NeuXus open-source tool for real-time artifact reduction in simultaneous EEG-fMRI

The simultaneous acquisition of electroencephalography and functional magnetic resonance imaging (EEG-fMRI) allows the complementary study of the brain's electrophysiology and hemodynamics with high temporal and spatial resolution. One application with great potential is neurofeedback training of targeted brain activity, based on the real-time analysis of the EEG and/or fMRI signals. This depends on the ability to reduce in real time the severe artifacts affecting the EEG signal acquired with fMRI, mainly the gradient and pulse artifacts. A few methods have been proposed for this purpose, but they are either slow, hardware-dependent, publicly unavailable, or proprietary software. Here, we present a fully open-source and publicly available tool for real-time EEG artifact reduction in simultaneous EEG-fMRI recordings that is fast and applicable to any hardware. Our tool is integrated in the Python toolbox NeuXus for real-time EEG processing and adapts to a real-time scenario well-established artifact average subtraction methods combined with a long short-term memory network for R peak detection. We benchmarked NeuXus on three different datasets, in terms of artifact power reduction and background signal preservation in resting state, alpha-band power reactivity to eyes closure, and event-related desynchronization during motor imagery. We showed that NeuXus performed at least as well as the only available real-time tool for conventional hardware setups (BrainVision's RecView) and a well-established offline tool (EEGLAB's FMRIB plugin). We also demonstrated NeuXus' real-time ability by reporting execution times under 250 ms. In conclusion, we present and validate the first fully open-source and hardware-independent solution for real-time artifact reduction in simultaneous EEG-fMRI studies.

Real-time detection of laryngopharyngeal cancer using an artificial intelligence-assisted system with multimodal data

BACKGROUND

Laryngopharyngeal cancer (LPC) includes laryngeal and hypopharyngeal cancer, whose early diagnosis can significantly improve the prognosis and quality of life of patients. Pathological biopsy of suspicious cancerous tissue under the guidance of laryngoscopy is the gold standard for diagnosing LPC. However, this subjective examination largely depends on the skills and experience of laryngologists, which increases the possibility of missed diagnoses and repeated unnecessary biopsies. We aimed to develop and validate a deep convolutional neural network-based Laryngopharyngeal Artificial Intelligence Diagnostic System (LPAIDS) for real-time automatically identifying LPC in both laryngoscopy white-light imaging (WLI) and narrow-band imaging (NBI) images to improve the diagnostic accuracy of LPC by reducing diagnostic variation among on-expert laryngologists.

METHODS

All 31,543 laryngoscopic images from 2382 patients were categorised into training, verification, and test sets to develop, validate, and internal test LPAIDS. Another 25,063 images from five other hospitals were used as external tests. Overall, 551 videos were used to evaluate the real-time performance of the system, and 200 randomly selected videos were used to compare the diagnostic performance of the LPAIDS with that of laryngologists. Two deep-learning models using either WLI (model W) or NBI (model N) images were constructed to compare with LPAIDS.

RESULTS

LPAIDS had a higher diagnostic performance than models W and N, with accuracies of 0·956 and 0·949 in the internal image and video tests, respectively. The robustness and stability of LPAIDS were validated in external sets with the area under the receiver operating characteristic curve values of 0·965-0·987. In the laryngologist-machine competition, LPAIDS achieved an accuracy of 0·940, which was comparable to expert laryngologists and outperformed other laryngologists with varying qualifications.

CONCLUSIONS

LPAIDS provided high accuracy and stability in detecting LPC in real-time, which showed great potential for using LPAIDS to improve the diagnostic accuracy of LPC by reducing diagnostic variation among on-expert laryngologists.

Development and application of a real-time polymerase chain reaction assay to detect lumpy skin disease virus belonging to the Kenyan sheep and goat pox group

