Search

Background The prediction of potentially pathogenic variant combinations in patients remains a key task in the field of medical genetics for the understanding and detection of oligogenic/multilocus diseases. Models tailored towards such cases can help shorten the gap of missing diagnoses and can aid researchers in dealing with the high complexity of the derived data. The predictor VarCoPP (Variant Combinations Pathogenicity Predictor) that was published in 2019 and identified potentially pathogenic variant combinations in gene pairs (bilocus variant combinations), was the first important step in this direction. Despite its usefulness and applicability, several issues still remained that hindered a better performance, such as its False Positive (FP) rate, the quality of its training set and its complex architecture. Results We present VarCoPP2.0: the successor of VarCoPP that is a simplified, faster and more accurate predictive model identifying potentially pathogenic bilocus variant combinations. Results from cross-validation and on independent data sets reveal that VarCoPP2.0 has improved in terms of both sensitivity (95% in cross-validation and 98% during testing) and specificity (5% FP rate). At the same time, its running time shows a significant 150-fold decrease due to the selection of a simpler Balanced Random Forest model. Its positive training set now consists of variant combinations that are more confidently linked with evidence of pathogenicity, based on the confidence scores present in OLIDA, the Oligogenic Diseases Database (https://olida.ibsquare.be). The improvement of its performance is also attributed to a more careful selection of up-to-date features identified via an original wrapper method. We show that the combination of different variant and gene pair features together is important for predictions, highlighting the usefulness of integrating biological information at different levels. Conclusions Through its improved performance and faster execution time, VarCoPP2.0 enables a more accurate analysis of larger data sets linked to oligogenic diseases. Users can access the ORVAL platform (https://orval.ibsquare.be) to apply VarCoPP2.0 on their data.

Short-channel Gallium Nitride (GaN) high-electron-mobility transistors (HEMTs) often utilize T-shape gates due to their large gate-line cross-sectional area and subsequent fMAX increase. In this paper, we report the linearity trade-offs associated with varying the T-gate geometries of AlGaN/GaN HEMTs on Si, specifically the gate extensions which serve as field plates and their impact on the large-signal performance. Small-signal characterization and modeling, in addition to TCAD, provide initial guidelines for the optimal dimensions for the gate field plates using the ratio of fT and the product of the gate resistance and the gate-to-drain capacitance. We utilize various characterization methods, including 6 GHz non-linear vector network analyzer characterization in addition to load-pull, to quantify the amplitude and phase distortion and their subsequent impact on the large-signal metrics of the devices under differing matching conditions and bias points. We deduce that the influence of the gate field plates on the amplitude and phase distortion is non-negligible, particularly under matched conditions.

Wireless networks are experiencing a surge in demand for data bandwidth, necessitating higher performance and faster data rates. However, conventional signal processing techniques are inadequate for potential future developments that require the use of higher carrier frequencies. The terahertz region of the spectrum has become an increasingly prominent area of research for wireless communication applications. However, components commonly utilized for lower frequency ranges exhibit inefficiencies and significant losses when operating at terahertz frequencies. Furthermore, the miniaturization of antenna sizes to the micrometer scale has made the concept of perfect electric conductor ineffective in the terahertz range. Therefore, it is imperative to establish a reference system of electromagnetic devices. Due to the essential role of polarization in wireless communication and the convenient and widely used properties of wire-grid polarizers, this article establishes a reference for the polarization effect of wire grid polarizers and obtains their optimal parameters. The optimal parameters are 61.58 μm distance, 20 μm thickness and 20 μm width within the frequency band from 0.5 THz to 0.75 THz. We proposed a three-layer structure consisting of copper wire-grids, monolayer graphene, and silicon dioxide substrate to combine the remarkable electromagnetic properties of graphene with wire-grids polarizer. The resulting graphene-enhanced polarizer achieved excellent polarization improvement, with a strongly reduced S21 transmission magnitude up to − 55 dB compared to the case without graphene, representing an approximately 20 dB decrease.

3D Anthropometric measurement extraction is of paramount importance for several applications such as clothing design, online garment shopping, and medical diagnosis, to name a few. State-of-the-art 3D anthropometric measurement extraction methods estimate the measurements either through some landmarks found on the input scan or by fitting a template to the input scan using optimization-based techniques. Finding landmarks is very sensitive to noise and missing data. Template-based methods address this problem, but the employed optimization-based template fitting algorithms are computationally very complex and time-consuming. To address the limitations of existing methods, we propose a deep neural network architecture which fits a template to the input scan and outputs the reconstructed body as well as the corresponding measurements. Unlike existing template-based anthropocentric measurement extraction methods, the proposed approach does not need to transfer and refine the measurements from the template to the deformed template, thereby being faster and more accurate. A novel loss function, especially developed for 3D anthropometric measurement extraction is introduced. Additionally, two large datasets of complete and partial front-facing scans are proposed and used in training. This results in two models, dubbed Anet-complete and Anet-partial, which extract the body measurements from complete and partial front-facing scans, respectively. Experimental results on synthesized data as well as on real 3D scans captured by a photogrammetry-based scanner, an Azure Kinect sensor, and the very recent TrueDepth camera system demonstrate that the proposed approach systematically outperforms the state-of-the-art methods in terms of accuracy and robustness.

This is an updated report of the work of the JPEG Committee formed by the International Telecommunication Union (ITU), the International Organization for Standardization (ISO), and the International Electrotechnical Commission (IEC), known as the Joint Photographic Experts Group (JPEG). A detailed introduction to JPEG's work was already presented in Ref. 1. In this report, the focus is more on updated work since 2020 and on new work items and explorations.

