Epilepsy Detection and Monitoring - Quantitative EEG Analysis Methods and Clinical Applications - page 153

()
()

δ
δ

xt
x

Δ

N

1

= ∑

ij

L = log (6.14)

2

Nt Δ 0

i 1

ij

A decrease in the Lyapunov exponent indicates this transition to a more ordered
state (Figure 6.4). The assumptions underlying this methodology have been experi-
mentally observed in the STL max time-series data from human patients [18, 26] and
rodents [35]. For instance, in an experimental rat model of temporal lobe epilepsy,
changes in the phase portrait of STL max can be readily identified for the preictal,
ictal, and postictal states, during a spontaneous limbic seizure (Figure 6.5). This
characterization by the Lyapunov exponent has, however, been successful only for

10 25

20

8

6

15

4 10

2

5

0

0 0 51015202530 35 50

Time (minutes)

100

Time (minutes)

Figure 6.4 Sample STL max profile for a 35-minute epoch including a grade 5 seizure from an epileptic
rat. Seizure onset and offset are indicated by dashed vertical lines. Note the drop in the STL max value
during the seizure period. (b) T-index profiles calculated from STL max values of a pair of electrodes
from rat A. The electrode pair includes a right hippocampus electrode and a left frontal electrode.
Vertical dotted lines represent seizure onset and offset. The horizontal dashed line represents the criti-
cal entrainment threshold. Note a decline in the T-index value several minutes before seizure
occurrence.

8

7 Preictal (1 hour)

Ictal (1.5 min)

6 Postictal (1 hour)

5

4

3

8

2

7

6

1

1

5

2

4

3

4

3

5

6

2

STLmax t+ τ

STLmax t

7

1

8

Figure 6.5 Phase portrait of STL max of a spontaneous rodent epileptic seizure (grade 5).