Lumpy skin disease (LSD) outbreaks in Southeast and South Asia are attributed to different lineages of LSD virus (LSDV). Variants belonging to the novel recombinant cluster 2.5 circulate in China and Thailand, while a Kenyan sheep and goat pox (KSGP) strain from cluster 1.1 circulates in India, Pakistan, and Bangladesh. The clusters representing these circulating strains are vastly different. However, if their distribution encroaches into each other's ranges, it will be impossible to differentiate between them due to the lack of suitable molecular tools. Thus, fit-for-purpose molecular tools are in demand to effectively and timeously diagnose and investigate the epidemiology of LSDVs in a region. These could significantly contribute to the phylogenetic delineation of LSDVs and the development of preventive measures against transboundary spillovers. This work aimed to develop a real-time polymerase chain reaction assay targeting open reading frame LW032, capable of specifically detecting KSGP-related isolates and recombinant LSDV strains containing the KSGP backbone. The analytical specificity was proven against the widest possible panel of recombinant vaccine-like LSDV strains known to date. The amplification efficiency was 91.08%, and the assay repeatability had a cycle threshold variation of 0.56-1.1 over five repetitions across three runs. This KSGP-specific assay is reliable and fast and is recommended for use in LSDV epidemiological studies where the accurate detection of KSGP genetic signatures is a priority, particularly in regions where KSGP-like and other lineages are circulating.

Highly sensitive and rapid determination of Mycobacterium leprae based on real-time multiple cross displacement amplification

BACKGROUND

Mycobacterium leprae (ML) is the pathogen that causes leprosy, which has a long history and still exists today. ML is an intracellular mycobacterium that dominantly induces leprosy by causing permanent damage to the skin, nerves, limbs and eyes as well as deformities and disabilities. Moreover, ML grows slowly and is nonculturable in vitro. Given the prevalence of leprosy, a highly sensitive and rapid method for the early diagnosis of leprosy is urgently needed.

RESULTS

In this study, we devised a novel tool for the diagnosis of leprosy by combining restriction endonuclease, real-time fluorescence analysis and multiple cross displacement amplification (E-RT-MCDA). To establish the system, primers for the target gene RLEP were designed, and the optimal conditions for E-RT-MCDA at 67 °C for 36 min were determined. Genomic DNA from ML, various pathogens and clinical samples was used to evaluate and optimize the E-RT-MCDA assay. The limit of detection (LoD) was 48.6 fg per vessel for pure ML genomic DNA, and the specificity of detection was as high as 100%. In addition, the detection process could be completed in 36 min by using a real-time monitor.

CONCLUSION

The E-RT-MCDA method devised in the current study is a reliable, sensitive and rapid technique for leprosy diagnosis and could be used as a potential tool in clinical settings.

Terahertz imaging for non-invasive classification of healthy and cimiciato-infected hazelnuts

The development of new non-invasive approaches able to recognize defective food is currently a lively field of research. In particular, a simple and non-destructive method able to recognize defective hazelnuts, such as cimiciato-infected ones, in real-time is still missing. This study has been designed to detect the presence of such damaged hazelnuts. To this aim, a measurement setup based on terahertz (THz) radiation has been developed. Images of a sample of 150 hazelnuts have been acquired in the low THz range by a compact and portable active imaging system equipped with a 0.14 THz source and identified as Healthy Hazelnuts (HH) or Cimiciato Hazelnut (CH) after visual inspection. All images have been analyzed to find the average transmission of the THz radiation within the sample area. The differences in the distribution of the two populations have been used to set up a classification scheme aimed at the discrimination between healthy and injured samples. The performance of the classification scheme has been assessed through the use of the confusion matrix on 50 samples. The False Positive Rate (FPR) and True Negative Rate (TNR) are 0% and 100%, respectively. On the other hand, the True Positive Rate (TPR) and False Negative Rate (FNR) are 75% and 25%, respectively. These results are relevant from the perspective of the development of a simple, automatic, real-time method for the discrimination of cimiciato-infected hazelnuts in the processing industry.

Real-time lumbosacral joint loading estimation in exoskeleton-assisted lifting conditions via electromyography-driven musculoskeletal models