In this paper, we introduce the new problem of extracting fine-grained traffic information from Twitter streams by also making publicly available the two (constructed) traffic-related datasets from Belgium and the Brussels capital region. In particular, we experiment with several models to identify (i) whether a tweet is traffic-related or not, and (ii) in the case that the tweet is traffic-related to identify more fine-grained information regarding the event (e.g., the type of the event, where the event happened). To do so, we frame (i) the problem of identifying whether a tweet is a traffic-related event or not as a text classification subtask, and (ii) the problem of identifying more fine-grained traffic-related information as a slot filling subtask, where fine-grained information (e.g., where an event has happened) is represented as a slot/entity of a particular type. We propose the use of several methods that process the two subtasks either separately or in a joint setting, and we evaluate the effectiveness of the proposed methods for solving the traffic event detection problem. Experimental results indicate that the proposed architectures achieve high performance scores (i.e., more than 95% in terms of F$_{1}$ score) on the constructed datasets for both of the subtasks (i.e., text classification and slot filling) even in a transfer learning scenario. In addition, by incorporating tweet-level information in each of the tokens comprising the tweet (for the BERT-based model) can lead to a performance improvement for the joint setting., Comment: Author pre-print version

Synthetic aperture radar (SAR) techniques are commonly used in spaceborne and airborne side-looking radar imaging applications, where the relatively high platform speeds enable the formation of very long synthetic apertures, which provide images with high angular resolution. However, the recent development of new autonomous ground vehicles, with relatively slow speeds, calls for new SAR imaging methods to obtain high angular resolution images. In this paper, the concept of forward-looking multiple-input multiple-output SAR (FL-MIMO-SAR) is analyzed and applied for short-range millimeter-wave radar forward-looking imaging in autonomous mobile robots (AMRs). Moreover, if the antenna inter-element spacing in the MIMO array is small, the FL-MIMO-SAR imaging system will provide low or no angular resolution refinement at the radar boresight region, i.e., Doppler beam sharpening (DBS) ineffective region. To solve that, the combination of FL-MIMO-SAR and sparse MIMO arrays with large inter-element spacing (FL-sparseMIMO-SAR) is proposed. Aspects such as the signal model, image reconstruction, complexity reduction, and the left-right ambiguity for FL-MIMO-SAR imaging systems are addressed. In addition, different imaging scenarios for AMRs are simulated using a frequency-modulated continuous wave radar simulator. Simulation results are validated with real measurements. It is shown that the proposed combination of FL-MIMO-SAR and MIMO arrays with large inter-element spacing manages to significantly suppress the grating lobes, which come from the sparsity in the antenna array. Moreover, the proposed FL-sparseMIMO-SAR helps to relax the requirements on the synthetic aperture length and solves the DBS-blind region problem, thanks to the larger real sparse MIMO array and the grating lobes suppression with the FL-SAR processing. Finally, the results show high potential for radar imaging systems, with FL-sparseMIMO-SAR capabilities, to be employed in the applications of AMRs towards smart factories and warehouses.

This paper addresses infant cry classification in multi-view settings, that is, settings where the typical low-level representations, commonly used for audio recognition tasks, are considered as different views of the target data. We show that through the use of multi-view methods, such as Structured Latent Multi-View Representation Learning, we are able to reliably discriminate between normal and pathological infant cry signals. Extensive experimental results on two benchmark infant cry data sets indicate that the proposed method outperforms state-of-the-art models.

Wind is a renewable energy source that has become more important in recent years. Wind turbines are equipped with a SCADA system, which allows for remote supervision of the wind farm. SCADA systems are customarily used to provide data averaged every 10 minutes. Nevertheless, recent literature suggests that more insights could be extracted with a higher granularity of data. In this work, a naive methodology based on Multi-Task Gaussian Process Regression is presented, in order to show how spatiotemporal modeling benefits power estimation. Using sparsity properties a model for possible power prediction is proposed. The model proposed performs better than the power curves provided by the manufacturer.

ith the ongoing trend towards digitisation, vast amounts of, often very fine-grained, data are being collected. The ultimate goal is to capture and understand the behaviour of a system, such as the traffic in a city. However, making sense of such data is not straightforward due to its high level of detail and complex dependencies in time and space. Exploring heuristic approaches is essential to arrive at data representations that enable better insights into the underlying system dynamics by zooming out from the detail. In this paper, a novel approach for representing and reasoning about traffic state transition behaviour via a multitude of parameterised Markov chains models, cleverly designed to fit in a cascade, is proposed. The benefits of working with a multitude of individual Markov chains are outlined and subsequently, it is illustrated how to combine them into daily transition graphs such that their graph representation can be exploited to extract insights about daily traffic behaviour. In addition, targeting context-specific studies, an alternative approach is introduced combining in a dynamic fashion a cascade of Markov chains covering longer and overlapping time windows. A recursive algorithm is conceived and validated allowing to exploit this cascade structure for computing state transition probabilities over time. The potential of the proposed approach for mining traffic state transitions is demonstrated on a use case derived from real-world data.

Carbon nanofibers (CNFs) is a typical electrical loss material with strong permittivity that has been widely used in the field of electromagnetic shielding, attenuation, and reception. However, studies have demonstrated that the unstable anisotropic EM behavior exhibited by CNFs in the terahertz domain hinders its application in the tremendously high frequency domain. In this study, ZnO is used to perform surface modification of CNFs based on the sputtering method, which effectively eliminate the anisotropic EM behavior of CNFs at different irradiation angles in the sub-terahertz band. ZnO coated the surface of fiber wall comprehensively in the form of nano-scale. The results show that after the surface modification, the composite sample exhibits a stable isotropic EM response in the sub-terahertz band of 500-750 GHz. Also, in the frequency range of 551-563 GHz (bandwidth: 12 GHz), ZnO not only acts as a stabilizer, but also enhances the absorption performance of the original CNFs by 6%, which is up to 92%. In the frequency range of 660-690 GHz (bandwidth: 30 GHz), the composite sample and the pure CNFs sample have similar wave absorption performance with about 75%, the maximum difference is less than 5%, and the minimum is about 2.6%.

International audience; The age of computational imaging is merging the physical hardware-driven approach of photonics with advanced signal processing methods from software-driven computer engineering and applied mathematics. The compressed sensing theory in particular established a practical framework for reconstructing the scene content using few linear combinations of complex measurements and a sparse prior for regularizing the solution. Compressed sensing found direct applications in digital holography for microscopy. Indeed, the wave propagation phenomenon in free space mixes in a natural way the spatial distribution of point sources from the 3-dimensional scene. As the 3-dimensional scene is mapped to a 2-dimensional hologram, the hologram samples form a compressed representation of the scene as well. This overview paper discusses contributions in the field of compressed digital holography at the micro scale. Then, an outreach on future extensions towards the real-size macro scale is discussed. Thanks to advances in sensor technologies, increasing computing power and the recent improvements in sparse digital signal processing, holographic modalities are on the verge of practical high-quality visualization at a macroscopic scale where much higher resolution holograms must be acquired and processed on the computer.

