Journal of neural engineering最新文献

Objective.The use of electromyogram (EMG) signals recorded from the wrist is emerging as a desirable input modality for human-machine interaction (HMI). Although forearm-based EMG has been used for decades in prosthetics, there has been comparatively little prior work evaluating the performance of wrist-based control, especially in online, user-in-the-loop studies. Furthermore, despite different motivating use cases for wrist-based control, research has mostly adopted legacy prosthesis control evaluation frameworks.Approach.Gaining inspiration from rhythm games and the Schmidt's law speed-accuracy tradeoff, this work proposes a new temporally constrained evaluation environment with a linearly increasing difficulty to compare the online usability of wrist and forearm EMG. Compared to the more commonly used Fitts' Law-style testing, the proposed environment may offer different insights for emerging use cases of EMG as it decouples the machine learning algorithm's performance from proportional control, is easily generalizable to different gesture sets, and enables the extraction of a wide set of usability metrics that describe a users ability to successfully accomplish a task at a certain time with different levels of induced stress.Main results.The results suggest that wrist EMG-based control is comparable to that of forearm EMG when using traditional prosthesis control gestures and can even be better when using fine finger gestures. Additionally, the results suggest that as the difficulty of the environment increased, the online metrics and their correlation to the offline metrics decreased, highlighting the importance of evaluating myoelectric control in real-time evaluations over a range of difficulties.Significance.This work provides valuable insights into the future design and evaluation of myoelectric control systems for emerging HMI applications.

Objective.To date, a comprehensive comparison of Riemannian decoding methods with deep convolutional neural networks for EEG-based brain-computer interfaces remains absent from published work. We address this research gap by using MOABB, The Mother Of All BCI Benchmarks, to compare novel convolutional neural networks to state-of-the-art Riemannian approaches across a broad range of EEG datasets, including motor imagery, P300, and steady-state visual evoked potentials paradigms.Approach.We systematically evaluated the performance of convolutional neural networks, specifically EEGNet, shallow ConvNet, and deep ConvNet, against well-established Riemannian decoding methods using MOABB processing pipelines. This evaluation included within-session, cross-session, and cross-subject methods, to provide a practical analysis of model effectiveness and to find an overall solution that performs well across different experimental settings.Main results.We find no significant differences in decoding performance between convolutional neural networks and Riemannian methods for within-session, cross-session, and cross-subject analyses.Significance.The results show that, when using traditional Brain-Computer Interface paradigms, the choice between CNNs and Riemannian methods may not heavily impact decoding performances in many experimental settings. These findings provide researchers with flexibility in choosing decoding approaches based on factors such as ease of implementation, computational efficiency or individual preferences.

Objective.This study aims to address the challenges associated with data-driven electroencephalography (EEG) data analysis by introducing a standardised library calledBIDSAlign. This library efficiently processes and merges heterogeneous EEG datasets from different sources into a common standard template. The goal of this work is to create an environment that allows to preprocess public datasets in order to provide data for the effective training of deep learning (DL) architectures.Approach.The library can handle both Brain Imaging Data Structure (BIDS) and non-BIDS datasets, allowing the user to easily preprocess multiple public datasets. It unifies the EEG recordings acquired with different settings by defining a common pipeline and a specified channel template. An array of visualisation functions is provided inside the library, together with a user-friendly graphical user interface to assist non-expert users throughout the workflow.Main results.BIDSAlign enables the effective use of public EEG datasets, providing valuable medical insights, even for non-experts in the field. Results from applying the library to datasets from OpenNeuro demonstrate its ability to extract significant medical knowledge through an end-to-end workflow, facilitating group analysis, visual comparison and statistical testing.Significance.BIDSAlign solves the lack of large EEG datasets by aligning multiple datasets to a standard template. This unlocks the potential of public EEG data for training DL models. It paves the way to promising contributions based on DL to clinical and non-clinical EEG research, offering insights that can inform neurological disease diagnosis and treatment strategies.

Objective: Electroencephalography (EEG) has evolved into an indispensable instrument for estimating cognitive workload in various domains. ML and DL techniques have been increasingly employed to develop accurate workload estimation and classification models based on EEG data. The goal of this systematic review is to compile the body of research on EEG workload estimation and classification using ML and DL approaches.

Methods: The PRISMA procedures were followed in conducting the review, searches were conducted through databases at SpringerLink, ACM Digital Library, IEEE Explore, PUBMED, and Science Direct from the beginning to the end of February 16, 2024. Studies were selected based on predefined inclusion criteria. Data were extracted to capture study design, participant demographics, EEG features, ML/DL algorithms, and reported performance metrics.

