Abstracts

Rationale: Approximately half of the patients with a diagnosis of pharmaco-resistant epilepsy are potential candidates for epilepsy surgery. Successful epilepsy surgery may substantially reduce or eliminate epilepsy and the associated disability. In 30-40% of drug-resistant patients with focal epilepsy the cerebral areas responsible for seizure generation can only be defined by intracranial recordings with intracerebral or subdural electrodes. The correct pre-surgical identification of the epileptogenic zone with intracranial recordings has a direct impact on post-surgical outcome. For this purpose, an accurate analysis of ictal and interictal activity during and between seizures is required. At present, the identification of the epileptic zone is based on visual inspection of the intracranial electroencephalographic (EEG) patterns. One of the principal impediments to computer-driven analysis of intracranial signals is the complexity and the quantity of data recorded during pre-surgical stereo-EEG sessions. Methods: We selected a population of patients with extra-temporal neocortical dysplasias that showed similar ictal patterns and performed retrospectively a computer-assisted detection of the cortical region involved in seizures. Results: Reproducible ictal patterns observed in more then 10 seizures per patient were evaluated. Fast activity at 20-40 Hz was typically observed at the onset of seizures and during interictal paroxysms. During seizures, this pattern was followed by the appearance of faster activity (40-100 Hz) of lower voltage, coupled with a very slow wave deflection. Seizure termination was associated with a recovery of fast activity at 20-40 Hz that in most cases evolved into high amplitude bursts separated by brief EEG flattenings. Fast activities were further analyzed in the same group of patients. Stereo-EEG data were exported in ASCII or binary form and imported to a program developed in LabView for elaboration. Data evaluation algorithms were mainly based on the time, frequency and spatial domain analysis. Prevalent frequency was evaluated by Fourier transformation and further integral algorithms. As a final representation of the area that generated the typical ictal patterns, 3-D matrix of the measurements obtained from each recording channels were constructed based on magnetic resonance (MR) coordinates to obtain a visual representation and the spatial distribution of the analysed parameters. Conclusions: The study was sponsored by a Mariani Foundation grant 50-06.