This work investigates the star discrepancies and squared integration errors of two quasi-random points constructions using a generator one-dimensional sequence and the Hilbert space-filling curve. This recursive fractal is proven to maximize locality and passes uniquely through all points of the d-dimensional space. The van der Corput and the golden ratio generator sequences are compared for random- ized integro-approximations of both Lipschitz continuous and piecewise constant functions. We found that the star discrepancy of the construction using the van der Corput sequence reaches the theoretical optimal rate when the number of samples is a power of two while using the golden ratio sequence performs optimally for Fibonacci numbers. Since the Fibonacci sequence increases at a slower rate than the exponential in base 2, the golden ratio sequence is preferable when the budget of samples is not known beforehand. Numerical experiments confirm this observation.

This letter presents a two-stage mm-wave power amplifier (PA) in 22nm fully depleted silicon-on-insulator (FD-SOI) technology. High output power is obtained through transistor stacking. Transistor segmentation and back-gate bias control techniques are used in this adaptive PA which can work in high gain mode (HGM) and high linearity mode (HLM). The proposed PA achieves 17.9 dBm saturated output power ( Psat ), 27.6 dB power gain, 11.6 dBm output 1-dB compression point (OP1dB), and 11.0% peak power-added efficiency (PAE) at 74 GHz in the HGM. In the HLM, 17.1 dBm Psat , 21.2 dB power gain, 14.6 dBm OP1dB, and 12.9% peak PAE are obtained at 74 GHz.

This article presents a 55-63-GHz fundamental multicore voltage-controlled oscillator (VCO) in a 28-nm bulk CMOS process. The single-core VCO utilizes stacking and magnetic coupling of two NMOS-based resonators, which increases the tank energy for phase noise reduction while maintaining low voltage swings for reliability. In the multi-core configuration, the oscillators are coupled in a way that also simplifies LO distribution toward the transmitter and receiver. The potential parasitic modes that occur both in single-core and multi-core configurations are suppressed by design. The implicit common-mode (CM) impedance control is utilized to reduce phase noise without any large extra tail inductor as used in a conventional complementary oscillator. The distributed nature of energy restoring elements in the proposed stacked oscillator also offers robust parasitic mode suppression compared to alternative implementation based on a conventional complementary oscillator. The working range of the prototype is between55 and 63 GHz corresponding to a tuning range of 13.4% with power consumption varying from 15.3 to 12 mW across the tuning range. The prototype achieves the best phase noise of -95 and -121.7 dBc/Hz at 1- and 10-MHz offsets from the carrier at 55GHz. The corresponding peak figure of merits (FOMs) achieved are 178 and 185 dBc/Hz at 1- and 10-MHz offsets from the carrier.

This article presents a delta sigma modulator (DSM) analog to digital (ADC) that uses ring amplifiers as integrators to relax speed and efficiency bottlenecks in discrete-time (DT) oversampled ADCs. Its multi-bit quantizer is based on split source (SS) comparators, adding flexibility and power efficiency. The complete oversampling ADC is designed as a 3rd-order cascade of integrator with feed forward (CIFF) with a 4-bit quantizer, and it achieves a peak signal-to-noise and distortion ratio (SNDR) of 67 dB and DR of 70.0 dB with 47.5-MHz bandwidth when clocked at 950 MHz. This is the highest bandwidth reported to date among single-channel DT DSM ADCs and demonstrates a viable alternative to continuous-time (CT) DSM ADCs for wideband oversampling applications. With a power consumption of 4.7 mW from a 1-V supply, figure of merit (FoM) Schreier and Walden are 167.0 dB and 27.0 fJ/c.s, respectively, demonstrating efficient DT delta-sigma conversion with high bandwidth.

Conditional random fields (CRFs), a particular type of graph neural networks (GNNs), can be used to make structured predictions in machine learning, with various applications from image processing and natural language processing to recommender systems. CRFs refine the prediction of a sample by taking into account its context information. However, there is a lack of work on post-hoc explanation approaches to CRFs, especially when the model is softmax-activated like the deep mean field network (DMFN). In this paper, we bridge this gap by proposing a layer-wise relevance propagation (LRP) method based on deep Taylor decomposition to explain CRFs, especially the DMFN model. The method considers the intermediate softmax activation layers in DMFN. We use two evaluation settings: top K\% deletion and insertion to evaluate the method. Experimental studies on fake news detection using the DMFN model prove the effectiveness of our explanation method compared to the other baseline methods.

Objective. To test and compare different intensity standardization approaches for whole-body multi-parametric MR images, aiming to compensate voxel intensity differences between scans. These differences, common for magnetic resonance imaging, pose problems in image quantification, assessment of changes between a baseline and follow-up scan, and hinder performance of image processing and machine learning algorithms. Approach. In this work, we present a comparison on the accuracy of intensity standardization approaches with increasing complexity, for intra- and inter-patient multi-parametric whole-body MRI. Several approaches were used: z-scoring of the intensities, piecewise linear mapping and deformable mapping of intensity distributions into established reference intensity space. For each method, the impact on standardization algorithm on the use of single image or average population distribution reference; as well as, whole image and region of interest were additionally investigated. All methods were validated on a data set of 18 whole-body anatomical and diffusion-weighted MR scans consisting of baseline and follow-up examinations acquired from advanced prostate cancer patients and healthy volunteers. Main results. The piecewise linear intensity standardisation approach provided the best compromise between standardization accuracy and method stability, with average deviations in intensity profile of 0.011-0.027 and mean absolute difference of 0.29-0.37 standard score (intra-patient) and 0.014-0.056 (inter-patient), depending on the type of used MR modality. Significance. Linear piecewise approaches showed the overall best performance across multiple validation metrics, mostly because of its robustness. The inter-patient standardization proved to perform better when using population average reference image; in contrary to intra-patient approach, where the best results were achieved by standardizing towards a reference image taken as the baseline scan.

