Biomedical Engineering Reference
In-Depth Information
The problem is to analyze the data and determine
whether there is interhemispheric differential activity
that shows up in any or all of these cognitive tasks. The
potential sites of differential activity for each task:
1. For the baseline/resting task there should be no
differential activity.
2. For the verbal tasks, it is possible that there is
differential activity in the left hemisphere in the
occipital area (O1 or O2).
3. For the mathematical tasks, it is possible that there is
differential activity in the left hemisphere, in the
parietal area (P3 or P4).
How does one model brain activation? For the anal-
ysis, one needs to find a signal representation that is as
small as possible, but it should contain the information
necessary to differentiate different mental states. The
MATLAB program in Example 2.2.3 was written to
analyze this large set of data and determine whether
there are differences in activation, using a mathematical
(and numerical) model of brain activation as frequency
content.
If the frequency content of a particular signal is pri-
marily in the alpha band (7 to 13 Hz), then the subject
must be resting and not performing any tasks. If the
subject is performing a task then the majority of the
power shifts to the beta frequency band. Of course,
artifact due to 60 Hz noise has to be removed and muscle
artifact due to eye blinking has to be removed. The 60 Hz
noise is removed from high beta and the muscle artifact
will show up in the delta and theta frequency bands.
Table 2.2-2 Demographic information on subjects tested in
Keirn (1988) .
Subject
Age
Handedness
Gender
Sessions
1
48
Left
Male
2
2
39
Right
Male
1
3
<30
Right
Male
2
4
<30
Right
Male
2
5
<30
Right
Female
3
6
<30
Right
Male
2
7
<30
Right
Male
1
integers, and character strings). The data set includes
a set of trials of EEG measurement and some annotation.
Each trial has four elements in the cell:
Subject number
Type of the cognitive task
Trial number
The digitized EEG data
Each digitized EEG sample is 10 seconds long and was
sampled at 250 Hz. Data was recorded from six channels:
C3, C4, O1, O2, P3 and P4. The EOG was measured at
A1 and A2. Each trial has
250 samples/second 10 seconds
6 channels ¼ 15,000 data
The sampled data has been bandpass filtered for 0.1 to
100 Hz. The frequency components of EEG data are:
delta waves, from 0 to 2 Hz
theta waves , from 2 to 7 Hz
alpha waves, from 7 to 13 Hz
low beta waves, from 14 to 20 Hz
high beta waves, above 20 Hz, typically up to 64 Hz
Example 2.2.3 Differential brain activity in the left and
right hemispheres.
Given the dataset described above, analyze the frequency
content of the EEG signals to determine if there are
differences in brain activity, between hemispheres, that
can be attributed to cognitive task, frequency band, or
any of the demographic characteristics.
% constants
TRIALS ¼ 5;
EXPERIMENTS ¼ 5;
SESSIONS ¼ 13;
SR ¼ 250; %sample rate
% subject/trial map
SUBJ{1} ¼ 1:50;
SUBJ{2} ¼ 51:75;
SUBJ{3} ¼ 76:125;
SUBJ{4} ¼ 126:175;
SUBJ{5} ¼ 176:250;
SUBJ{6} ¼ 251:300;
SUBJ{7} ¼ 301:325;
load eegdata.mat
Search WWH ::




Custom Search