Visualizing spikes in source-space: Rapid and efficient evaluation of magnetoencephalography

Sándor Beniczky*, Lene Duez, Michael Scherg, Peter Orm Hansen, Hatice Tankisi, Per Sidenius, Anne Sabers, Lars Hageman Pinborg, Peter Uldall, Anders Fuglsang-Frederiksen

*Corresponding author af dette arbejde

Publikation: Bidrag til tidsskriftArtikelForskningpeer review


OBJECTIVE: Reviewing magnetoencephalography (MEG) recordings is time-consuming: signals from the 306 MEG-sensors are typically reviewed divided into six arrays of 51 sensors each, thus browsing each recording six times in order to evaluate all signals. A novel method of reconstructing the MEG signals in source-space was developed using a source-montage of 29 brain-regions and two spatial components to remove magnetocardiographic (MKG) artefacts. Our objective was to evaluate the accuracy of reviewing MEG in source-space.

METHODS: In 60 consecutive patients with epilepsy, we prospectively evaluated the accuracy of reviewing the MEG signals in source-space as compared to the classical method of reviewing them in sensor-space.

RESULTS: All 46 spike-clusters identified in sensor-space were also identified in source-space. Two additional spike-clusters were identified in source-space. As 29 source-channels can be easily displayed simultaneously, MEG recordings had to be browsed only once. Yet, this yielded a global coverage of the recorded signals and enhanced detectability of epileptiform discharges because MKG-artefacts were suppressed and did not impede evaluation in source-space.

CONCLUSIONS: Our results show that reviewing MEG recordings in source-space is accurate and much more rapid than the classical method of reviewing in sensor-space.

SIGNIFICANCE: This novel method facilitates the clinical use of MEG.

Sider (fra-til)1067-1072
Antal sider6
TidsskriftClinical Neurophysiology
Udgave nummer2
StatusUdgivet - feb. 2016


Udforsk hvilke forskningsemner 'Visualizing spikes in source-space: Rapid and efficient evaluation of magnetoencephalography' indeholder.