In this paper, the three-dimensional (3-D) imaging problem of monostatic forward-looking synthetic aperture radar (FL-SAR) is analyzed. A 3-D guided-and-decimated backprojection (3-D GDBP) algorithm is proposed for reducing the computational complexity of 3-D FL-SAR image reconstruction. This is done by combining range and Doppler processing together with decimation along the slow-time samples and backprojection along the fast-time samples. In addition, the geometry and frequency-modulated continuous wave (FMCW) signal model for the 3-D FL-SAR problem are presented. Finally, the performance of the proposed method is tested and compared against the 3-D decimated backprojection algorithm.

In this paper, we address the problem of ensuring that autonomous learning agents are in alignment with multiple moral values. Specifically, we present the theoretical principles and algorithmic tools necessary for creating an environment where an agent is assured of learning a behaviour (or policy) that corresponds to multiple moral values while striving to achieve its individual objective. To address this value alignment problem, we adopt the Multi-Objective Reinforcement Learning framework and propose a novel algorithm that combines techniques from Multi-Objective Reinforcement Learning and Linear Programming. In addition to providing theoretical guarantees, we illustrate our value alignment process with an example involving an autonomous vehicle. Here, we demonstrate that the agent learns to behave in alignment with the ethical values of safety, achievement, and comfort. Additionally, we use a synthetic multi-objective environment generator to evaluate the computational costs associated with guaranteeing ethical learning in situations with an increasing numbers of values.

obotic manipulation in cluttered environments requires synergistic planning among prehensile and non-prehensile actions. Previous works on sampling-based Task and Motion Planning (TAMP) algorithms, e.g. PDDLStream, provide a fast and generalizable solution for multi-modal manipulation. However, they are likely to fail in cluttered scenarios where no collision-free grasping approaches can be sampled without preliminary manipulations. To extend the ability of sampling-based algorithms, we integrate a vision-based Reinforcement Learning (RL) non-prehensile procedure, pusher. The pushing actions generated by pusher can eliminate interlocked situations and make the grasping problem solvable. Also, the sampling-based algorithm evaluates the pushing actions by providing rewards in the training process, thus the pusher can learn to avoid situations leading to irreversible failures. The proposed hybrid planning method is validated on a cluttered bin-picking problem and implemented in both simulation and real world. Results show that the pusher can effectively improve the success ratio of the previous sampling-based algorithm, while the sampling-based algorithm can help the pusher learn pushing skills.

Real-world sequential decision-making tasks are usually complex, and require trade-offs between multiple -- often conflicting -- objectives. However, the majority of research in reinforcement learning (RL) and decision-theoretic planning assumes a single objective, or that multiple objectives can be handled via a predefined weighted sum over the objectives. Such approaches may oversimplify the underlying problem, and produce suboptimal results. This extended abstract outlines the limitations of using a semi-blind iterative process to solve multi-objective decision making problems. Our extended paper serves as a guide for the application of explicitly multi-objective methods to difficult problems.

We present a detailed analysis of Nash equilibria in multi-objective normal-form games, which are normal-form games with vectorial payoffs. Our approach is based on modelling each player's utility using a utility function that maps a vector to a scalar utility. For mixed strategies, we can apply the utility function before calculating the expectation of the payoff vector as well as after, resulting in two distinct optimisation criteria. We show that when computing the utility from the expected payoff, a Nash equilibrium can be guaranteed when players have quasiconcave utility functions. In addition, we show that when players have quasiconvex utility functions, pure strategy Nash equilibria are equal under both optimisation criteria. We extend this to settings where some players optimise for one criterion, while others optimise for the second. We combine these results and formulate an algorithm that computes all pure strategy Nash equilibria given quasiconvex utility functions.

Flash flood early warning requires accurate rainfall forecasts with a high spatial and temporal resolution. As the first few hours ahead are already not sufficiently well captured by the rainfall forecasts of numerical weather prediction (NWP) models, rainfall nowcasting can provide an alternative. This observation-based method, however, quickly loses skill after the first few hours of the forecast due to growth and dissipation processes that are not accounted for. In addition, providing an additional forecasting method can let users drown in the amount of available information. A promising way forward is a seamless forecasting system, which combines the aforementioned forecasting methods. By optimally combining (blending) rainfall nowcasts with NWP forecasts, we can extend the skillful lead time of short-term rainfall forecasts and provide users with more consistent, seamless forecasts.We implemented an adaptive scale-dependent ensemble blending method in the open-source pysteps library. In this implementation, the blending of the extrapolation (ensemble) nowcast, (ensemble) NWP and noise components is performed level-by-level, which means that the blending weights vary per spatial cascade level. These scale-dependent blending weights are computed from the recent skill of the forecast components, and converge to a climatological value, which is computed from a multi-day rolling window and can be adjusted to the (operational) needs of the user. To constrain the (dis)appearance of rain in the ensemble members to regions around the rainy areas, we have developed a Lagrangian blended probability matching scheme and incremental masking strategy.We evaluate the method using three heavy and extreme (July 2021) rainfall events in four Belgian and Dutch catchments, focusing on both the rainfall forecasts and the resulting discharge forecasts using the fully distributed wflow_sbm hydrological model. We benchmark the results of the 48-member blended forecasts against the deterministic Belgian NWP forecast, a 48-member nowcast and a simple 48-member linear blending approach. When focusing on the resulting rainfall forecasts, the introduced blending approach predominantly performs similarly or better than only nowcasting (in terms of event-averaged CRPS and CSI values) and adds value compared to NWP for the first hours of the forecast. This holds for both the radar domain and catchment scale, although the difference, particularly with the linear blending method, reduces when we focus on catchment-average cumulative rainfall sums instead of instantaneous rainfall rates. We find similar results for the resulting discharge forecasts, although the effect of the catchment size and corresponding lag times becomes influential and determines the added value of nowcasting over NWP. By properly combining observations and NWP forecasts, blending methods such as these are a crucial component of seamless hydrometeorological forecasting systems.

