Magnetoencephalography (MEG) measures human neural activity non-invasively with spatio-temporal precision, and has been foundational in enabling impactful discoveries in cognitive neuroscience. New on-scalp MEG sensor technologies, such as OPM-MEG, offer the opportunity to capture more information about the neuronal magnetic fields with higher sensitivity to more complex, higher order spatial components, leading to improved source localization. The accuracy of MEG and OPM-MEG source localization relies on data preprocessing techniques to isolate the neuronal fields from other magnetic and biomagnetic sources through signal space separation, rejection, and suppression methods. Current preprocessing methods risk rejecting brain signals of interest, or can spread sensor noise artifacts unknowingly. Here we propose a novel preprocessing method for MEG, and test the extent to which it overcomes limitations of prior methods. Specifically, we derive, apply, and assess a novel signal space separation (SSS) method with Foster's inverse, a weighted matrix inversion protocol that can utilize information about the MEG sensor noise and artifacts to reconstruct neuronal activity. With simulations, phantom head cryogenic MEG recordings, and subject recordings with two OPM-MEG systems, we show that Foster's inverse with SSS offers a more robust and stable reconstruction of the neuronal magnetic fields, especially in the face of sensor noise and artifacts. As the field of cognitive neuroscience continues to embrace MEG and OPM-MEG, Foster's inverse with SSS offers a robust and powerful data preprocessing technique for reducing noise and improving source localization of the underlying neuronal currents.

Electromyography (EMG) is fundamental to clinical assessment, rehabilitation, neuromuscular research, and human-machine interfaces. Despite decades of use, no widely adopted standard exists for organizing and sharing EMG data, limiting reusability and large-scale data aggregation. We present EMG-BIDS, an extension to the Brain Imaging Data Structure (BIDS) that standardizes the organization of EMG recordings. EMG-BIDS addresses challenges unique to EMG, including diverse electrode types (surface or intramuscular, single channel to high-density arrays), heterogeneous electrode placements across anatomical locations, montages (e.g., monopolar or bipolar sensor designs), and the critical need for transparent documentation of sensor positioning. The specification introduces hierarchical coordinate systems that link local electrode grids to anatomical landmarks, enabling precise and reproducible placement documentation. EMG-BIDS is now part of BIDS as of version 1.11.0, supported by existing tools, including MNE-BIDS and EEGLAB. We demonstrate the specification through public datasets, including high-density surface EMG recordings. EMG-BIDS provides the foundation for FAIR (Findable, Accessible, Interoperable, Reusable) EMG data sharing, enabling meta-analyses, multi-site studies, and machine learning applications that require standardized, well-documented datasets.

We address three key challenges in learning continuous kernel representations: computational efficiency, parameter efficiency, and spectral bias. Continuous kernels have shown significant potential, but their practical adoption is often limited by high computational and memory demands. Additionally, these methods are prone to spectral bias, which impedes their ability to capture high-frequency details. To overcome these limitations, we propose a novel approach that leverages sparse learning in the Fourier domain. Our method enables the efficient scaling of continuous kernels, drastically reduces computational and memory requirements, and mitigates spectral bias by exploiting the Gibbs phenomenon.

Adolescents are particularly tuned to social cues, but how does this sensitivity affect their learning? This study examined whether social feedback conveyed through human faces modulates implicit learning of auditory speech patterns. Magnetoencephalography (MEG) was used to assess neural measures of learning in response to auditory linguistic input as participants engaged in a computer-controlled guessing task and received incidental social feedback in the form of smiling (positive) or frowning (negative) faces. The results showed that smiling faces produced neural response patterns consistent with learning statistical regularities in left-hemisphere regions involved in social and language processing, including the amygdala, anterior auditory cortex, inferior frontal cortex, and insula, whereas no significant learning-related effect was observed in the right hemisphere. In contrast, negative social feedback (frowning faces) did not produce a significant learning-related effect in either hemisphere, indicating no reliable neural effect of learning under negative feedback. Direct comparisons for positive and negative feedback revealed that the neural effect of learning was significantly larger for positive than for negative feedback, with this effect remaining robust after correction for multiple comparisons in the left insula. These results occurred despite the visual cues being incidental and unrelated to the auditory input. Together, the findings suggest that incidental social feedback from human faces influences neural mechanisms of learning during adolescence.