Lumbar joint compression forces have been linked to the development of chronic low back pain, which is specially present in occupational environments. Offline methodologies for lumbosacral joint compression force estimation are not commonly integrated in occupational or medical applications due to the highly time-consuming and complex post-processing procedures. Hence, applications such as real-time adjustment of assistive devices (i.e., back-support exoskeletons) for optimal modulation of compression forces remains unfeasible. Here, we present a real-time electromyography (EMG)-driven musculoskeletal model, capable of estimating accurate lumbosacral joint moments and plausible compression forces. Ten participants performed box-lifting tasks (5 and 15 kg) with and without the Laevo Flex back-support exoskeleton using squat and stoop lifting techniques. Lumbosacral kinematics and EMGs from abdominal and thoracolumbar muscles were used to drive, in real-time, subject-specific EMG-driven models, and estimate lumbosacral joint moments and compression forces. Real-time EMG-model derived moments showed high correlations (R2 = 0.76 - 0.83) and estimation errors below 30% with respect to reference inverse dynamic moments. Compared to unassisted lifting conditions, exoskeleton liftings showed mean lumbosacral joint moments and compression forces reductions of 11.9 - 18.7 Nm (6 - 12% of peak moment) and 300 - 450 N (5 - 10%), respectively. Our modelling framework was capable of estimating in real-time, valid lumbosacral joint moments and compression forces in line with in vivo experimental data, as well as detecting the biomechanical effects of a passive back-support exoskeleton. Our presented technology may lead to a new class of bio-protective robots in which personalized assistance profiles are provided based on subject-specific musculoskeletal variables.

Epidemic monitoring in real-time based on dynamic grid search and Monte Carlo numerical simulation algorithm

Building upon the foundational principles of the grid search algorithm and Monte Carlo numerical simulation, this article introduces an innovative epidemic monitoring and prevention plan. The plan offers the capability to accurately identify the sources of infectious diseases and predict the final scale and duration of the epidemic. The proposed plan is implemented in schools and society, utilizing computer simulation analysis. Through this analysis, the plan enables precise localization of infection sources for various demographic groups, with an error rate of less than 3%. Additionally, the plan allows for the estimation of the epidemic cycle duration, which typically spans around 14 days. Notably, higher population density enhances fault tolerance and prediction accuracy, resulting in smaller errors and more reliable simulation outcomes. Overall, this study provides highly valuable theoretical guidance for effective epidemic prevention and control efforts.

Clinical benefit of AI-assisted lung ultrasound in a resource-limited intensive care unit

BACKGROUND

Interpreting point-of-care lung ultrasound (LUS) images from intensive care unit (ICU) patients can be challenging, especially in low- and middle- income countries (LMICs) where there is limited training available. Despite recent advances in the use of Artificial Intelligence (AI) to automate many ultrasound imaging analysis tasks, no AI-enabled LUS solutions have been proven to be clinically useful in ICUs, and specifically in LMICs. Therefore, we developed an AI solution that assists LUS practitioners and assessed its usefulness in  a low resource ICU.

METHODS

This was a three-phase prospective study. In the first phase, the performance of four different clinical user groups in interpreting LUS clips was assessed. In the second phase, the performance of 57 non-expert clinicians with and without the aid of a bespoke AI tool for LUS interpretation was assessed in retrospective offline clips. In the third phase, we conducted a prospective study in the ICU where 14 clinicians were asked to carry out LUS examinations in 7 patients with and without our AI tool and we interviewed the clinicians regarding the usability of the AI tool.

RESULTS

The average accuracy of beginners' LUS interpretation was 68.7% [95% CI 66.8-70.7%] compared to 72.2% [95% CI 70.0-75.6%] in intermediate, and 73.4% [95% CI 62.2-87.8%] in advanced users. Experts had an average accuracy of 95.0% [95% CI 88.2-100.0%], which was significantly better than beginners, intermediate and advanced users (p < 0.001). When supported by our AI tool for interpreting retrospectively acquired clips, the non-expert clinicians improved their performance from an average of 68.9% [95% CI 65.6-73.9%] to 82.9% [95% CI 79.1-86.7%], (p < 0.001). In prospective real-time testing, non-expert clinicians improved their baseline performance from 68.1% [95% CI 57.9-78.2%] to 93.4% [95% CI 89.0-97.8%], (p < 0.001) when using our AI tool. The time-to-interpret clips improved from a median of 12.1 s (IQR 8.5-20.6) to 5.0 s (IQR 3.5-8.8), (p < 0.001) and clinicians' median confidence level improved from 3 out of 4 to 4 out of 4 when using our AI tool.

CONCLUSIONS

AI-assisted LUS can help non-expert clinicians in an LMIC ICU improve their performance in interpreting LUS features more accurately, more quickly and more confidently.

