progress has allowed researchers to advance from the scientific goals of description and explanation to prediction and control. Resulting knowledge promises to improve human health and well-being. As an alternative to the extremes of reductive versus emergent approaches, however, we propose a middle way of "expansion." This expansionist approach promises to leverage the specific spatial localization, temporal precision, and directed connectivity of central neural components to ultimately link levels of analysis.

Regional recurrence of malignant peripheral nerve sheath tumor in upper extremity – a rare case report

Introduction: Malignant Peripheral Nerve Sheath Tumor (MPNST) are aggressive soft tissue sarcoma (STS), originally arising from neural cell components. MPNST had a relatively high local recurrence rate of 22% and a disease survival rate of 45 - 60%. In this study, we would like to report a rare case of recurrent MPNST affecting the upper arm region, with a history of previously resected MPNST at the patient's ipsilateral elbow. Case Presentation: A 38-years old male patient complained of an enlarging mass in his left anterior axilla Patient had a history of left elbow mass 2 years before and had resected twice. Eight months afterward, a new tumor grew from his left axilla. From the physical examination, the mass was fixated and firm with an oval shape, size 8 x 6 cm. From chest CT scan examination, it appeared a soft tissue mass surrounded with stranding fat had infiltrated his left teres major muscle. The mass was removed successfully and the pathology result revealed a grade 2 MPNST. Post-operative radiotherapy was then scheduled. Among many STS, MPNSTs are notorious for having a high recurrence rate. NF1 mutation, tumors deeper than fascia, large tumors, high-grade tumors, and tumors arising from the trunk, head, and neck are related to poor prognosis in MPNST. Surgeons' inadequate knowledge, unplanned resections, and their dependency on radiotherapy are other iatrogenic factors for recurrences. Conclusion: MPNST has high recurrent risk, especially in high-grade tumors, large or deep tumors, surgery with positive resection margins, and NF1 mutations. The primary option for treatment is surgery and radiation therapy.

Targeting the Receptor at Synapse of Keratinocyte and Peripheral Nerve Ending: A Therapeutic Hope for Post-Burn Pruritus

Chronic itch or Pruritus is an unpleasant sensation that provokes the desire to scratch. In post burn patients, it is a significant health burden with few effective treatments. Pruritus is a common distressing consequence of postburn scars which affects burn survivors’ quality of life by causing sleep disturbance, daily activity impairment, and psychological problems. The mechanism of such abnormal scars with pruritus is not yet clear. The field of pruritus research is a dynamic field which involves neural component and various receptors at skin. The receptor present at the synapse of keratinocytes and peripheral nerve ending are responsible for itching. Many recent studies have worked to define the various receptors and the itch mediations in skin. However, considering that post-burn patients mainly report itch to be localized in the burn scars, spontaneous itch might be strongly based on peripheral input. The detailed characteristic and the factors for persistence of pruritus in post burn patients have not yet been clearly understood. Experimental evidences suggest that transient receptor potential (TRP) channels in skin and their expression patterns, activation mechanisms play regulatory roles in pruritus pathogenesis. In this article author collected the pruritus related data from post burn pruritus patients, isolated keratinocytes checked the receptor expression by immunocytochemistry and discussed the complex interplay between the central and peripheral factors that causes pruritus. It would be helpful to identify novel anti-pruritic agents that target the molecular itch pathogenesis pathways and the clinical assessment for burn patients with pruritus.

End-to-End Implementation of a Convolutional Neural Network on a 3D-Integrated Image Sensor with Macropixel Array.

Three-dimensional-integrated focal-plane array image processor chips offer new opportunities to implement highly parallelised computer vision algorithms directly inside sensors. Neural networks in particular can perform highly complex machine vision tasks, and therefore their efficient implementation in such imagers are of significant interest. However, studies with existing pixel-processor array chips have focused on the implementation of a subset of neural network components-notably convolutional kernels-on pixel processor arrays. In this work, we implement a continuous end-to-end pipeline for a convolutional neural network from the digitisation of incoming photons to the output prediction vector on a macropixel processor array chip (where a single processor acts on group of pixels). Our implementation performs inference at a rate between 265 and 309 frames per second, directly inside of the sensor, by exploiting the different levels of parallelism available.

Nerves within bone and their application in tissue engineering of bone regeneration.

