Abstracts

Temporal lobe seizures are preceded by preictal transitions detectable by the quantitative analysis of intracranial EEG recordings ([italic]Epilepsia 31(5): 621, 1990; Epilepsia 35S: 133, 1994; The Neuroscientist, 2: 118, 1996; J. Combinatorial Optimization 5: 9, 2001; Nature Medicine 4: 1173, 1998; Europ. J. Neurosci. 10: 786, 1998; Lancet 357: 183, 2001; Neuron 30: 51, 2001[/italic]). These transitions may be observed seconds, minutes, or even hours before the clinical seizure onset. These findings suggested that it is possible to develop an implantable seizure prediction device for diagnostic and therapeutic purposes. In the present study, based on the convergence pattern of a nonlinear dynamical measure in the preictal period, we sought to test the hypothesis that the preictal dynamical transitions are also detectable by the analysis of scalp EEG recordings.
Long-term scalp EEG data from 3 temporal-lobe epileptic patients, including 11, 11 and 14 medically intractable partial seizures per patient, were analyzed to test our hypothesis. The short-term maximum Lyapunov exponents ([italic]STLmax[/italic]) were calculated for each EEG channel and each sequential 10.24-second non-overlapping data segment, according to methods previously described [italic](Brain Topogr. 2: 187, 1990[/italic]). The paired-T statistic was estimated continuously over each 10-minute sliding overlapping window to test the mean difference of [italic]STLmax[/italic] values between two electrode sites. Electrode pairs were considered not entrained during any 10-minute period if the mean [italic]STLmax[/italic] values were significantly different (p[lt]0.05). The T-index curve was generated for the 3-hour time interval before each seizure. A preictal dynamical transition was considered detectable if the corresponding T-index curve had only one decreasing trend (i.e., non-entrained electrode pairs gradually becoming entrained) in the 3-hour interval before a seizure. Otherwise, the preictal transition was considered undetectable.
The preictal dynamical transitions characterized by the convergence of [italic]STLmax[/italic] values were ovserved in 9 (/11) seizures in patient 1; 11 (/11) seizures in patient 2, and 14 (/14) seizures in patient 3 (overall 94.4%). The transitions were not observed in only 2 (out of 36) seizures. Further, the preictal transitions began on the average of 74.7 minutes (27.7, 90.2, and 92.8 minutes in patient 1, 2, and 3 respectively) prior to the upcoming seizures.
A preictal transition similar to that previously observed in the intracranial EEG recordings was also detectable in the scalp EEG. This suggests that it may be possible to develop a seizure warning method by the analysis of scalp EEG recordings.
[Supported by: NIH/NIBIB 8R01EB002089-03
U.S.Veterans Affairs
The Whitaker Foundation
NSF]