Cooperative simultaneous inversion of satellite-based real-time PM2.5 and ozone levels using an improved deep learning model with attention mechanism

Ground-level fine particulate matter (PM_2.5) and ozone (O₃) are air pollutants that can pose severe health risks. Surface PM_2.5 and O₃ concentrations can be monitored from satellites, but most retrieval methods retrieve PM_2.5 or O₃ separately and disregard the shared information between the two air pollutants, for example due to common emission sources. Using surface observations across China spanning 2014-2021, we found a strong relationship between PM_2.5 and O₃ with distinct spatiotemporal characteristics. Thus, in this study, we propose a new deep learning model called the Simultaneous Ozone and PM_2.5 inversion deep neural Network (SOPiNet), which allows for daily real-time monitoring and full coverage of PM_2.5 and O₃ simultaneously at a spatial resolution of 5 km. SOPiNet employs the multi-head attention mechanism to better capture the temporal variations in PM_2.5 and O₃ based on previous days' conditions. Applying SOPiNet to MODIS data over China in 2022, using 2019-2021 to construct the network, we found that simultaneous retrievals of PM_2.5 and O₃ improved the performance compared with retrieving them independently: the temporal R² increased from 0.66 to 0.72 for PM_2.5, and from 0.79 to 0.82 for O₃. The results suggest that near-real time satellite-based air quality monitoring can be improved by simultaneous retrieval of different but related pollutants. The codes of SOPiNet and its user guide are freely available online at https://github.com/RegiusQuant/ESIDLM.

Using drone soundings to study the impacts and compositions of plumes from a gigantic coal-fired power plant

The immense impacts of coal-fired power plant plumes on the atmospheric environment have caused great concern linked to climate and health issues. However, studies on the field observations of aerial plumes are relatively limited, mainly due to the lack of suitable plume observation tools and techniques. In this study, we use a multicopter unmanned aerial vehicle (UAV) sounding technique to study the influences of the aerial plumes of the world's fourth-largest coal-fired power plant on the atmospheric physical/chemical conditions and air quality. A set of species, including 106 volatile organic compounds (VOCs), CO, CO₂, CH₄, PM_2.5, and O₃, and meteorological variables of temperature (T), specific humidity (SH), and wind data, are collected by the UAV sounding technique. The results reveal that the large-scale plumes of the coal-fired power plant cause local temperature inversion and humidity changes, and even affect the dispersion of pollutants below. The chemical compositions of coal-fired power plant plumes are significantly different from those of another ubiquitous vehicular source. High fractions of ethane, ethene, and benzene and low fractions of n-butane and isopentane found in plumes could serve as the key features to help distinguish the influences of coal-fired power plant plumes from other pollution sources in a particular area. By taking the ratios of pollutants (e.g., PM_2.5, CO, CH₄, and VOCs) to CO₂ in plumes and the CO₂ emission amounts of the power plant into calculation, we enable the easy quantification of the specific pollutant emissions released from power plant plumes to the atmosphere. In summary, observation by using drone soundings dissecting the aerial plumes provides a new methodology that allows aerial plumes to be readily detected and characterized. Furthermore, the influences of the plumes on the atmospheric physical/chemical conditions and air quality can be assessed rather straightforwardly, which was not easily achievable in the past.

Real-time simulations of ADF STEM probe position-integrated scattering cross-sections for single element fcc crystals in zone axis orientation using a densely connected neural network

Quantification of annular dark field (ADF) scanning transmission electron microscopy (STEM) images in terms of composition or thickness often relies on probe-position integrated scattering cross sections (PPISCS). In order to compare experimental PPISCS with theoretically predicted ones, expensive simulations are needed for a given specimen, zone axis orientation, and a variety of microscope settings. The computation time of such simulations can be in the order of hours using a single GPU card. ADF STEM simulations can be efficiently parallelized using multiple GPUs, as the calculation of each pixel is independent of other pixels. However, most research groups do not have the necessary hardware, and, in the best-case scenario, the simulation time will only be reduced proportionally to the number of GPUs used. In this manuscript, we use a learning approach and present a densely connected neural network that is able to perform real-time ADF STEM PPISCS predictions as a function of atomic column thickness for most common face-centered cubic (fcc) crystals (i.e., Al, Cu, Pd, Ag, Pt, Au and Pb) along [100] and [111] zone axis orientations, root-mean-square displacements, and microscope parameters. The proposed architecture is parameter efficient and yields accurate predictions for the PPISCS values for a wide range of input parameters that are commonly used for aberration-corrected transmission electron microscopes.

