Professional Documents
Culture Documents
116
International Journal of Digital Content Technology and its Applications
Volume 3, Number 4, December 2009
As Figure 2 show, these subjects were asked to movement according to an experimental paradigm
execute an imagined left(L) or right(R) index finger
Figure 3. Averaged time-frequency maps of right and left hand movement imagination of K3b
show, The task was to perform imagery left hand, right movement imagination, the energy concentrated on the
hand according to a cue.. The experiment consists of 10 to 15 Hz frequency band on electrode C3, and it
several runs (>= 6) with 40 trials; after trial begin, the maintained to the end of imagining. In the same band
first 2s were quite, at t=2s an acoustic stimulus on electrode C4, the energy only concentrated in the
indicated the beginning of the trial, and a cross “+” is early stage of imagination (about 0 ~ 0.6 s). On the
displayed; then from t=3s an arrow to the left, right, as contrary during the right-hand movement imagination,
displayed for 1 s; at the same time the subject was the energy concentrated on the 10 to 15 Hz frequency
asked to imagine a left hand, right hand, respectively, band on electrode C3 only in the early stage of
until the cross disappeared. imagination (about 0 ~ 0.6 s), but it maintained to the
end of imagining on electrode C4.
Step 2: Calculated the Fisher distance[23] between
two classes. Fisher distance F is calculated as
( µ1 − µ 2 ) 2
F= (2)
Figure 2. Timing of the paradigm σ 12 + σ 22
In equation (2), μ and σ are the mean and the
3. Method standard deviation of the feature they correspond to.
Step 3: Sort F (vector of Fisher distance
In this paper the EEG data from the C3 ,Cz and )ascending and select the first k<n(n is the number of
C4 electrodes only are analyzed. In order to get the trials) as feature.
local activity ,the EEG data are converted to Hjort In this paper , we define category line to classify,
category line y is defined as
derivation. The Hjort derivation
C iH is calculated as k k
1 ∑ Pi − ∑ Qi
C iH = ci − ∑ sc j
8 j∈Si
(1)
y=
j =1 j =1
(3)
2
where ci is the reading of the center electrode scj, with In equation (3), Pi andQi are the set of feature(right
i=3 ,0 or 4 and j is the set of indices corresponding to hand movement imagination and left hand movement
the eight electrodes surrounding electrode ci. imagination).k is the number of feature
The algorithm we use in this paper as followed:
Step 1: Convert the time domain signal to
4. Result and discuss
time-frequency domain signal using STFTs. Here we
use matlab function “spectrogram”. The EEG data are
In this paper, the EEG data was separated into two
bandpass filtered and only signals between 1 Hz and 45
datasets, one dataset was used to learning, another
Hz was used. We use the 73 trials available of the
dataset was used to testing. in order to make the
subject k3b’s task to test.
selected features as far as possible including the
Figure 3 shows, the energy distribution on electrode
C3, Cz and C4 is different. During the left-hand essential features of motor imagery, we analyzed
117
Classification of Motor Imagery EEG Signals Based on Time–Frequency Analysis
Zhendong Mu, Dan Xiao, Jianfeng Hu
feature’s Fisher distance selected during the following three different frequency bands: θ rhythm(4~7Hz),
118
International Journal of Digital Content Technology and its Applications
Volume 3, Number 4, December 2009
119