Biomedical Engineering Reference
In-Depth Information
predict a seizure by detecting the state change from interictal to preictal. Although
unable to predict seizures, statistical moments may prove valuable as seizure detec-
tors in recordings with large amplitude seizures.
6.4.4 Recurrence Time Statistics
The recurrence time statistic (RTS), T1, is a characteristic of trajectories in an
abstract dynamical system. Stated informally, it is a measure of how often a given
trajectory of the dynamical system visits a certain neighborhood in the phase space.
T1 has been calculated for ECoG data in an effort to detect seizures, with significant
success. With two different patients and a total of 79 hours of data, researchers were
able to detect 97% of the seizures with only an average of 0.29 false negatives per
hour [21]. They did not, however, indicate any attempts to predict seizures. Results
from our preliminary studies on human EEG signals showed that the RTS exhibited
significant change during the ictal period that is distinguishable from the back-
ground interictal period (Figure 6.3). In addition, through the observations over
multichannel RTS features, the spatial pattern from channel to channel can also be
traced. Existence of these spatiotemporal patterns of RTS suggests that it is possible
to utilize RTS to develop an automated seizure-warning algorithm.
150
RTS
Seizure
Intracranial EEG
(patient)
100
50
0
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
500
Scalp EEG
(patient)
400
300
200
100
0
0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
15
Rat EEG
10
5
0
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
Hours
Figure 6.3
Studies on human EEG signals show that the recurrence time statistics exhibit changes during the
ictal period that is distinguishable from the background interictal period.
Search WWH ::
Custom Search