Nerves within bone play an irreplaceable role in promoting bone regeneration. Crosstalk between the nerve system and bone has arisen to the attention of researchers in the field of basic medicine, clinical medicine, and biomaterials science. Successful bone regeneration relies on the appropriate participation of neural system components including nerve fibers, signaling molecules, and neural-related cells. Furthermore, more about the mechanisms through which nerves took part in bone regeneration and how these mechanisms could be integrated into tissue engineering scaffolds were under exploration. In the present review, we aimed to systematically elaborate on the structural and functional interrelationship between the nerve system and bone. In particular, peripheral nerves interact with the bone through innervated axons, multiple neurotrophins, and bone resident cells. Also, we aimed to summarize research that took advantage of the neuro-osteogenic network to design tissue engineering scaffolds for bone repair.

An enhancement of transformer-based architecture with randomized regularization for wind speed prediction

The wind power is considered as a potential renewable energy resource which requires less management cost and effort than the others like as tidal, geothermal, etc. However, the natural randomization and volatility aspects of wind in different regions have brought several challenges for efficiently as well as reliably operating the wind-based power supply grid. Thus, it is necessary to have centralized monitoring centers for managing as well as optimizing the performance of wind power farms. Among different management task, wind speed prediction is considered as an important task which directly support for further wind-based power supply resource planning/optimization, hence towards power shortage risk and operating cost reductions. Normally, considering as traditional time-series based prediction problem, most of previous deep learning-based models have demonstrated significant improvement in accuracy performance of wind speed prediction problem. However, most of recurrent neural network (RNN) as well as sequential auto-encoding (AE) based architectures still suffered several limitations related to the capability of sufficient preserving the spatiotemporal and long-range time dependent information of complex time-series based wind datasets. Moreover, previous RNN-based wind speed predictive models also perform poor prediction results within high-complex/noised time-series based wind speed datasets. Thus, in order to overcome these limitations, in this paper we proposed a novel integrated convolutional neural network (CNN)-based spatiotemporal randomization mechanism with transformer-based architecture for wind speed prediction problem, called as: RTrans-WP. Within our RTrans-WP model, we integrated the deep neural encoding component with a randomized CNN learning mechanism to softy align temporal feature within the long-range time-dependent learning context. The utilization of randomized CNN component at the data encoding part also enables to reduce noises and time-series based observation uncertainties which are occurred during the data representation learning and wind speed prediction-driven fine-tuning processes.

Early detection of neurological abnormalities using a combined phase space reconstruction and deep learning approach

The scientific literature on depression detection using electroencephalogram (EEG) signals is extensive and offers numerous innovative approaches. However, these existing state-of-the-art (SOTA) have limitations that hinder their overall efficacy. They rely significantly on datasets with limited scope and accessibility, which introduces potential biases and diminishes generalizability. In addition, they concentrate on analyzing a single dataset, potentially overlooking the inherent variability and complexity of EEG patterns associated with depression. Moreover, certain SOTA methods employ deep learning architectures with exponential time complexity, resulting in computationally intensive and time-consuming training procedures. Therefore, their practicability and applicability in real-world scenarios are compromised. To address these limitations, a novel integrated methodology that combines the advantages of phase space reconstruction and deep neural networks is proposed. It employs publicly available EEG datasets, mitigating the inherent biases of exclusive data sources. Moreover, the method incorporates reconstructed phase space analysis, a feature engineering technique that captures more accurately the complex EEG patterns associated with depression. Simultaneously, the incorporation of a deep neural network component guarantees optimal efficiency and accurate, seamless classification. Using publicly available datasets, cross-dataset validation, and a novel combination of reconstructed phase space analysis and deep neural networks, the proposed method circumvents the shortcomings of current state-of-the-art (SOTA) approaches. This innovation represents a significant advance in enhancing the accuracy of depression detection and provides the base for EEG-based depression assessment tools applicable to real-world settings. The findings of the study provide a more robust and efficient model, which increases classification precision and decreases computing burden. The study findings layout the foundation for scalable, accessible mental health solutions, identification of the pathological deficits in affected brain tissues, and demonstrate the potential of technology-driven approaches to support and guide depressed individuals and enhance mental health outcomes.

Multimodal Data Matters: Language Model Pre-Training Over Structured and Unstructured Electronic Health Records.