We propose a novel multi-objective reinforcement learning algorithm that successfully learns the optimal policy even for non-linear utility functions. Non-linear utility functions pose a challenge for SOTA approaches, both in terms of learning efficiency as well as the solution concept. A key insight is that, by proposing a critic that learns a multi-variate distribution over the returns, which is then combined with accumulated rewards, we can directly optimize on the utility function, even if it is non-linear. This allows us to vastly increase the range of problems that can be solved compared to those which can be handled by single-objective methods or multi-objective methods requiring linear utility functions, yet avoiding the need to learn the full Pareto front. We demonstrate our method on multiple multi-objective benchmarks, and show that it learns effectively where baseline approaches fail.

Flash flood early warning requires accurate rainfall forecasts with a high spatial and temporal resolution. As the first few hours ahead are already not sufficiently well captured by the rainfall forecasts of numerical weather prediction (NWP) models, radar rainfall nowcasting can provide an alternative. Because this observation-based method quickly loses skill after the first 2 hr of the forecast, it needs to be combined with NWP forecasts to extend the skillful lead time of short-term rainfall forecasts, which should increase decision-making times. We implemented an adaptive scale-dependent ensemble blending method in the open-source pysteps library, based on the Short-Term Ensemble Prediction System scheme. In this implementation, the extrapolation (ensemble) nowcast, (ensemble) NWP, and noise components are combined with skill-dependent weights that vary per spatial scale level. To constrain the (dis)appearance of rain in the ensemble members to regions around the rainy areas, we have developed a Lagrangian blended probability matching scheme and incremental masking strategy. We describe the implementation details and evaluate the method using three heavy and extreme (July 2021) rainfall events in four Belgian and Dutch catchments. We benchmark the results of the 48-member blended forecasts against the Belgian NWP forecast, a 48-member nowcast, and a simple 48-member linear blending approach. Both on the radar domain and catchment scale, the introduced blending approach predominantly performs similarly or better than only nowcasting (in terms of event-averaged continuous ranked probability score and critical success index values) and adds value compared with NWP for the first hours of the forecast, although the difference, particularly with the linear blending method, reduces when we focus on catchment-average cumulative rainfall sums instead of instantaneous rainfall rates. By properly combining observations and NWP forecasts, blending methods such as these are a crucial component of seamless prediction systems.

Brain–computer interfaces (BCIs) have the potential to enable individuals to interact with devices by detecting their intention from brain activity. A common approach to BCI is to decode movement intention from motor imagery (MI), the mental representation of an overt action. However, research-grade electroencephalogram (EEG) acquisition devices with a high number of sensors are typically necessary to achieve the spatial resolution required for reliable analysis. This entails high monetary and computational costs that make these approaches impractical for everyday use. This study investigates the trade-off between accuracy and complexity when decoding MI from fewer EEG sensors. Data were acquired from 15 healthy participants performing MI with a 64-channel research-grade EEG device. After performing a quality assessment by identifying visually evoked potentials, several decoding pipelines were trained on these data using different subsets of electrode locations. No significant differences (p = [0.18–0.91]) in the average decoding accuracy were found when using a reduced number of sensors. Therefore, decoding MI from a limited number of sensors is feasible. Hence, using commercial sensor devices for this purpose should be attainable, reducing both monetary and computational costs for BCI control.

Background Clinical pathways are one of the main tools to manage the health care’s quality and concerned with the standardization of care processes. They have been used to help frontline healthcare workers by presenting summarized evidence and generating clinical workflows involving a series of tasks performed by various people within and between work environments to deliver care. Integrating clinical pathways into Clinical Decision Support Systems (CDSSs) is a common practice today. However, in a low-resource setting (LRS), this kind of decision support systems is often not readily accessible or even not available. To fill this gap, we developed a computer aided CDSS that swiftly identifies which cases require a referral and which ones may be managed locally. The computer aided CDSS is designed primarily for use in primary care settings for maternal and childcare services, namely for pregnant patients, antenatal and postnatal care. The purpose of this paper is to assess the user acceptance of the computer aided CDSS at the point of care in LRSs. Methods For evaluation, we used a total of 22 parameters structured in to six major categories, namely “ease of use, system quality, information quality, decision changes, process changes, and user acceptance.” Based on these parameters, the caregivers from Jimma Health Center's Maternal and Child Health Service Unit evaluated the acceptability of a computer aided CDSS. The respondents were asked to express their level of agreement using 22 parameters in a think-aloud approach. The evaluation was conducted in the caregiver's spare-time after the clinical decision. It was based on eighteen cases over the course of two days. The respondents were then asked to score their level of agreement with some statements on a five-point scale: strongly disagree, disagree, neutral, agree, and strongly agree. Results The CDSS received a favorable agreement score in all six categories by obtaining primarily strongly agree and agree responses. In contrast, a follow-up interview revealed a variety of reasons for disagreement based on the neutral, disagree, and strongly disagree responses. Conclusions Though the study had a positive outcome, it was limited to the Jimma Health Center Maternal and Childcare Unit, and hence a wider scale evaluation and longitudinal measurements, including computer aided CDSS usage frequency, speed of operation and impact on intervention time are needed.

BackgroundBefore starting surgery for the resection of an intracranial tumor, its outlines are typically marked on the skin of the patient. This allows for the planning of the optimal skin incision, craniotomy, and angle of approach. Conventionally, the surgeon determines tumor borders using neuronavigation with a tracked pointer. However, interpretation errors can lead to important deviations, especially for deep-seated tumors, potentially resulting in a suboptimal approach with incomplete exposure. Augmented reality (AR) allows displaying of the tumor and critical structures directly on the patient, which can simplify and improve surgical preparation.MethodsWe developed an AR-based workflow for intracranial tumor resection planning deployed on the Microsoft HoloLens II, which exploits the built-in infrared-camera for tracking the patient. We initially performed a phantom study to assess the accuracy of the registration and tracking. Following this, we evaluated the AR-based planning step in a prospective clinical study for patients undergoing resection of a brain tumor. This planning step was performed by 12 surgeons and trainees with varying degrees of experience. After patient registration, tumor outlines were marked on the patient's skin by different investigators, consecutively using a conventional neuronavigation system and an AR-based system. Their performance in both registration and delineation was measured in terms of accuracy and duration and compared.ResultsDuring phantom testing, registration errors remained below 2.0 mm and 2.0° for both AR-based navigation and conventional neuronavigation, with no significant difference between both systems. In the prospective clinical trial, 20 patients underwent tumor resection planning. Registration accuracy was independent of user experience for both AR-based navigation and the commercial neuronavigation system. AR-guided tumor delineation was deemed superior in 65% of cases, equally good in 30% of cases, and inferior in 5% of cases when compared to the conventional navigation system. The overall planning time (AR = 119 ± 44 s, conventional = 187 ± 56 s) was significantly reduced through the adoption of the AR workflow (p < 0.001), with an average time reduction of 39%.ConclusionBy providing a more intuitive visualization of relevant data to the surgeon, AR navigation provides an accurate method for tumor resection planning that is quicker and more intuitive than conventional neuronavigation. Further research should focus on intraoperative implementations.