Most scientists need software to perform their research (Barker et al., 2020;Carver et al., 2022;Hettrick, 2014;Hettrick et al., 2014;Switters & Osimo, 2019), and neuroscientists are no exception. Whether we work with reaction times, electrophysiological signals, or magnetic resonance imaging data, we rely on software to acquire, analyze, and statistically evaluate the raw data we obtain-or to generate such data if we work with simulations. In recent years, there has been a shift toward relying on free, open-source scientific software (FOSSS) for neuroscience data analysis (Poldrack et al., 2019), in line with the broader open science movement in academia (McKiernan et al., 2016) and wider industry trends (Eghbal, 2016). Importantly, FOSSS is typically developed by working scientists (not professional software developers), which sets up a precarious situation given the nature of the typical academic workplace wherein academics, especially in their early careers, are on short- and fixed-term contracts. In this paper, we argue that the existing ecosystem of neuroscientific open-source software is brittle, and discuss why and how the neuroscience community needs to come together to ensure a healthy software ecosystem to the benefit of all.

Objective.The reliability of biomagnetic measurements is improved by data processing techniques like the signal space separation (SSS) method, which transforms multichannel signals into device-independent channels with separate components for internal biomagnetic and external interference signals based on sensor geometry. Newer on-scalp sensors, such as optically-pumped magnetometers (OPMs), have recently been deployed in magnetoencephalography (MEG) systems, bringing a need for refined SSS variants to capture the potentially improved spatial resolution provided by the on-scalp sensors. Standard single-origin SSS may fail to capture the full brain-space when the sensors are on scalp. In this paper, we propose potential solutions to this problem including novel multi-origin SSS (mSSS). With multiple optimized origins and radii used together, the basis can span the brain-space without encroaching on the sensor space. Other adaptations to SSS include vector spheroidal harmonics, which create signal space expansions using ellipsoidal geometry to model the brain-space. This adaptation is further modified to combine an interior spheroidal with exterior single-SSS.Approach.Focusing on two-origin mSSS, the spheroidal constructions and the single-origin SSS are investigated with simulated data from an internal current dipole source coupled with an external interference signal with geometry from the 432-channel Kernel Flux OPM system, the 306-channel MEGIN/Elekta Neuromag SQUID system, and the 192-Channel Triaxial QuSpin OPM system. Finally, each variant is used to process collected data including auditory evoked data measured at the University of Washington with the Kernel Flux OPM system, previously recorded empty-room data collected in a lightly-shielded magnetically shielded room with 192-channel third generation triaxial QuSpin Zero Field Magnetometers, and publicly available single-subject audiovisual data collected with an 86-Channel dual-axis QuSpin OPM system at the University College London.Main results.The mSSS method has comparable or better stability to the SSS method in all sensor geometries and reconstructs interior simulated signals while successfully suppressing exterior interference, and performs better in simulated cases with variably placed on-scalp MEG systems. Additionally, results with Kernel and QuSpin data show the mSSS basis provides a lower noise floor than other SSS variants and had the best performance with on-scalp systems, even with low-channel-count OPM systems.Significance.With on-scalp MEG systems becoming more widely available, the MEG community needs updated data analysis techniques. mSSS is a straightforward and robust modification to the SSS method which functions for novel on-scalp sensor systems without needing drastic modification to the underlying mathematical method.

