Neurophysiological parameters and neuroimaging data in predicting the course of structural focal epilepsy

Abstract

The complex interactions of the epileptic focus and the brain systemic response require an integrated approach to predicting the course of focal epilepsy. Objective: to investigate the role of interictal neurophysiological parameters and neuroimaging data in predicting the course of epilepsy. Patients and methods. Eighty-two patients with focal structural epilepsy and 82 healthy participants (a control group) were examined. Clinical, psychological, and social characteristics and the data of neuroimaging and comprehensive neurophysiological studies (electroencephalography (EEG), recording exogenous and cognitive evoked potentials, motor testing, and heart rate variability) were assessed. Results and discussion. The investigators identified the prognostic factors of the unfavorable course of epilepsy: temporal lobe epilepsy, left temporal lobe lesion, nontraumatic intracerebral hemorrhages. They proposed algorithms for predicting the course of epilepsy based on neurophysiological and neuroimaging data. An analysis of physiological parameters in patients with the unfavorable course of epilepsy demonstrated the slowing of the background rhythm on the EEG, a decrease in the power of specific afferentation, and an increase in the decision-making time for the stimulus, as evidenced by the P300 potential, and insufficiency of the central mechanisms in providing a motor reaction. These patients also showed the enhanced activity of stress-implementing systems. Taking into account not only neurophysiological parameters, but also neuroimaging data could improve the prognostic capabilities of an artificial neural network that determines the type of a disease course. Conclusion. The unfavorable course of focal epilepsy is associated with a number of clinical and physiological parameters; in this case, it is possible to identify the specific physiological pattern of this course of the disease. The integrated clinical and physiological approach and neuroimaging data make it possible to successfully predict the course of the disease through machine learning technology.