Aarhus University Seal / Aarhus Universitets segl

Ictal EEG source imaging in presurgical evaluation: High agreement between analysis methods

Research output: Contribution to journal/Conference contribution in journal/Contribution to newspaperJournal articleResearchpeer-review

  • Sándor Beniczky
  • Ivana Rosenzweig
  • ,
  • Michael Scherg
  • ,
  • Todor Jordanov
  • ,
  • Benjamin Lanfer
  • ,
  • Göran Lantz
  • ,
  • Pål Gunnar Larsson

PURPOSE: To determine the agreement between five different methods of ictal EEG source imaging, and to assess their accuracy in presurgical evaluation of patients with focal epilepsy. It was hypothesized that high agreement between methods was associated with higher localization-accuracy.

METHODS: EEGs were recorded with a 64-electrode array. Thirty-eight seizures from 22 patients were analyzed using five different methods phase mapping, dipole fitting, CLARA, cortical-CLARA and minimum norm. Localization accuracy was determined at sub-lobar level. Reference standard was the final decision of the multidisciplinary epilepsy surgery team, and, for the operated patients, outcome one year after surgery.

RESULTS: Agreement between all methods was obtained in 13 patients (59%) and between all but one methods in additional six patients (27%). There was a trend for minimum norm being less accurate than phase mapping, but none of the comparisons reached significance. Source imaging in cases with agreement between all methods was not more accurate than in the other cases. Ictal source imaging achieved an accuracy of 73% (for operated patients: 86%).

CONCLUSION: There was good agreement between different methods of ictal source imaging. However, good inter-method agreement did not necessarily imply accurate source localization, since all methods faced the limitations of the inverse solution.

Original languageEnglish
JournalSeizure - European Journal of Epilepsy
Pages (from-to)1-5
Number of pages5
Publication statusPublished - Dec 2016

See relations at Aarhus University Citationformats

ID: 109013934