Real-time face mask position recognition system based on MobileNet model

COVID-19 is a highly contagious disease that was first identified in 2019, and has since taken more than six million lives world wide till date, while also causing considerable economic, social, cultural and political turmoil. As a way to limit its spread, the World Health Organization and medical experts have advised properly wearing face masks, social distancing and hand sanitization, besides vaccination. However, people wear masks sometimes uncovering their mouths and/or noses consciously or unconsciously, thereby lessening the effectiveness of the protection they provide. A system capable of automatic recognition of face mask position could alert and ensure that an individual is wearing a mask properly before entering a crowded public area and putting themselves and others at risk. We first develop and publicly release a dataset of face mask images, which are collected from 391 individuals of different age groups and gender. Then, we study six different architectures of pre-trained deep learning models, and finally propose a model developed by fine tuning the pre-trained state of the art MobileNet model. We evaluate the performance (accuracy, F1-score, and Cohen's Kappa) of this model on the proposed dataset and MaskedFace-Net, a publicly available synthetic dataset created by image editing. Its performance is also compared to other existing methods. The proposed MobileNet is found as the best model providing an accuracy, F1-score, and Cohen's Kappa of 99.23%, 99.22%, and 99.19%, respectively for face mask position recognition. It outperforms the accuracy of the best existing model by about 2%. Finally, an automatic face mask position recognition system has been developed, which can recognize if an individual is wearing a mask correctly or incorrectly. The proposed model performs very well with no drop in recognition accuracy from real images captured by a camera.

EEG-LLAMAS: A low-latency neurofeedback platform for artifact reduction in EEG-fMRI

Simultaneous EEG-fMRI is a powerful multimodal technique for imaging the brain, but its use in neurofeedback experiments has been limited by EEG noise caused by the MRI environment. Neurofeedback studies typically require analysis of EEG in real time, but EEG acquired inside the scanner is heavily contaminated with ballistocardiogram (BCG) artifact, a high-amplitude artifact locked to the cardiac cycle. Although techniques for removing BCG artifacts do exist, they are either not suited to real-time, low-latency applications, such as neurofeedback, or have limited efficacy. We propose and validate a new open-source artifact removal software called EEG-LLAMAS (Low Latency Artifact Mitigation Acquisition Software), which adapts and advances existing artifact removal techniques for low-latency experiments. We first used simulations to validate LLAMAS in data with known ground truth. We found that LLAMAS performed better than the best publicly-available real-time BCG removal technique, optimal basis sets (OBS), in terms of its ability to recover EEG waveforms, power spectra, and slow wave phase. To determine whether LLAMAS would be effective in practice, we then used it to conduct real-time EEG-fMRI recordings in healthy adults, using a steady state visual evoked potential (SSVEP) task. We found that LLAMAS was able to recover the SSVEP in real time, and recovered the power spectra collected outside the scanner better than OBS. We also measured the latency of LLAMAS during live recordings, and found that it introduced a lag of less than 50 ms on average. The low latency of LLAMAS, coupled with its improved artifact reduction, can thus be effectively used for EEG-fMRI neurofeedback. A limitation of the method is its use of a reference layer, a piece of EEG equipment which is not commercially available, but can be assembled in-house. This platform enables closed-loop experiments which previously would have been prohibitively difficult, such as those that target short-duration EEG events, and is shared openly with the neuroscience community.

Gaussian Processes for real-time 3D motion and uncertainty estimation during MR-guided radiotherapy

Respiratory motion during radiotherapy causes uncertainty in the tumor's location, which is typically addressed by an increased radiation area and a decreased dose. As a result, the treatments' efficacy is reduced. The recently proposed hybrid MR-linac scanner holds the promise to efficiently deal with such respiratory motion through real-time adaptive MR-guided radiotherapy (MRgRT). For MRgRT, motion-fields should be estimated from MR-data and the radiotherapy plan should be adapted in real-time according to the estimated motion-fields. All of this should be performed with a total latency of maximally 200 ms, including data acquisition and reconstruction. A measure of confidence in such estimated motion-fields is highly desirable, for instance to ensure the patient's safety in case of unexpected and undesirable motion. In this work, we propose a framework based on Gaussian Processes to infer 3D motion-fields and uncertainty maps in real-time from only three readouts of MR-data. We demonstrated an inference frame rate up to 69 Hz including data acquisition and reconstruction, thereby exploiting the limited amount of required MR-data. Additionally, we designed a rejection criterion based on the motion-field uncertainty maps to demonstrate the framework's potential for quality assurance. The framework was validated in silico and in vivo on healthy volunteer data (n=5) acquired using an MR-linac, thereby taking into account different breathing patterns and controlled bulk motion. Results indicate end-point-errors with a 75th percentile below 1 mm in silico, and a correct detection of erroneous motion estimates with the rejection criterion. Altogether, the results show the potential of the framework for application in real-time MR-guided radiotherapy with an MR-linac.

