Independent component analysis of subdurally recorded occipital seizures

Objective To study the components of occipital seizures using independent component analysis (ICA) of subdurally recorded electroencephalogram (EEG) data. Methods Twenty-seven subdurally recorded occipital seizures from eight patients were studied. ICA was performed, and the resulting independent components (ICs) were compared with respect to their power, frequency, degree of spread within the epileptogenic zone, and propagation patterns. The ictal ICs were further subdivided into propagating and non-propagating types. Results ICA provided information over and above that supplied by standard visual EEG analysis, confirming that all the seizures originated in the occipital lobe, and affording additional knowledge regarding the nature of the seizure onset zone. Each seizure was composed of multiple ICs, some of which propagated while others remained within the epileptogenic zone. There was no statistical difference between the propagating and non-propagating ICs with respect to power or frequency. However, propagating ICs involved a significantly greater number of recording electrodes at their onset when compared to the non-propagating ICs. Conclusions The propagation likelihood of ICs in occipital seizures is independent of their power or frequency, but it is dependent on the volume of brain giving rise to that signal; ICs that are generated by a greater volume of brain within the epileptogenic zone are more likely to propagate. Significance This study shows that ICA has the potential to be used to redefine the epileptogenic zone and guide the extent of cortisectomy for the treatment of patients with medically intractable occipital epilepsy.