Epilepsy Signal Analysis And Lesion Location

Epileptic focus localization in the treatment of drug-resistant epilepsy is a critical issue. However, the method used in the clinical treatment of epilepsy depends on the invasive detection electrode to achieve detection of the lesion. But invasive detection methods will introduce greater clinical risk. More and more research is gradually inclined to non-invasive detection methods. This article is proposed to achieve non-invasive detection of epileptic foci solutions EEG and head model.In this paper, firstly, the boundary element model was used to build the individual head geometry based on the MRI. Secondly, three analysis methods, Power spectral density based on FFT, continuous wavelet transform and multivariate empirical mode decomposition were used for features extraction from the 19-channel scalp EEG recordings of nine patients(3-18 yr) with epilepsy. Thirdly, two inverse solution, Beamformer and eLORETA were used for epileptic foci localization. Finally, the sensitivity of different localization measures are evaluated. Combing the different feature selection measures with two inverse solutions, we can achieve eight epileptic foci location measures. In all these measures, the Beamformer based on MEMD(MEMD-BF) has the highest sensitivity in all localization measures. However, almost all of the foci location based on the e LORETA are invalid.In this paper, the 19 channels scalp EEG signals of 9 pediatric patients are analuzed, and the different analysis methods and different inverse solutions are caompared. At last the MEMD-BF is a promising localization measures for tracing the seizure source.