Holding times to maintain quasi-regular headways and reduce real-time bus bunching

Real-time control strategies deal with the day's dynamics in bus rapid transit systems. This work focuses on minimizing the number of buses of the same line cruising head-to-tail or arriving at a stop simultaneously by implementing bus holding times at the stops as a control strategy. We propose a new mathematical model to determine the bus holding times. It has quadratic constraints but a linear objective function that minimizes the bus bunching penalties. We also propose a beam-search heuristic to reduce computational solution time to solve large instances. Experimental results on a bus rapid transit system simulation in Monterrey, Mexico, show a bus bunching reduction of 45% compared to the case without optimization. Moreover, passenger waiting times are reduced by 30% in some scenarios. For real-world instances with 60 buses, the beam-search approach provides solutions with an optimality gap of less than 5% in less than 3 s.

A non-intrusive and reactive architecture to support real-time ETL processes in data warehousing environments

Nowadays, organizations are very interested to gather data for strategic decision-making. Data are disposable in operational sources, which are distributed, heterogeneous, and autonomous. These data are gathered through ETL processes, which occur traditionally in a pre-defined time, that is, once a day, once a week, once a month or in a specific period of time. On the other hand, there are special applications for which data needs to be obtained in a faster way and sometimes even immediately after the data are generated in the operation data sources, such as health systems and digital agriculture. Thus, the conventional ETL process and the disposable techniques are incapable of making the operational data delivered in real-time, providing low latency, high availability, and scalability. As our proposal, we present an innovative architecture, named Data Magnet, to cope with real-time ETL processes. The experimental tests performed in the digital agriculture domain using real and synthetic data showed that our proposal was able to deal in real-time with the ETL process. The Data Magnet provided great performance, showing an almost constant elapsed time for growing data volumes. Besides, Data Magnet provided significant performance gains over the traditional trigger technique.

Towards European automatic bioaerosol monitoring: Comparison of 9 automatic pollen observational instruments with classic Hirst-type traps

To benefit allergy patients and the medical practitioners, pollen information should be available in both a reliable and timely manner; the latter is only recently possible due to automatic monitoring. To evaluate the performance of all currently available automatic instruments, an international intercomparison campaign was jointly organised by the EUMETNET AutoPollen Programme and the ADOPT COST Action in Munich, Germany (March-July 2021). The automatic systems (hardware plus identification algorithms) were compared with manual Hirst-type traps. Measurements were aggregated into 3-hourly or daily values to allow comparison across all devices. We report results for total pollen as well as for Betula, Fraxinus, Poaceae, and Quercus, for all instruments that provided these data. The results for daily averages compared better with Hirst observations than the 3-hourly values. For total pollen, there was a considerable spread among systems, with some reaching R² > 0.6 (3 h) and R² > 0.75 (daily) compared with Hirst-type traps, whilst other systems were not suitable to sample total pollen efficiently (R² < 0.3). For individual pollen types, results similar to the Hirst were frequently shown by a small group of systems. For Betula, almost all systems performed well (R² > 0.75 for 9 systems for 3-hourly data). Results for Fraxinus and Quercus were not as good for most systems, while for Poaceae (with some exceptions), the performance was weakest. For all pollen types and for most measurement systems, false positive classifications were observed outside of the main pollen season. Different algorithms applied to the same device also showed different results, highlighting the importance of this aspect of the measurement system. Overall, given the 30 % error on daily concentrations that is currently accepted for Hirst-type traps, several automatic systems are currently capable of being used operationally to provide real-time observations at high temporal resolutions. They provide distinct advantages compared to the manual Hirst-type measurements.