Partially Observable Markov Decision Processes (POMDPs) are useful tools to model environments where the full state cannot be perceived by an agent. As such the agent needs to reason taking into account the past observations and actions. However, simply remembering the full history is generally intractable due to the exponential growth in the history space. Keeping a probability distribution that models the belief over what the true state is can be used as a sufficient statistic of the history, but its computation requires access to the model of the environment and is also intractable. Current state-of-the-art algorithms use Recurrent Neural Networks (RNNs) to compress the observation-action history aiming to learn a sufficient statistic, but they lack guarantees of success and can lead to suboptimal policies. To overcome this, we propose the Wasserstein-Belief-Updater (WBU), an RL algorithm that learns a latent model of the POMDP and an approximation of the belief update. Our approach comes with theoretical guarantees on the quality of our approximation ensuring that our outputted beliefs allow for learning the optimal value function.

Model output from climate projections often requires bias-adjustment to compensate for systematic model errors. A bias-adjustment method for extreme precipitation intensity is proposed that preserves the scaling equation for different accumulation levels from hourly to daily, using intensity-duration-frequency (IDF) modeling. A validation is performed within a pseudo-reality setting, based on hourly precipitation from 28 regional climate model projections of the EURO-CORDEX ensemble over Belgium. The scaling-based adjustment methods improve upon previous methods, an optimal method is identified, and, analytical quantile mapping methods must be avoided due to three identified problems. The ensemble mean of the adjusted extreme precipitation intensity obeys the above-mentioned scale-invariance property, which is consistent with observed extreme intensities. We thus show that IDF modeling provides added value in the context of bias-adjustment, and, that the particular IDF model proposed balances well between accuracy and the preservation of desired properties such as scale invariance and consistency among rainfall durations.

Background Long-term sequelae of COVID-19 can result in reduced functionality of the central nervous system and substandard quality of life. Gaining insight into the recovery trajectory of admitted COVID-19 patients on their cognitive performance and global structural brain connectivity may allow a better understanding of the diseases' relevance. Objectives To assess whole-brain structural connectivity in former non-intensive-care unit (ICU)- and ICU-admitted COVID-19 survivors over 2 months following hospital discharge and correlate structural connectivity measures to cognitive performance. Methods Participants underwent Magnetic Resonance Imaging brain scans and a cognitive test battery after hospital discharge to evaluate structural connectivity and cognitive performance. Multilevel models were constructed for each graph measure and cognitive test, assessing the groups' influence, time since discharge, and interactions. Linear regression models estimated whether the graph measurements affected cognitive measures and whether they differed between ICU and non-ICU patients. Results Six former ICU and six non-ICU patients completed the study. Across the various graph measures, the characteristic path length decreased over time (β = 0.97, p = 0.006). We detected no group-level effects (β = 1.07, p = 0.442) nor interaction effects (β = 1.02, p = 0.220). Cognitive performance improved for both non-ICU and ICU COVID-19 survivors on four out of seven cognitive tests 2 months later (p Conclusion Adverse effects of COVID-19 on brain functioning and structure abate over time. These results should be supported by future research including larger sample sizes, matched control groups of healthy non-infected individuals, and more extended follow-up periods.

Neurological disorders are the leading cause of physical and cognitive disability across the globe, currently affecting approximately 15% of the worldwide population [...]

Driven by the increased consciousness in data ownership and privacy, zero-knowledge proofs (ZKPs) have become a popular tool to convince a third party of the truthfulness of a statement without disclosing any further information. As ZKPs are rather complex to design, frameworks that transform high- level languages into ZKPs have been proposed. We propose Circuitree, a Datalog reasoner in zero-knowledge. Datalog is a high-level declarative logic language that is generally used for querying. Furthermore, as a logic language, it can also be used to solve logic problems. An application using Circuitree can efficiently generate ZKPs, based on Datalog rules and encrypted data, to prove that a certain conclusion follows from a Datalog ruleset and encrypted input data. Compared to existing frameworks, which generally use their own limited imperative languages, Circuitree uses an existing high-level declarative language. We point out several applications for Circuitree, including EU Digital COVID Certificates and privacy- preserving access control for peer-to-peer (p2p) networks. Circuitree’s performance is evaluated for access control in a p2p network. First results show that our approach allows for fast proofs and proof verification for this application.

A novel method for noise removal in multicamera depth sensing systems is proposed in this work. The method uses a combination of convolutional neural networks applied on each depth camera separately, followed by depth reprojection and joint processing of the resulting point clouds using a 3-D neural network. Both depth and point cloud processing networks are designed to preserve the structure of the depth maps by appropriately correcting noise in the direction of the viewing rays. The proposed method accommodates any depth unit and noise intensity, thanks to adequate normalization in both the processing steps. The proposed approach is shown to outperform the state-of-the-art methods for both synthetic and real data captured with a multicamera setup and can reduce intercamera inconsistencies while preserving depth map structures.

Pooling layers are essential building blocks of convolutional neural networks (CNNs), to reduce computational overhead and increase the receptive fields of proceeding convolutional operations. Their goal is to produce downsampled volumes that closely resemble the input volume while, ideally, also being computationally and memory efficient. Meeting both these requirements remains a challenge. To this end, we propose an adaptive and exponentially weighted pooling method: adaPool. Our method learns a regional-specific fusion of two sets of pooling kernels that are based on the exponent of the Dice-Sørensen coefficient and the exponential maximum, respectively. AdaPool improves the preservation of detail on a range of tasks including image and video classification and object detection. A key property of adaPool is its bidirectional nature. In contrast to common pooling methods, the learned weights can also be used to upsample activation maps. We term this method adaUnPool.We evaluate adaUnPool on image and video super-resolution and frame interpolation. For benchmarking, we introduce Inter4K, a novel high-quality, high frame-rate video dataset. Our experiments demonstrate that adaPool systematically achieves better results across tasks and backbones, while introducing a minor additional computational and memory overhead.

