Abstracts

Semi-automated selection and characterisation of interictal events in high density EEG for ESI-fMRI investigation

Abstract number : 3.275
Submission category : 5. Neuro Imaging
Year : 2011
Submission ID : 15341
Source : www.aesnet.org
Presentation date : 12/2/2011 12:00:00 AM
Published date : Oct 4, 2011, 07:57 AM

Authors :
S. J. Vogrin, S. Lau, , W. J. D'Souza, J. Haueisen, M. J. Cook

Rationale: To improve comprehensive identification and characterisation of interictal events in patients with focal epilepsy through semi-automated measurements of variability prior to electrical source imaging (ESI) and refined EEG-fMRI analysis.Methods: Participants' high density EEGs were recorded using a 128-channel cap in an electrically shielded room and during simultaneous EEG-fMRI acquired in a 1.5T Avanto Magnetom (Siemens, Erlangen Germany) at St Vincent's Hospital Melbourne (ethics approval HREC-A034/08). The MR-gradient and cardiac pulse-related artifacts were removed using average artifact subtraction (Scan 4.5, Neuroscan Compumedics, Charlotte NC, USA ) and PCA projection (Curry 6.0 Neuroscan Compumedics, Hamburg Germany). An experienced EEG reviewer initially identified discharges, which were averaged and used as a template for template matching to discover larger quantities of interictal events in the same recording. The average was subtracted from every event and the correlation matrix between any two such residuals was computed. Results: Artifact-corrupted events could be rejected using their correlation characteristics and the homogeneity of events within the set could be determined. The characterized inter-event variability provided an objective measure for subsequent ESI and EEG-fMRI analyses of the localisation of the refined event set.Conclusions: This novel, semi-automated approach offers improved characterization and interpretation of true epileptogenic features versus artifact compromised discharges, while assessable in a multimodal context. The benefit of enhanced concordant simultaneous ESI and fMRI localisation for interictal events shows promise in improving non-invasive characterisation in the presurgical evaluation of epilepsy.
Neuroimaging