System derived spatial-temporal CNN for high-density fNIRS BCI DOI Creative Commons
Robin Dale, Thomas D. O’Sullivan, Scott S. Howard

et al.

IEEE Open Journal of Engineering in Medicine and Biology, Journal Year: 2023, Volume and Issue: 4, P. 85 - 95

Published: Jan. 1, 2023

An intuitive and generalisable approach to spatial-temporal feature extraction for high-density (HD) functional Near-Infrared Spectroscopy (fNIRS) brain-computer interface (BCI) is proposed, demonstrated here using Frequency-Domain (FD) fNIRS motor-task classification. Enabled by the HD probe design, layered topographical maps of Oxy/deOxy Haemoglobin changes are used train a 3D convolutional neural network (CNN), enabling simultaneous spatial temporal features. The proposed CNN shown effectively exploit relationships in measurements improve classification haemodynamic response, achieving an average F1 score 0.69 across seven subjects mixed training scheme, improving subject-independent as compared standard CNN.

Language: Английский

What happens in the prefrontal cortex? Cognitive processing of novel and familiar stimuli in soccer: An exploratory fNIRS study DOI Creative Commons

Lena F. Schmaderer,

Mathilda Meyer,

Rüdiger Reer

et al.

European Journal of Sport Science, Journal Year: 2023, Volume and Issue: 23(12), P. 2389 - 2399

Published: Aug. 3, 2023

The importance of both general and sport-specific perceptual-cognitive abilities in soccer players has been investigated several studies. Although these skills could contribute significantly to players' expertise, the underlying cortical mechanisms have not clarified yet. Examining activity changes prefrontal cortex under different cognitive demands may help better understand sports expertise. aim this study was analyse experts during tasks. For purpose, 39 semi-professional performed four tests, two which assessed cognition, other cognition. Since is a movement-intensive sport, tests were motion. While performing recorded using functional near-infrared spectroscopy (fNIRS) (NIRSport, NIRx Medical Technologies, USA). Differences tasks analysed paired t-tests. results showed significant increases (novel stimuli) compared (familiar stimuli). comparatively lower change cognition might be due learned automatisms field. These seem line with previous findings on novel automated "repetition suppression theory" "neural efficiency theory". Furthermore, processes caused by altered structures represent decisive factor for expertise team sports. However, further research needed clarify involvement cognition.This fNIRS examines differences tasks.In tasks, increased detected, whereas found.These support “repetition theory” earlier processing stimuli (PFC).The special soccer.

Language: Английский

Citations

4
A Unified Local–Global Feature Extraction Network for Human Gait Recognition Using Smartphone Sensors DOI Creative Commons
Sonia Das, Sukadev Meher, Upendra Kumar Sahoo

et al.

Sensors, Journal Year: 2022, Volume and Issue: 22(11), P. 3968 - 3968

Published: May 24, 2022

Smartphone-based gait recognition has been considered a unique and promising technique for biometric-based identification. It is integrated with multiple sensors to collect inertial data while person walks. However, captured may be affected by several covariate factors due variations of sequences such as holding loads, wearing types, shoe etc. Recent approaches either work on global or local features, causing failure handle these covariate-based features. To address issues, novel weighted multi-scale CNN (WMsCNN) architecture designed extract features boosting accuracy. Specifically, weight update sub-network (Ws) proposed increase reduce the weights concerning their contribution final classification task. Thus, sensitivity toward decreases using updated technique. Later, are fed fusion module used produce overall classification. Extensive experiments have conducted four different benchmark datasets, demonstrated results model superior other state-of-the-art deep learning approaches.

Language: Английский

Citations

7
Brain-Controlled Lower-Limb Exoskeleton to Assist Elderly and Disabled DOI
Hammad Nazeer, Noman Naseer

2022 8th International Conference on Control, Decision and Information Technologies (CoDIT), Journal Year: 2022, Volume and Issue: unknown

Published: May 17, 2022

Disability limits an individual's ability to participate in everyday activities. Rehabilitation is specialized healthcare improve, maintain or restore physical strength, cognition, and mobility. Exoskeletons are the external devices used aid disabled performing daily life activities restoring their strength capability. Brain-computer interface (BCI) a technique use brain signals controlling directly. In this paper, BCI-based control of lower limb exoskeleton proposed using functional near-infrared spectroscopy (fNIRS). Brain for waking nine healthy subjects on treadmill recorded pre-processed, followed by channel selection feature extraction. Linear discriminant analysis classify walking rest achieved significantly (p < 0.05) higher accuracy 75.5 ± 13.0%. Furthermore, system showed better performance as compared all channels classification. The methodology step forward achieve intuitive exoskeletons.

Language: Английский

Citations

7
Deep Learning in Design of Semi-Automated 3D Printed Chainmail with Pre-Programmed Directional Functions for Hand Exoskeleton DOI Creative Commons
Izabela Rojek, Jakub Kopowski, Piotr Kotlarz

et al.

Applied Sciences, Journal Year: 2022, Volume and Issue: 12(16), P. 8106 - 8106

Published: Aug. 12, 2022

The aim of this paper is to refine a scientific solution the problem automated or semi-automated efficient and practical design 3D printed chainmails exoskeletons with pre-programmed properties (variable stiffness/flexibility depending on direction) reflecting individual user needs, including different types degrees deficit. We demonstrate example using chainmail in hand exoskeleton, where components can be arranged single-layer structure adjustable one- two-way bending modulus. novelty proposed approach consists combining use real data from research exoskeleton hand, new methods their analysis deep neural networks, clear scalable fabric product that personalized (mechanical parameters such as stiffness bend angles various directions) needs goals therapy particular patient. So far, unique, having no equivalent literature. This paves way for wider implementation adaptive based machine learning, more complex designs.

Language: Английский

Citations

6
System derived spatial-temporal CNN for high-density fNIRS BCI DOI Creative Commons
Robin Dale, Thomas D. O’Sullivan, Scott S. Howard

et al.

IEEE Open Journal of Engineering in Medicine and Biology, Journal Year: 2023, Volume and Issue: 4, P. 85 - 95

Published: Jan. 1, 2023

An intuitive and generalisable approach to spatial-temporal feature extraction for high-density (HD) functional Near-Infrared Spectroscopy (fNIRS) brain-computer interface (BCI) is proposed, demonstrated here using Frequency-Domain (FD) fNIRS motor-task classification. Enabled by the HD probe design, layered topographical maps of Oxy/deOxy Haemoglobin changes are used train a 3D convolutional neural network (CNN), enabling simultaneous spatial temporal features. The proposed CNN shown effectively exploit relationships in measurements improve classification haemodynamic response, achieving an average F1 score 0.69 across seven subjects mixed training scheme, improving subject-independent as compared standard CNN.

Language: Английский

Citations

3