Correct traceability of muscle identity within a predefined set of muscles in EMG studies is relevant in the periodic evaluation process of muscle training programs (for athletes), and in routine reviews for muscle rehabilitation. This article proposes hybrid deep learning CNN-LSTM models to classify the muscle directly from sEMG signals. These models allow for effective feature extraction and learning of short-term and long-term sequential dependencies. Two training setups are proposed: one using weight initialization provided from layer-wise unsupervised pretraining and another one using random initialization. Two validation scenarios are described to assess performance: testing on new contraction bursts from already-seen subjects in the training step (intrapersonal validation, useful in follow-up), and testing on a leave-one-out subject (interpersonal validation). Results indicate that the model can correctly classify different muscle groups in patients that already have been screened, but fails in distinguishing between symmetrical muscles.

A classic model to study strategic decision making in multi-agent systems is the normal-form game. This model can be generalised to allow for an infinite number of pure strategies leading to continuous games. Multi-objective normal-form games are another generalisation that model settings where players receive separate payoffs in more than one objective. We bridge the gap between the two models by providing a theoretical guarantee that a game from one setting can always be transformed to a game in the other. We extend the theoretical results to include guaranteed equivalence of Nash equilibria. The mapping makes it possible to apply algorithms from one field to the other. We demonstrate this by introducing a fictitious play algorithm for multi-objective games and subsequently applying it to two well-known continuous games. We believe the equivalence relation will lend itself to new insights by translating the theoretical guarantees from one formalism to another. Moreover, it may lead to new computational approaches for continuous games when a problem is more naturally solved in the succinct format of multi-objective games., Comment: Accepted to AAMAS 2023

Active load impedance variations in a phased array transmitter cause significant power amplifier (PA) performance degradation, in terms of output power, linearity, and power-added efficiency, which are key parameters to enable high-speed data throughputs using spectrally efficient modulation schemes. The system performance can be restored by using PAs having active or passive reconfigurability with the help of antenna impedance sensors. This article presents a low-power reflection-coefficient sensor for 5G millimeter-wave phased-array applications. The complex load impedance of the PA is determined based on the complex voltage over a sensing element, which can be integrated and co-designed with the PA output matching network, with minimal loss (

The human foot is an incredibly complex combination of bones, joints, and muscles which helps humans with their balance, posture, and mobility. Foot measurement extraction plays an essential role in many applications ranging from medical science to fashion industry. Conventional foot measurement extraction methods require manual interventions using a measuring tape or its digital twins. Recent advancements in 3D scanning technologies and deep learning strategies enable us to propose, to the best of our knowledge, the first deep-learning-based approach to automatic foot measurement extraction from a single 3D scan. The proposed method involves three steps: (i) 3D foot data acquisition, (ii) template fitting, and (iii) measurement extraction. The foot is scanned with the Occipital Structure Sensor Pro. The template fitting process is performed using a deep neural network trained for foot template fitting. Finally, the measurement defined on the template are transferred and refined to the fitted template. The template fitting method is trained on a novel large dataset of \textit{dense} synthetic foot samples. The experimental measurement results demonstrate that the proposed method performs well on both unseen synthetic and real scans.

Advanced structural brain imaging techniques, such as diffusion tensor imaging (DTI), have been used to study the relationship between DTI-parameters and cognitive scores in multiple sclerosis (MS). In this study, we assessed cognitive function in 61 individuals with MS and a control group of 35 healthy individuals with the Symbol Digit Modalities Test, the California Verbal Learning Test-II, the Brief Visuospatial Memory Test-Revised, the Controlled Oral Word Association Test, and Stroop-test. We also acquired diffusion-weighted images (b = 1000; 32 directions), which were processed to obtain the following DTI scalars: fractional anisotropy, mean, axial, and radial diffusivity. The relation between DTI scalars and cognitive parameters was assessed through permutations. Although fractional anisotropy and axial diffusivity did not correlate with any of the cognitive tests, mean and radial diffusivity were negatively correlated with all of these tests. However, this effect was not specific to any specific white matter tract or cognitive test and demonstrated a general effect with only low to moderate individual voxel-based correlations of

Many people in the world are living with chronic diseases, demanding continuous monitoring, diagnosis, and treatment. Continuous physiological monitoring is key to providing preventive healthcare and accurate disease diagnosis, which leads to a growing demand for autonomous wearable technology. Wearable devices acquiring physiological information from the patient demand high-power efficiency to operate in a continuous acquisition mode. While power-saving techniques are applied in wearable devices for many application, very few are considered for biomedical applications. In this work, we explore existing techniques of power reduction for wearable medical devices. Our analysis addresses the power reduction of wearable medical devices and their generalization for different medical signal processing applications. In addition, we propose a taxonomy for power-saving techniques. The common categories of power-saving techniques are task scheduling, clock management, signal compression, and energy awareness. The presented analysis identifies the most appropriate and combined low-power techniques in wearable devices to reduce power consumption.

OBJECTIVES: To quantitatively evaluate the impact of a cardiac acquisition CT mode on motion artifacts in comparison to a conventional cine mode for dynamic musculoskeletal (MSK) imaging. METHODS: A rotating PMMA phantom with air-filled holes drilled at varying distances from the disk center corresponding to linear hole speeds of 0.75 cm/s, 2.0 cm/s, and 3.6 cm/s was designed. Dynamic scans were obtained in cardiac and cine modes while the phantom was rotating at 48°/s in the CT scanner. An automated workflow to compute the Jaccard distance (JD) was established to quantify degree of motion artifacts in the reconstructed phantom images. JD values between the cardiac and cine scan modes were compared using a paired sample t-test. In addition, three healthy volunteers were scanned with both modes during a cyclic flexion-extension motion of the knee and analysed using the proposed metric. RESULTS: For all hole sizes and speeds, the cardiac scan mode had significantly lower (p-value

This work provides a data-oriented overview of the rapidly growing research field covering machine learning (ML) applied to predicting electrochemical corrosion. Our main aim was to determine which ML models have been applied and how well they performed depending on the corrosion topic considered. From an extensive review of corrosion articles presenting comparable performance metrics, a ‘Machine learning for corrosion database’ was created, guiding corrosion experts and model developers in their applications of ML to corrosion. Potential research gaps and recommendations are discussed, and a broad perspective for future research paths is provided.

