Biomedical Signal Processing and Control, Journal Year: 2025, Volume and Issue: 104, P. 107503 - 107503
Published: Jan. 18, 2025
Language: Английский
Brain Sciences, Journal Year: 2024, Volume and Issue: 14(8), P. 820 - 820
Published: Aug. 16, 2024
Electroencephalogram (EEG) and functional near-infrared spectroscopy (fNIRS) can objectively reflect a person’s emotional state have been widely studied in emotion recognition. However, the effective feature fusion discriminative learning from EEG–fNIRS data is challenging. In order to improve accuracy of recognition, graph convolution capsule attention network model (GCN-CA-CapsNet) proposed. Firstly, signals are collected 50 subjects induced by video clips. And then, features EEG fNIRS extracted; fused generate higher-quality primary capsules with Pearson correlation adjacency matrix. Finally, module introduced assign different weights capsules, selected better classification dynamic routing mechanism. We validate efficacy proposed method on our dataset an ablation study. Extensive experiments demonstrate that GCN-CA-CapsNet achieves more satisfactory performance against state-of-the-art methods, average increase 3–11%.
Language: Английский
Citations
4
Neurophotonics, Journal Year: 2022, Volume and Issue: 10(01)
Published: Oct. 22, 2022
SignificanceAdvances in electronics have allowed the recent development of compact, high channel count time domain functional near-infrared spectroscopy (TD-fNIRS) systems. Temporal moment analysis has been proposed for increased brain sensitivity due to depth selectivity higher order temporal moments. We propose a general linear model (GLM) incorporating TD data and auxiliary physiological measurements, such as short separation channels, improve recovery HRF.AimsWe compare performance previously reported multi-distance techniques commonly used continuous wave (CW) fNIRS hemodynamic response function (HRF) recovery, namely block averaging CW GLM. Additionally, we technique GLM.ApproachWe augmented resting TD-fNIRS (six subjects) with known synthetic HRFs. then employed GLM "short-separation regression" designed both recover calculated root mean square error (RMSE) correlation recovered HRF ground truth. compared equivalent paired t-tests.ResultsWe found that, on average, improves correlations by 98% 48% HbO HbR respectively, over The improvement is 12% (HbO) 27% (HbR). RMSE decreases 56% 52% (HbO HbR) no statistically significant moment.ConclusionsProperly covariance-scaled outperform their equivalents Furthermore, our based moments outperforms regular analysis, while allowing incorporation measurements confounding signals from scalp.
Language: Английский
Citations
19
Computational Intelligence and Neuroscience, Journal Year: 2023, Volume and Issue: 2023(1)
Published: Jan. 1, 2023
This umbrella review is motivated to understand the shift in research themes on brain-computer interfacing (BCI) and it determined that a away from focus medical advancement system development applications included education, marketing, gaming, safety, security has occurred. The background of this examined aspects BCI categorisation, neuroimaging methods, brain control signal classification, applications, ethics. specific area software hardware was not examined. A search using One Search undertaken 92 reviews were selected for inclusion. Publication demographics indicate average number authors papers considered 4.2 ± 1.8. results also rapid increase 2003, with only three before period, two 1972, one 1996. While predominantly Euro-American early reviews, shifted more global authorship, which China dominated by 2020-2022. revealed six disciplines associated systems: life sciences biomedicine (
Language: Английский
Citations
10
Biomedical Signal Processing and Control, Journal Year: 2023, Volume and Issue: 86, P. 105160 - 105160
Published: June 20, 2023
Language: Английский
Citations
10
Biomedical Signal Processing and Control, Journal Year: 2025, Volume and Issue: 104, P. 107503 - 107503
Published: Jan. 18, 2025
Language: Английский
Citations
0