We introduce a differentiable statistical moment aggregation layer, enabling networks to learn the optimal method of statistical moment pooling for automatic modulation classification. Statistical pooling, a cornerstone of convolutional networks, consolidates activations into fixed-length representations. Traditionally, this entails mean, variance, and higher-ordered statistics pooling defined as fixed hyperparameters. By enabling the statistics layer to become differentiable, networks are able to optimize the method of statistical aggregations, transcending predefined hyperparameters. With our approach, the statistical moment order is differentiable. Our results demonstrate learned statistical moments are able to outperform fixed-moments—improving modulation classification performance of a time-domain signal. 1

Physically unclonable functions (PUFs) are designed to act as device 'fingerprints.' Given an input challenge, the PUF circuit should produce an unpredictable response for use in situations such as root-of-trust applications and other hardware-level cybersecurity applications. PUFs are typically subcircuits present within integrated circuits (ICs), and while conventional IC PUFs are well-understood, several implementations have proven vulnerable to malicious exploits, including those perpetrated by machine learning (ML)-based attacks. Such attacks can be difficult to prevent because they are often designed to work even when relatively few challenge-response pairs are known in advance. Hence the need for both more resilient PUF designs and analysis of ML-attack susceptibility. Previous work has developed a PUF for photonic integrated circuits (PICs). A PIC PUF not only produces unpredictable responses given manufacturing-introduced tolerances, but is also less prone to electromagnetic radiation eavesdropping attacks than a purely electronic IC PUF. In this work, we analyze the resilience of the proposed photonic PUF when subjected to ML-based attacks. Specifically, we describe a computational PUF model for producing the large datasets required for training ML attacks; we analyze the quality of the model; and we discuss the modeled PUF's susceptibility to ML-based attacks. We find that the modeled PUF generates distributions that resemble uniform white noise, explaining the exhibited resilience to neural-network-based attacks designed to exploit latent relationships between challenges and responses. Preliminary analysis suggests that the PUF exhibits similar resilience to generative adversarial networks, and continued development will show whether more-sophisticated ML approaches better compromise the PUF and -- if so -- how design modifications might improve resilience.

Physically unclonable functions (PUFs) identify integrated circuits using nonlinearly-related challenge-response pairs (CRPs). Ideally, the relationship between challenges and corresponding responses is unpredictable, even if a subset of CRPs is known. Previous work developed a photonic PUF offering improved security compared to non-optical counterparts. Here, we investigate this PUF's susceptibility to Multiple-Valued-Logic-based machine learning attacks. We find that approximately 1,000 CRPs are necessary to train models that predict response bits better than random chance. Given the significant challenge of acquiring a vast number of CRPs from a photonic PUF, our results demonstrate photonic PUF resilience against such attacks.

Objective.Measures of functional connectivity (FC) can elucidate which cortical regions work together in order to complete a variety of behavioral tasks. This study's primary objective was to expand a previously published model of measuring FC to include multiple subjects and several regions of interest. While FC has been more extensively investigated in vision and other sensorimotor tasks, it is not as well understood in audition. The secondary objective of this study was to investigate how auditory regions are functionally connected to other cortical regions when attention is directed to different distinct auditory stimuli.Approach.This study implements a linear dynamic system (LDS) to measure the structured time-lagged dependence across several cortical regions in order to estimate their FC during a dual-stream auditory attention task.Results.The model's output shows consistent functionally connected regions across different listening conditions, indicative of an auditory attention network that engages regardless of endogenous switching of attention or different auditory cues being attended.Significance.The LDS implemented in this study implements a multivariate autoregression to infer FC across cortical regions during an auditory attention task. This study shows how a first-order autoregressive function can reliably measure functional connectivity from M/EEG data. Additionally, the study shows how auditory regions engage with the supramodal attention network outlined in the visual attention literature.

Automated assessing prosody of oral reading fluency presents challenges due to the inherent difficulty of quantifying prosody. This study proposed and evaluated an approach focusing on specific prosodic features using a deep-learning neural network. The current work focuses on cross-domain performance, researching how generalizable the prosody scoring is across students and text passages. The results demonstrated that the model with selected prosodic features had better cross-domain performance with an accuracy of 62.5% compared to 57% from the previous research. Our findings also indicate that students’ reading patterns influence cross-domain performance more than specific text passage patterns. In other words, letting the student read at least one passage is more important than having others read all passage texts. The specific prosodic features had a high generalization to capture the typical prosody characteristics for achieving a satisfactorily high accuracy and classification agreement rate. This result provides valuable information for developing future automated scoring algorithms of prosody. This study is an essential demonstration of estimating the prosody score using fewer selected features, which would be more efficient and interpretable.