SiCN thin film is one of the most attractive silicon-based materials due to its excellent electrical, mechanical and optical properties. In this study, SiCN thin films have been prepared by the radio frequency reactive magnetron sputtering method under different sputtering power, pressure, and substrate temperature. The microstructure, morphology, and the optical and field emission properties of SiCN thin films were performed. The results indicated that the high-quality SiCN thin films have been successfully prepared and it is proved that these properties can be tailored by the preparation conditions. A main near ultraviolet light emission line at around 370 nm in SiCN thin films makes it suitable for the development of optoelectronic devices. This study can provide novel guidance for the controlled preparation of high-quality SiCN thin films by magnetron sputtering.

The aim of this study was to evaluate and present an automated method for registration of magnetic resonance imaging (MRI) and computed tomography (CT) or cone beam CT (CBCT) images of the mandibular region for patients with oral squamous cell carcinoma (OSCC). Registered MRI and (CB)CT could facilitate the three-dimensional virtual planning of surgical guides employed for resection and reconstruction in patients with OSCC with mandibular invasion. MRI and (CB)CT images were collected retrospectively from 19 patients. MRI images were aligned with (CB)CT images employing a rigid registration approach (stage 1), a rigid registration approach using a mandibular mask (stage 2), and two non-rigid registration approaches (stage 3). Registration accuracy was quantified by the mean target registration error (mTRE), calculated over a set of landmarks annotated by two observers. Stage 2 achieved the best registration result, with an mTRE of 2.5 ± 0.7 mm, which was comparable to the inter- and intra-observer variabilities of landmark placement in MRI. Stage 2 was significantly better aligned compared to all approaches in stage 3. In conclusion, this study demonstrated that rigid registration with the use of a mask is an appropriate image registration method for aligning MRI and (CB)CT images of the mandibular region in patients with OSCC.

Purpose: Feasability of a no-reference image quality metric was assessed on patient-like images using a patient-specific phantom simulating a frame of a coronary angiogram. Methods: One background and one contrast-filled frame of a coronary angiogram, acquired using a clinical imaging protocol, were selected from a Philips Integris Allura FD (Philips Healthcare, Best, The Netherlands). The background frame's pixels were extruded to a thickness proportional to their grey value. One phantom was 3D printed using composite 80% bronze filament (max. thickness of 5.1 mm), the other was a custom PMMA cast (max thickness of 8.5 cm). A vessel mold was created from the contrast-filled frame and injected with a solution of 320 mg I/ml contrast fluid (75%), water and gelatin. Still X-ray frames of the vessel mold + background phantom + 16 cm PMMA were acquired at manually selected different exposure settings using a Philips Azurion (Philips Healthcare, Best, The Netherlands) in User Quality Control Mode and were exportedas RAW images. The signal-difference-to-noise-ratio-squared (SDNR2) and a spatial-domain-equivalent of the noise equivalent quanta (NEQSDE) were calculated. The Spearman's correlation of the latter parameters with a no-reference perceptual image quality metric (NIQE) was investigated. Results: The bronze phantom showed better resemblance to the original patient frame selected from a coronary angiogram of an actual patient, with better contrast and less blur than the PMMA phantom. Both phantoms were imaged using a comparable imaging protocol to the one used to acquire the original frame. The bronze phantom was hence used together with the vessel mold for image quality measurements on the 165 still phantom frames. A strong correlation was noted between NEQSDE and NIQE (SROCC = –0.99, p < 0.0005) and between SDNR2 and NIQE (SROCC = –0.97, p < 0.0005). Conclusion: Using a cost-effective and easy to realize patient-specific phantom we were able to generate patient-like X-ray frames. NIQE as a no-reference image quality model has the potential to predict physical image quality from patient images.

Background Patient referral is a process in which a healthcare provider decides to seek assistance due to the limitations of available skills, resources and services offered locally. Paper-based referrals predominantly used in low-income countries hardly follow any procedure. This causes a major gap in communication, coordination, and continuity of care between primary and specialized levels, leading to poor access, delay, duplication and unnecessary costs. The goal of this study is to assess the formats and completeness of existing paper-based referral letters in order to improve health information exchange, coordination, and continuity of care. Methods A retrospective exploratory research was conducted in eight public and three private healthcare facilities in the city of Kigali from May to October 2021. A purposive sampling method was used to select hospitals and referral letters from patients’ files. A data capture sheet was designed according to the contents of the referral letters and the resulting responses were analyzed descriptively. Results In public hospitals, five types of updated referral letters were available, in total agreement with World Health Organization (WHO) standards of which two (neonatal transfer form and patient monitoring transfer form) were not used. There was also one old format that was used by most hospitals and another format designed and used by a district hospital (DH) separately. Three formats were designed and used by private hospitals (PH) individually. A total of 2,304 referral letters were perused and the results show that “external transfer” forms were completed at 58.8%; “antenatal, delivery, and postnatal external transfer” forms at 47.5%; “internal transfer” forms at 46.6%; “Referral/counter referral” forms at 46.0%; district hospital referrals (DH2) at 73.4%. Referrals by private hospitals (PH1, PH2 and PH3) were completed at 97.7%, 70.7%, and 0.0% respectively. The major completeness deficit was observed in counter referral information for all hospitals. Conclusion We observed inconsistencies in the format of the available referral letters used by public hospitals, moreover some of them were incompatible with WHO standards. Additionally, there were deficits in the completeness of all types of paper-based referral letters in use. There is a need for standardization and to disseminate the national patient referral guideline in public hospitals with emphasis on referral feedback, referral registry, triage, archiving and a need for regular training in all organizations.