Determination of monophenolase activity based on backpropagation neural network analysis of three-dimensional fluorescence spectroscopy

Tyrosinase is pivotal for melanin formation. Measuring monophenolase activity is of great importance for both fundamental research and industrial applications. For the first time, a backpropagation (BP) artificial neural network with three-dimensional fluorescence spectroscopy was applied for the real-time determination of tyrosinase monophenolase activity. Principal component analysis (PCA) was utilized for the dimension reduction of three-dimensional fluorescence data. The four principal components served as inputs for the neural network. Network parameters were optimized using a genetic algorithm (GA). BP learning algorithm was applied to train the network model to determine tyrosine levels in a binary mixture containing tyrosine and L-DOPA without any chemical separation. The time course of tyrosine consumption by monophenolase was determined to calculate the initial velocity of the enzymatic reaction. The limit of detection of the monophenolase assay was 0.0615 U·mL^-1. This combined strategy of PCA, GAs, and BP artificial neural networks for three-dimensional fluorescence spectroscopy was efficient for the real-time and in-situ determination of monophenolase activity in a cascade reaction.

Performance estimation for the memristor-based computing-in-memory implementation of extremely factorized network for real-time and low-power semantic segmentation

Nowadays many semantic segmentation algorithms have achieved satisfactory accuracy on von Neumann platforms (e.g., GPU), but the speed and energy consumption have not meet the high requirements of certain edge applications like autonomous driving. To tackle this issue, it is of necessity to design an efficient lightweight semantic segmentation algorithm and then implement it on emerging hardware platforms with high speed and energy efficiency. Here, we first propose an extremely factorized network (EFNet) which can learn multi-scale context information while preserving rich spatial information with reduced model complexity. Experimental results on the Cityscapes dataset show that EFNet achieves an accuracy of 68.0% mean intersection over union (mIoU) with only 0.18M parameters, at a speed of 99 frames per second (FPS) on a single RTX 3090 GPU. Then, to further improve the speed and energy efficiency, we design a memristor-based computing-in-memory (CIM) accelerator for the hardware implementation of EFNet. It is shown by the simulation in DNN+NeuroSim V2.0 that the memristor-based CIM accelerator is ∼63× (∼4.6×) smaller in area, at most ∼9.2× (∼1000×) faster, and ∼470× (∼2400×) more energy-efficient than the RTX 3090 GPU (the Jetson Nano embedded development board), although its accuracy slightly decreases by 1.7% mIoU. Therefore, the memristor-based CIM accelerator has great potential to be deployed at the edge to implement lightweight semantic segmentation models like EFNet. This study showcases an algorithm-hardware co-design to realize real-time and low-power semantic segmentation at the edge.

A multimodal facial cues based engagement detection system in e-learning context using deep learning approach

Due to the COVID-19 crisis, the education sector has been shifted to a virtual environment. Monitoring the engagement level and providing regular feedback during e-classes is one of the major concerns, as this facility lacks in the e-learning environment due to no physical observation of the teacher. According to present study, an engagement detection system to ensure that the students get immediate feedback during e-Learning. Our proposed engagement system analyses the student's behaviour throughout the e-Learning session. The proposed novel approach evaluates three modalities based on the student's behaviour, such as facial expression, eye blink count, and head movement, from the live video streams to predict student engagement in e-learning. The proposed system is implemented based on deep-learning approaches such as VGG-19 and ResNet-50 for facial emotion recognition and the facial landmark approach for eye-blinking and head movement detection. The results from different modalities (for which the algorithms are proposed) are combined to determine the EI (engagement index). Based on EI value, an engaged or disengaged state is predicted. The present study suggests that the proposed facial cues-based multimodal system accurately determines student engagement in real time. The experimental research achieved an accuracy of 92.58% and showed that the proposed engagement detection approach significantly outperforms the existing approaches.

Automatic real-time monitoring of fungal spores: the case of Alternaria spp

We present the first implementation of the monitoring of airborne fungal spores in real-time using digital holography. To obtain observations of Alternaria spp. spores representative of their airborne stage, we collected events measured in the air during crop harvesting in a contaminated potato field, using a Swisens Poleno device. The classification algorithm used by MeteoSwiss for operational pollen monitoring was extended by training the system using this additional dataset. The quality of the retrieved concentrations is evaluated by comparison with parallel measurements made with a manual Hirst-type trap. Correlations between the two measurements are high, especially over the main dispersion period of Alternaria spp., demonstrating the potential for automatic real-time monitoring of fungal spores.