As two important textual modalities in electronic health records (EHR), both structured data (clinical codes) and unstructured data (clinical narratives) have recently been increasingly applied to the healthcare domain. Most existing EHR-oriented studies, however, either focus on a particular modality or integrate data from different modalities in a straightforward manner, which usually treats structured and unstructured data as two independent sources of information about patient admission and ignore the intrinsic interactions between them. In fact, the two modalities are documented during the same encounter where structured data inform the documentation of unstructured data and vice versa. In this paper, we proposed a Medical Multimodal Pre-trained Language Model, named MedM-PLM, to learn enhanced EHR representations over structured and unstructured data and explore the interaction of two modalities. In MedM-PLM, two Transformer-based neural network components are firstly adopted to learn representative characteristics from each modality. A cross-modal module is then introduced to model their interactions. We pre-trained MedM-PLM on the MIMIC-III dataset and verified the effectiveness of the model on three downstream clinical tasks, i.e., medication recommendation, 30-day readmission prediction and ICD coding. Extensive experiments demonstrate the power of MedM-PLM compared with state-of-the-art methods. Further analyses and visualizations show the robustness of our model, which could potentially provide more comprehensive interpretations for clinical decision-making.

Algorithms for Estimating Time-Locked Neural Response Components in Cortical Processing of Continuous Speech.

The Temporal Response Function (TRF) is a linear model of neural activity time-locked to continuous stimuli, including continuous speech. TRFs based on speech envelopes typically have distinct components that have provided remarkable insights into the cortical processing of speech. However, current methods may lead to less than reliable estimates of single-subject TRF components. Here, we compare two established methods, in TRF component estimation, and also propose novel algorithms that utilize prior knowledge of these components, bypassing the full TRF estimation. We compared two established algorithms, ridge and boosting, and two novel algorithms based on Subspace Pursuit (SP) and Expectation Maximization (EM), which directly estimate TRF components given plausible assumptions regarding component characteristics. Single-channel, multi-channel, and source-localized TRFs were fit on simulations and real magnetoencephalographic data. Performance metrics included model fit and component estimation accuracy. Boosting and ridge have comparable performance in component estimation. The novel algorithms outperformed the others in simulations, but not on real data, possibly due to the plausible assumptions not actually being met. Ridge had slightly better model fits on real data compared to boosting, but also more spurious TRF activity. Results indicate that both smooth (ridge) and sparse (boosting) algorithms perform comparably at TRF component estimation. The SP and EM algorithms may be accurate, but rely on assumptions of component characteristics. This systematic comparison establishes the suitability of widely used and novel algorithms for estimating robust TRF components, which is essential for improved subject-specific investigations into the cortical processing of speech.

РОЛЬ ШВАННОВСКИХ КЛЕТОК В ПАТОЛОГИИ ЧЕРВЕОБРАЗНОГО ОТРОСТКА У ДЕТЕЙ

Objective: To compare the immunohistochemical (IHC) characteristics of appendices removed from children with acute appendicitis and chronic pain syndrome (CP) in the right lower quadrant of the abdomen (RLQA), as well as to define clinical and morphological correlations. Methods: The structure of fifty-one appendices of children aged 5-14 years who underwent appendectomy for chronic appendicitis/CP in the RLQA (24 patients, Group 1) and acute appendicitis (27 patients, Group 2) was assessed. Image analysis of histological sections stained with hematoxylin-eosin and immunohistochemically for S100 protein and proliferating cell nuclear antigen (PCNA) was performed. Results: Image analysis showed the presence of a significantly higher volume density (VD) and numerical density (ND) (p0.05) of immunoreactive cells when stained for S100 protein, as well as the ND of intramural ganglia (p0.05) against the background of minimal histological changes in the wall of the appendix in patients of Group 1 compared to Group 2. On the contrary, the ND of PCNA-immunoreactive cells was significantly higher in patients of Group 2 against the background of pronounced inflammatory changes in the appendix (p0.01). In the biopsy samples of the Group 1 patients, a significantly higher ND of the lymphoid nodule germinal centers was also noted (p0.05), while the ND of tingible body macrophages was higher in Group 2 patients (p0.001). These results demonstrate differences in neuroimmune parameters in the appendices in the two described nosologic groups and indicate the possible role of the appendiceal neural component as a prerequisite for developing CP in the RLQA in children. Keywords: Appendix, Schwann cell, S100 protein, PCNA, appendicitis, chronic abdominal syndrome.

