射频识别在SSVEP-BCI中的MULTISET典型相关分析应用

29 浏览量更新于2024-07-14 1 收藏 473KB PDF 举报

本文是一篇研究论文，标题为《射频识别在SSVEP-BCI中的应用：基于MULTISET典型相关分析》（Frequency Recognition in SSVEP-Based BCI Using MULTISET Canonical Correlation Analysis）。该研究聚焦于一种名为脑机接口（Brain-Computer Interface, BCI）的技术，特别是眼神控制（Synchronised Pupillary Velocity, SSVEP）类型的BCI，这是一种利用人眼对不同频率视觉刺激产生的瞬态生理反应来实现脑部与外部设备交互的方法。多集典型相关分析（MULTISET Canonical Correlation Analysis, MultiCCA）作为一种统计方法，被作者们应用于SSVEP-BCI系统的设计中，以提高信号处理的精度和有效性。MultiCCA能够同时分析多个集合（或多传感器数据）之间的关联，这对于处理多源脑电图（Electroencephalogram, EEG）信号特别有价值，因为SSVEP-BCI依赖于多导联记录到的视觉事件引发的瞬态眼部信号变化。研究者们来自多个机构，包括中国东华理工大学的先进化学过程控制与优化实验室，日本RIKEN脑科学研究所的高级脑信号处理实验室，以及波兰科学院的系统研究 institute。这些作者的研究团队可能共同开发了一种新的算法或改进了现有的技术，以增强SSVEP-BCI系统的性能，使其在实际应用中如残疾人士的康复、游戏控制或虚拟现实交互等领域展现更高效和精确的控制能力。论文发表于2013年，但在2014年进行了修订，并可能在国际神经系统期刊（International Journal of Neural Systems）上发表。论文内容涉及理论分析、实验设计、数据分析方法和可能的结果评估。由于具体内容没有提供，我们无法详述细节，但可以推测，论文的核心贡献可能是如何通过MultiCCA优化了SSVEP信号的特征提取，从而提高了SSVEP-BCI的信噪比和识别精度。这篇论文在射频识别领域具有较高的学术价值，尤其是在非侵入式脑机接口技术的发展中，展示了多集典型相关分析这一工具如何提升人机交互的效率和可靠性。

Multiset CCA for SSVEP Recognition

mensions of EEG tensor data for SSVEP frequency

recognition. The MwayCCA method has shown im-

proved recognition performance of SSVEP frequency

compared to the CCA method. On the other hand,

a pha se constrained CC A method (PCCA)

has

also been proposed for SSVEP frequency recogni-

tion. The PCCA method achieved signiﬁcant ac-

curacy improvement in comparison with the CCA

method, by embedding the phas e information esti-

mated from training data into the reference signals.

Howe ver, the procedures of reference signal optimiza-

tion in both the MwayCCA and the PCCA meth-

ods are not completely based on training data but

still need to resort to the pre-constructed sine-cosine

waves.

In the present study, we introduce a multi-

set canonical correlation analy sis (MsetCCA)-based

method to reference signal optimizatio n for SSVEP

frequency recognition. MsetCCA was developed as

an extension of CCA to ﬁnd multiple linear trans-

forms that maximize the overall correlation among

canonical variates from multiple sets of random vari-

ables

. Recently, MsetCCA has been successfully

applied to the joint blind source separation of multi-

subject fMRI data

30,31

, the functional connectiv-

ity analys is of fMRI data

, and also the fusion of

concurrent single trial ERP and fMRI data

. In

this study, the proposed MsetCCA method imple-

ments reference signal optimization for SSVEP fre-

quency recognition through extracting SSVEP com-

mon features from the joint spa tial ﬁltering of multi-

ple sets of EEG data recorded at the same stimulus

frequency. Diﬀerent from the MwayCCA and the

PCCA methods, the procedure of reference signa l

optimization in the MsetCCA method is completely

based on training data. EEG data recorded from ten

healthy subjects are used to validate the Ms etCCA

method in comparison with the CCA, the Mway-

CCA and the PCCA methods. Experimental results

indicate that the proposed MsetCCA method out-

performs the three c omp eting methods for SSVEP

recognition, especially for a small number of chan-

nels and a short time window length.

2. Materials and Methods

2.1. Experiments and EEG recordings

The experiments were performed by ten healthy sub-

jects (S1-S10, aged from 21 to 27, all males) who re-

ceived their remuneration. All of them had normal

or corrected to normal vision. In the experiments,

the subjects were seated in a comfor table chair 60

cm from a standard 17 inch CRT monitor (85 Hz re-

fresh rate, 1024 × 768 screen resolution) in a shielded

room. Four red squares were presented on the screen

as stimuli (see Fig. 1 (a)) and ﬂickered at diﬀerent

four frequencies 6 Hz, 8 Hz, 9 Hz and 10 Hz, respec-

tively. There were 20 experimental runs completed

by each subject. In each run, the subject was asked

to focus attention on each of the four stimuli once for

4 s, respectively, preceded by each target cue dur a-

tion of 2 s. A total of 80 trials (4 trials in each run)

were therefore performed by each subject.

EEG signals were recorded by using the Nuamps

ampliﬁer (NuAmp, Neurosca n, Inc .) at 250 Hz sam-

pling rate with high-pass and low-pass ﬁlters of 0.1

and 70 Hz from 30 channels arranged ac cording to

standard positions of the 10-20 international system

(see Fig. 1 (b)). The average of two mas toid elec-

trodes (A1, A2) was used as reference and the elec-

trode on the forehead (GND) as ground. With a

sixth-order Butterworth ﬁlter, a band-pass ﬁltering

from 4 to 45 Hz was implemented on the recorde d

EEG signals before further analysis.

2.2. CCA for SSVEP Recognition

As a multivariate statistical method, canonical

correlation analysis (CCA)

explores the underly-

ing cor relation between two se ts of data. Given two

sets of random variables X ∈ R

×J

and Y ∈ R

×J

which are norma lized to have zero mean and unit

variance, CC A is to seek a pair of linear transforms

∈ R

and w

∈ R

such that the corr e lation be-

tween linear combinations ˜x = w

X and ˜y = w

is maximized as

max

ρ =



˜x˜y



E [˜x˜x

] E [˜y˜y

]

. (1)

The maximum of correlation coeﬃcient ρ with re-

spect to w

and w

is the maximum canonica l cor-

relation.

A CCA-based frequency recognition method

was ﬁrst introduced by Lin et al. to SSVEP-based

剩余14页未读，继续阅读

weixin_38545923

粉丝: 4
资源: 933

射频识别在SSVEP-BCI中的MULTISET典型相关分析应用

射频识别技术

射频识别原理

SSVEP-BCI-研究：高中SSVEP-BCI研究项目，旨在提高捕获的EEG信号的分类准确性

基于时频联合表示ssvep-bci的刺激传递

SSVEP-BCI algorithm introduction v5.pdf

A novel stimulation method for multi-class SSVEP-BCI using intermodulation frequencies

基于SSVEP-BCI的脑机音乐接口系统，可以让用户使用眼睛注视选择音符，再由AI拓展生成音乐

时频联合分析在SSVEP-BCI刺激传递中的应用研究

任务判别成分分析法（TDCA）在SSVEP-BCI中的应用研究

MATLAB实现SSVEP-BCI直接ITR优化及代码示例

最新资源