Results: Out of the 125 items that emerged, 33 scientific papers were fully evaluated. The study designs, participant demographics, and EEG workload measurement and categorization techniques used in the investigations differed. SVM, CNN, and hybrid networks are examples of ML and DL approaches that were often used. Analyzing the accuracy scores achieved by different ML/DL models. Furthermore, a relationship was noted between sample frequency and model accuracy, with higher sample frequencies generally leading to improved performance. The percentage distribution of ML/DL methods revealed that SVMs, CNNs, and RNNs were the most commonly utilized techniques, reflecting their robustness in handling EEG data.

Significance: The comprehensive review emphasizes how ML may be used to identify mental workload across a variety of disciplines using EEG data. Optimizing practical applications requires multimodal data integration, standardization efforts, and real-world validation studies. These systems will also be further improved by addressing ethical issues and investigating new EEG properties, which will improve human-computer interaction and performance assessment.

Objective: The effective treatment of central nervous system (CNS) disorders remains a significant challenge, primarily due to its molecular and structural complexity. Clinical translation of promising therapeutic agents is limited by the lack of optimal drug delivery systems capable of targeted, localized release of drugs to the brain and spinal cord.Approach: This review provides an overview of the potential of affinity-based drug delivery systems, which leverage molecular interactions to enhance the delivery and efficacy of therapeutic agents within the CNS.Main results: Various approaches, including hydrogels, micro- and nanoparticles, and functionalized biomaterials, are examined for their ability to provide local, sustained release of proteins, growth factors and other drugs.Significance: Furthermore, we present a detailed analysis of design considerations for developing effective affinity-based systems, incorporating insights from both existing literature and our group's research. These considerations include the biochemical modification of delivery vehicles and the optimization of physical and chemical properties to improve therapeutic outcomes.

Objective. Macrophages and astrocytes play a crucial role in the aftermath of a traumatic spinal cord injury (SCI). Infiltrating macrophages adopt a pro-inflammatory phenotype while resident astrocytes adopt a neurotoxic phenotype at the injury site, both of which contribute to neuronal death and inhibit axonal regeneration. The cytokine interleukin-4 (IL-4) has shown significant promise in preclinical models of SCI by alleviating the macrophage-mediated inflammation and promoting functional recovery. However, its effect on neurotoxic reactive astrocytes remains to be elucidated, which we explored in this study. We also studied the beneficial effects of a sustained release of IL-4 from an injectable biomaterial compared to bolus administration of IL-4.Approach. We fabricated a heparin-based coacervate capable of anchoring and releasing bioactive IL-4 and tested its efficacyin vitroandin vivo. Main results. We show that IL-4 coacervate is biocompatible and drives a robust anti-inflammatory macrophage phenotype in culture. We also show that IL-4 and IL-4 coacervate can alleviate the reactive neurotoxic phenotype of astrocytes in culture. Finally, using a murine model of contusion SCI, we show that IL-4 and IL-4 coacervate, injected intraspinally 2 d post-injury, can reduce macrophage-mediated inflammation, and alleviate neurotoxic astrocyte phenotype, acutely and chronically, while also promoting neuroprotection with significant improvements in hindlimb locomotor recovery. We observed that IL-4 coacervate can promote a more robust regenerative macrophage phenotypein vitro, as well as match its efficacyin vivo,compared to bolus IL-4.Significance. Our work shows the promise of coacervate as a great choice for local and prolonged delivery of cytokines like IL-4. We support this by showing that the coacervate can release bioactive IL-4, which acts on macrophages and astrocytes to promote a pro-regenerative environment following a SCI leading to robust neuroprotective and functional outcomes.

Objective.Implanted neural microelectrodes are an important tool for recording from and stimulating the cerebral cortex. The performance of chronically implanted devices, however, is often hindered by the development of a reactive tissue response. Previous computational models have investigated brain strain from micromotions of neural electrodes after they have been inserted, to investigate design parameters that might minimize triggers to the reactive tissue response. However, these models ignore tissue damage created during device insertion, an important contributing factor to the severity of inflammation. The objective of this study was to evaluate the effect of electrode geometry, insertion speed, and surface friction on brain tissue strain during insertion.Approach. Using a coupled Eulerian-Lagrangian approach, we developed a 3D finite element model (FEM) that simulates the dynamic insertion of a neural microelectrode in brain tissue. Geometry was varied to investigate tip bluntness, cross-sectional shape, and shank thickness. Insertion velocities were varied from 1 to 8 m s^-1. Friction was varied from frictionless to 0.4. Tissue strain and potential microvasculature hemorrhage radius were evaluated for brain regions along the electrode shank and near its tip.Main results. Sharper tips resulted in higher mean max principal strains near the tip except for the bluntest tip on the square cross-section electrode, which exhibited high compressive strain values due to stress concentrations at the corners. The potential vascular damage radius around the electrode was primarily a function of the shank diameter, with smaller shank diameters resulting in smaller distributions of radial strain around the electrode. However, the square shank interaction with the tip taper length caused unique strain distributions that increased the damage radius in some cases. Faster insertion velocities created more strain near the tip but less strain along the shank. Increased friction between the brain and electrode created more strain near the electrode tip and along the shank, but frictionless interactions resulted in increased tearing of brain tissue near the tip.Significance. These results demonstrate the first dynamic FEM study of neural electrode insertion, identifying design factors that can reduce tissue strain and potentially mitigate initial reactive tissue responses due to traumatic microelectrode array insertion.