Random number generators (RNG) are essential elements in many cryptographic systems. True random number generators (TRNG) rely upon sources of randomness from natural processes such as those arising from quantum mechanics phenomena. We demonstrate that a quantum computer can serve as a high-quality, weakly random source for a generalized user-defined probability mass function (PMF). Specifically, QC measurement implements the process of variate sampling according to a user-specified PMF resulting in a word comprised of electronic bits that can then be processed by an extractor function to address inaccuracies due to non-ideal quantum gate operations and other system biases. We introduce an automated and flexible method for implementing a TRNG as a programmed quantum circuit that executes on commercially-available, gate-model quantum computers. The user specifies the desired word size as the number of qubits and a definition of the desired PMF. Based upon the user specification of the PMF, our compilation tool automatically synthesizes the desired TRNG as a structural OpenQASM file containing native gate operations that are optimized to reduce the circuit's quantum depth. The resulting TRNG provides multiple bits of randomness for each execution/measurement cycle; thus, the number of random bits produced in each execution is limited only by the size of the QC. We provide experimental results to illustrate the viability of this approach.

Scintillators are the primary devices used for radiation detection., especially at national borders and other ports of entry. Reducing the size of these detectors for placement on mobile devices such as drones can allow for better detection and localization of radiation sources. Detection of radiation with supervised machine learning can be a challenge when looking for previously unobserved radiation sources. Therefore, anomaly detection methods are investigated. In this work we employ adversarial autoencoders trained to classify spectra from radioactive sources as either background or anomalous. This allows the model to detect anomalies regardless of radiation source and outperform supervised methods on newly encountered sources.

Purpose The purpose of this study was to describe practices and experiences of rurally oriented physician assistant (PA) training programs in providing rural clinical training to PA students. Methods A survey of PA program directors (PDs) included questions about program characteristics, student and clinical preceptor (CP) recruitment in rural areas, and barriers to, and facilitators of, rural clinical training. Programs that considered rural training “very important” to their goals were identified. We interviewed PDs from rurally oriented programs about their rural clinical training and rural CPs about their experiences training PA students for rural practice. We identified key themes through content analysis. Results Of 178 programs surveyed, 113 (63.5%) responded, 61 (54.0%) of which were rurally oriented and more likely than other programs to recruit rural students or those with rural practice interests and to address rural issues in didactic curriculum. The 13 PDs interviewed linked successful rural training to finding and supporting rural preceptors who enjoy teaching and helping students understand rural communities. The 13 rural CPs identified enthusiastic and rurally interested students as key elements to successful rural training. Interviewees identified systemic barriers to rural training, including student housing, decreased productivity, competition for training slots, and administrative burden. Conclusions Physician assistant students can be coached to capitalize on their rural clinical experiences. Knowing how to “jump in” to rotations and having genuine interest in the community are particularly important. Student housing, competition for training slots, and lack of financial incentives are major system-level challenges for sustaining and increasing the availability of PA rural clinical training.

We propose a method to improve steganography by increasing the resilience of stego-media to discovery through steganalysis. Our approach enhances a class of steganographic approaches through the inclusion of a steganographic assistant convolutional neural network (SA-CNN). Previous research showed success in discovering the presence of hidden information within stego-images using trained neural networks as steganalyzers that are applied to stego-images. Our results show that such steganalyzers are less effective when SA-CNN is employed during the generation of a stego-image. We also explore the advantages and disadvantages of representing all the possible outputs of our SA-CNN within a smaller, discrete space, rather than a continuous space. Our SA-CNN enables certain classes of parametric steganographic algorithms to be customized based on characteristics of the cover media in which information is to be embedded. Thus, SA-CNN is adaptive in the sense that it enables the core steganographic algorithm to be especially configured for each particular instance of cover media. Experimental results are provided that employ a recent steganographic technique, S-UNIWARD, both with and without the use of SA-CNN. We then apply both sets of stego-images, those produced with and without SA-CNN, to an exmaple steganalyzer, Yedroudj-Net, and we compare the results. We believe that this approach for the integration of neural networks with hand-crafted algorithms increases the reliability and adaptability of steganographic algorithms.