An Index for Rail Weld Health Assessment in Urban Metro Using In-Service Train

Rail welds are considered as the weak part of a railway track. Their defects and health can directly affect wheel-rail interaction, track safety, and reliability. Current practices for rail welds health assessment are based on 2D vertical and lateral wear measurement which needs time and track blocking. The development of inertia-based condition monitoring methods such as measuring axle box acceleration (ABA) comes with a crucial question on criteria or index for each rail track component health monitoring. In this study, an index for evaluation of rail weld health is proposed through integrated numerical and field experiment data within a metro line using the ABA technique. The relationship between the speed, wheel structural vibration, and acceleration amplitude is investigated using fast Fourier transformation (FFT) and a nonlinear neural network principal component analysis (PCA) model. An index is introduced to assess weld severity level based on the statistical method. This index is simple and applicable for maintenance practice.

Locally controlled release of immunosuppressive promotes survival of transplanted adult spinal cord tissue.

Transplantation of adult spinal cord tissue (aSCT) is a promising treatment for spinal cord injury (SCI) basing on various types of neural cells and matrix components inside aSCT. However, long-term systemic administration of immunosuppressors (e.g. tacrolimus, TAC) is required for the survival of allogeneic tissue, which often associated with severe side effects such as infection, liver damageand renal failure. In this study, a triglycerol monostearate (TGM)-based TAC delivery system (e.g. TAC@TGM) with high drug loading concentration was developed, which possessed injectable properties as well as sustainable and immune-responsive drug release behaviors. In complete transected SCI model, locally injected TAC@TGM could reduce the infiltration of inflammation cells, enhance the survival of transplanted aSCT (e.g. Tuj-1+ and NF+ neurons) and promote the recovery of locomotor function. Moreover, controlled release of TAC by TAC@TGM attenuated side effects of TAC on liver and kidneys compared with traditional systemic administration. More importantly, the developed TAC@TGM system provided a facile single dose of long-term immunosuppressive effect not just for aSCT transplantation, but also for other tissue/organ and cell transplantations.

Du-Bus: A Realtime Bus Waiting Time Estimation System Based On Multi-Source Data

Realtime bus waiting time information is of great importance to the intelligent public transportation system and is beneficial for improving user satisfaction by online map services. While there are limited realtime bus waiting time services in a city, because of the expensive cost of sensor deployment and sophisticated traffic conditions. To address the above problem, we propose Du-Bus, a multi-source data fusion based system, which estimates the realtime bus waiting time based on approximating the realtime locations of buses without GPS sensors, by a variety of urban datasets, including historical bus trip data reported by a limited number of GPS equipped buses, transportation network data, traffic condition data, user mobility data, and temporal data. Du-Bus approximates the realtime locations of buses without GPS sensors by jointly modeling the bus timetable and the bus realtime travel time, which can be estimated by a variety of data sources. Specifically, we first propose a BiLSTM based end-to-end model for each bus route to estimate the bus departure interval and generate the corresponding departure timetable. Then, we estimate the travel time for each individual bus via a deep neural network component by incorporating the traffic conditions, geolocation, and map query information. Finally, we estimate the bus waiting time for arbitrary stations in the city by jointly modeling the estimated bus departure timetable and travel time. We evaluate our system on two real-world datasets, and the results verify the effectiveness of Du-Bus compared with historical average based and headway based methods. Since early 2019, Du-Bus has been deployed on Baidu Maps, one of the world’s largest map services, servicing over 20 major cities in China.

Layover Detection Using Neural Network Based on Expert Knowledge

Layover detection is crucial in 3D array SAR topography reconstruction. However, existing algorithms are not automated and accurate enough in practice. To solve this problem, this paper proposes a novel layover detection method that combines the complex-valued (cv) neural network and expert knowledge to extract features in the amplitude and phase of multi-channel SAR. First, inspired by expert knowledge, a fast Fourier transform (FFT) residual convolutional neural network was developed to eliminate the training divergence of the cv network, deepen networks without extra parameters, and facilitate network learning. Then, another innovative component, phase convolution, was designed to extract phase features of the layover. Subsequently, various cv neural network components were integrated with FFT residual learning blocks and phase convolution on the skeleton of U-Net. Due to the difficulty of obtaining SAR images marked with layover truths, a simulation was performed to gather the required dataset for training. The experimental results indicated that our approach can efficiently determine the layover area with higher precision and fewer noises. The proposed method achieves an accuracy of 97% on the testing dataset, which surpasses previous methods.