Objective.Optogenetics allows the manipulation of neural circuitsin vivowith high spatial and temporal precision. However, combining this precision with control over a significant portion of the brain is technologically challenging (especially in larger animal models).Approach.Here, we have developed, optimised, and testedin vivo, the Utah Optrode Array (UOA), an electrically addressable array of optical needles and interstitial sites illuminated by 181μLEDs and used to optogenetically stimulate the brain. The device is specifically designed for non-human primate studies.Main results.Thinning the combinedμLED and needle backplane of the device from 300μm to 230μm improved the efficiency of light delivery to tissue by 80%, allowing lowerμLED drive currents, which improved power management and thermal performance. The spatial selectivity of each site was also improved by integrating an optical interposer to reduce stray light emission. These improvements were achieved using an innovative fabrication method to create an anodically bonded glass/silicon substrate with through-silicon vias etched, forming an optical interposer. Optical modelling was used to demonstrate that the tip structure of the device had a major influence on the illumination pattern. The thermal performance was evaluated through a combination of modelling and experiment, in order to ensure that cortical tissue temperatures did not rise by more than 1 °C. The device was testedin vivoin the visual cortex of macaque expressing ChR2-tdTomato in cortical neurons.Significance.It was shown that the UOA produced the strongest optogenetic response in the region surrounding the needle tips, and that the extent of the optogenetic response matched the predicted illumination profile based on optical modelling-demonstrating the improved spatial selectivity resulting from the optical interposer approach. Furthermore, different needle illumination sites generated different patterns of low-frequency potential activity.

Objective.Producing realistic numerical models of neurostimulation electrodes in contact with the electrolyte and tissue, for use in time-domain finite element method simulations while maintaining a reasonable computational burden remains a challenge. We aim to provide a straightforward experimental-theoretical hybrid approach for common electrode materials (Ti, TiN, ITO, Au, Pt, IrOx) that are relevant to the research field of bioelectronics, along with all the information necessary to replicate our approach in arbitrary geometry for real-life experimental applications.Approach.We used electrochemical impedance spectroscopy (EIS) to extract the electrode parameters in the AC regime under different DC biases. The pulsed electrode response was obtained by fast amperometry (FA) to optimize and verify the previously obtained electrode parameters in a COMSOL Multiphysics model. For optimization of the electrode parameters a constant phase element (CPE) needed to be implemented in time-domain.Main results.We find that the parameters obtained by EIS can be used to accurately simulate pulsed response only close to the electrode open circuit potential, while at other potentials we give corrections to the obtained parameters, based on FA measurements. We also find that for many electrodes (Au, TiN, Pt, and IrOx), it is important to implement a distributed CPE rather than an ideal capacitor for estimating the electrode double-layer capacitance. We outline and provide examples for the novel time-domain implementation of the CPE for finite element method simulations in COMSOL Multiphysics.Significance.An overview of electrode parameters for some common electrode materials can be a valuable and useful tool in numerical bioelectronics models. A provided FEM implementation model can be readily adapted to arbitrary electrode geometries and used for various applications. Finally, the presented methodology for parametrization of electrode materials can be used for any materials of interest which were not covered by this work.

Objective.Due to the difficulty in acquiring motor imagery electroencephalography (MI-EEG) data and ensuring its quality, insufficient training data often leads to overfitting and inadequate generalization capabilities of deep learning-based classification networks. Therefore, we propose a novel data augmentation method and deep learning classification model to enhance the decoding performance of MI-EEG further.Approach.The raw EEG signals were transformed into the time-frequency maps as the input to the model by continuous wavelet transform. An improved Wasserstein generative adversarial network with gradient penalty data augmentation method was proposed, effectively expanding the dataset used for model training. Additionally, a concise and efficient deep learning model was designed to improve decoding performance further.Main results.It has been demonstrated through validation by multiple data evaluation methods that the proposed generative network can generate more realistic data. Experimental results on the BCI Competition IV 2a and 2b datasets and the actual collected dataset show that classification accuracies are 83.4%, 89.1% and 73.3%, and Kappa values are 0.779, 0.782 and 0.644, respectively. The results indicate that the proposed model outperforms state-of-the-art methods.Significance.Experimental results demonstrate that this method effectively enhances MI-EEG data, mitigates overfitting in classification networks, improves MI classification accuracy, and holds positive implications for MI tasks.