We use machine learning to evaluate surgical skill from videos during the tumor resection and renography steps of a robotic assisted partial nephrectomy (RAPN). This expands previous work using synthetic tissue to include actual surgeries. We investigate cascaded neural networks for predicting surgical proficiency scores (OSATS and GEARS) from RAPN videos recorded from the DaVinci system. The semantic segmentation task generates a mask and tracks the various surgical instruments. The movements from the instruments found via semantic segmentation are processed by a scoring network that regresses (predicts) GEARS and OSATS scoring for each subcategory. Overall, the model performs well for many subcategories such as force sensitivity and knowledge of instruments of GEARS and OSATS scoring, but can suffer from false positives and negatives that would not be expected of human raters. This is mainly attributed to limited training data variability and sparsity.

Reducing the size of intermediate feature maps within various neural network architectures is critical for generalization performance, and memory and computational complexity. Until recently, most methods required downsampling rates (i.e., decimation) to be predefined and static during training, with optimal downsampling rates requiring a vast hyper-parameter search. Recent work has proposed a novel and differentiable method for learning strides named DiffStride which uses the discrete Fourier transform (DFT) to learn strides for decimation. However, in many cases the DFT does not capture signal properties as efficiently as the discrete cosine transform (DCT). Therefore, we propose an alternative method for learning decimation strides, DCT-DiffStride, as well as new regularization methods to reduce model complexity. Our work employs the DCT and its inverse as a low-pass filter in the frequency domain to reduce feature map dimensionality. Leveraging the well-known energy compaction properties of the DCT for natural signals, we evaluate DCT-DiffStride with its competitors on image and audio datasets demonstrating a favorable tradeoff in model performance and model complexity compared to competing methods. Additionally, we show DCT-DiffStride and DiffStride can be applied to data outside the natural signal domain, increasing the general applications of such methods.

The excellent temporal resolution and advanced spatial resolution of magnetoencephalography (MEG) makes it an excellent tool to study the neural dynamics underlying cognitive processes in the developing brain. Nonetheless, a number of challenges exist when using MEG to image infant populations. There is a persistent belief that collecting MEG data with infants presents a number of limitations and challenges that are difficult to overcome. Due to this notion, many researchers either avoid conducting infant MEG research or believe that, in order to collect high-quality data, they must impose limiting restrictions on the infant or the experimental paradigm. In this article, we discuss the various challenges unique to imaging awake infants and young children with MEG, and share general best-practice guidelines and recommendations for data collection, acquisition, preprocessing, and analysis. The current article is focused on methodology that allows investigators to test the sensory, perceptual, and cognitive capacities of awake and moving infants. We believe that such methodology opens the pathway for using MEG to provide mechanistic explanations for the complex behavior observed in awake, sentient, and dynamically interacting infants, thus addressing core topics in developmental cognitive neuroscience.

Objective Analyze a publicly available sample of rule-based phenotype definitions to characterize and evaluate the types of logical constructs used. Materials & Methods A sample of 33 phenotype definitions used in research and published to the Phenotype KnowledgeBase (PheKB), that are represented using Fast Healthcare Interoperability Resources (FHIR) and Clinical Quality Language (CQL) was analyzed using automated analysis of the computable representation of the CQL libraries. Results Most of the phenotype definitions include narrative descriptions and flowcharts, while few provide pseudocode or executable artifacts. Most use 4 or fewer medical terminologies. The number of codes used ranges from 5 to 6865, and value sets from 1 to 19. We found the most common expressions used were literal, data, and logical expressions. Aggregate and arithmetic expressions are the least common. Expression depth ranges from 4 to 27. Discussion Despite the range of conditions, we found that all of the phenotype definitions consisted of logical criteria, representing both clinical and operational logic, and tabular data, consisting of codes from standard terminologies and keywords for natural language processing. The total number and variety of expressions is low, which may be to simplify implementation, or authors may limit complexity due to data availability constraints. Conclusion The phenotypes analyzed show significant variation in specific logical, arithmetic and other operators, but are all composed of the same high-level components, namely tabular data and logical expressions. A standard representation for phenotype definitions should support these formats and be modular to support localization and shared logic.