Center of Neurosciences, Swammerdam Institute for Life Sciences, University of Amsterdam, 1098 SM Amsterdam, The Netherlands.
IEEE Rev Biomed Eng. 2008;1:143-56. doi: 10.1109/RBME.2008.2008246.
This overview covers recent advances in the field of EEG/MEG signal processing and modeling in epilepsy regarding both interictal and ictal phenomena. In the first part, the main methods used in the analysis of interictal EEG/MEG epileptiform spikes are presented and discussed. Source and volume conductor models are passed in review, namely the equivalent dipole source concept, the requirements for adequate time and spatial sampling, the question of how to validate source solutions, particularly by comparing solutions obtained using scalp and intracranial EEG signals, EEG & MEG data, or EEG simultaneously recorded with fMRI (BOLD signals). In the second part, methods used for the characterization of seizures are considered, namely dipolar modeling of spikes at seizure onset, decomposition of seizure EEG signals into sets of orthogonal spatio-temporal components, and also methods (linear and nonlinear) of estimating seizure propagation. In the third part, the crucial issue of how the transition between interictal and seizure activity takes place is examined. In particular the vicissitudes of the efforts along the road to seizure prediction are shortly reviewed. It is argued that this question can be reduced to the problem of estimating the excitability state of neuronal populations in the course of time as a seizure approaches. The value of active probing methods in contrast with passive analytical methods is emphasized. In the fourth part modeling aspects are considered in the light of two special kinds of epilepsies, absences characterized by spike-and-wave discharges and mesial temporal lobe epilepsy. These two types correspond to different scenarios regarding the transition to epileptic seizures, namely the former is a case of a jump transition and the latter is a typical case of gradual transition. In conclusion, the necessity of developing comprehensive computational models of epileptic seizures is emphasized.
本综述涵盖了癫痫的 EEG/MEG 信号处理和建模领域的最新进展，涉及发作间期和发作期现象。在第一部分中，介绍和讨论了分析发作间期 EEG/MEG 癫痫样棘波的主要方法。通过回顾源和容积导体模型，即等效偶极子源概念、充分的时间和空间采样要求、如何验证源解的问题，特别是通过比较头皮和颅内 EEG 信号、EEG 和 MEG 数据或同时记录的 EEG 与 fMRI（BOLD 信号）获得的解，来验证源解。在第二部分，考虑了用于描述发作的方法，即发作起始时棘波的偶极子建模、将发作 EEG 信号分解为一组正交时空分量，以及估计发作传播的线性和非线性方法。在第三部分，检查了发作间期和发作活动之间的转换的关键问题。特别是简要回顾了沿着发作预测道路努力的变化。有人认为，这个问题可以归结为随着发作的临近，估计神经元群体兴奋性状态的问题。强调了主动探测方法相对于被动分析方法的价值。在第四部分中，根据两种特殊类型的癫痫（以棘波和尖波放电为特征的失神发作和内侧颞叶癫痫），从建模的角度进行了考虑。这两种类型对应于向癫痫发作转变的不同情况，即前者是跳跃转变的情况，后者是典型的逐渐转变的情况。总之，强调了开发全面的癫痫发作计算模型的必要性。