Real-time digital data of international passengers will shine in the precaution of epidemics

International movement plays an important role in spatial spread of infectious diseases. Here, we share two successful COVID-19 interventions based on real-time digital information collected from international passengers, which have been performed in Greece and China respectively. Both of the interventions demonstrated good performance and showed the potential of real-time digital data in containing the spread. However, several key points should not be ignored when we promote similar strategies.

Video based oil palm ripeness detection model using deep learning

Research on oil palm detection has been carried out for years, but there are only a few research that have conducted research using video datasets and only focus on development using non-sequential image. The use of the video dataset aims to adjust to the detection conditions carried out in real time so that it can automatically harvest directly from oil palm trees to increase efficiency in harvesting. To solve this problem, in this research, we develop an object detection model using a video dataset in training and testing. We used the 3 series YOLOv4 architecture to develop the model using video. Model development is done by means of hyperparameter tuning and frozen layer with data augmentation consisting of photometric and geometric augmentation experiment. To validate the outcomes of the YOLOv4 model development, a comparison of SSD-MobileNetV2 FPN and EfficientDet-D0 was performed. The results obtained show that YOLOv4-Tiny 3L is the most suitable architecture for use in real time object detection conditions with an mAP of 90.56% for single class category detection and 70.21% for multi class category detection with a detection speed of almost 4× faster than YOLOv4-CSPDarknet53, 5× faster than SSD-MobileNetV2 FPN, and 9× faster than EfficientDet-D0.

Development of diagnostic tests for pathogen identification and detection of antimicrobial resistance on WHO global priority pathogens using modular real-time nucleic acid amplification test

BACKGROUND

Concerns regarding antimicrobial resistance (AMR) have resulted in the World Health Organization (WHO) designating so-called global priority pathogens (GPPs). However, little discussion has focused on the diagnosis of GPPs. To enable the simultaneous identification of pathogens and AMR, we developed a modular real-time nucleic acid amplification test (MRT-NAAT).

METHODS

Sequence-specific primers for each modular unit for MRT-NAAT pathogen identification and AMR sets were designed. The composition of the reaction mixture and the real-time PCR program were unified irrespective of primer type so to give MRT-NAAT modularity. Standard strains and clinical isolates were used to evaluate the performance of MRT-NAAT by real-time PCR and melting curve analysis. Probit analysis for the MRT-NAAT pathogen identification set was used to assess the limit of detection (LoD).

RESULTS

The MRT-NAAT pathogen identification set was made up of 15 modular units 109-199 bp in product size and with a T_ms of 75.5-87.5 °C. The LoD was < 15.548 fg/μL, and nine modular units successfully detected the target pathogens. The MRT-NAAT AMR set included 24 modular units 65-785 bp in product size with a T_ms of 75.5-87.5 °C; it showed high performance for detecting GPP target genes and variants.

CONCLUSIONS

MRT-NAAT enables pathogen identification and AMR gene detection and is time-effective. By unifying the reaction settings of each modular unit, the modularity where combinations of primers can be used according to need could be achieved. This would greatly help in reflecting the researcher's need and the AMR status of a certain region while successfully detecting pathogens and AMR genes.

DSENet: Directional Signal Extraction Network for Hearing Improvement on Edge Devices

In this paper, we propose a directional signal extraction network (DSENet). DSENet is a low-latency, real-time neural network that, given a reverberant mixture of signals captured by a microphone array, aims at extracting the reverberant signal whose source is located within a directional region of interest. If there are multiple sources situated within the directional region of interest, DSENet will aim at extracting a combination of their reverberant signals. As such, the formulation of DSENet circumvents the well-known crosstalk problem in beamforming while providing an alternative and perhaps more practical approach to other spatially constrained signal extraction methods proposed in the literature. DSENet is based on a computationally efficient and low-distortion linear model formulated in the time domain. As a result, an important application of our work is hearing improvement on edge devices. Simulation results show that DSENet outperforms oracle beamformers, as well as state-of-the-art in low-latency causal speech separation, while incurring a system latency of only 4 ms. Additionally, DSENet has been successfully deployed as a real-time application on a smartphone.