Authors :
Presenting Author: Aaron Pope, MD – University of Texas Southwestern
Afsaneh Talai, MD – University of Texas Southwestern; Natascha Cardoso Da Fonseca, PhD – University of Texas Southwestern; Angela Price, MD – Pediatric Neurosurgery – University of Texas Southwestern; Elizabeth Davenport, PhD – University of Texas Southwestern; Andrea Lowden, MD – Associate Professor, Pediatric neurology and Epilepsy, University of Texas Southwestern
Rationale: Magnetoencephalography (MEG) is increasingly utilized as an integral part of epilepsy surgery evaluation. MEG detects changes in magnetic fields outside the head that are created by electrical neuronal activity. Functional data obtained with MEG is combined with magnetic resonance (MR) imaging to create magnetic source imaging (MSI). Source localization can be achieved via multiple techniques, with single equivalent current dipole (sECD) analysis being the most widely used and accepted method. However, sECD is not without limitations that could be overcome by alternative source location techniques, such as beamforming with power spectrum density mapping. The advantages of this alternative method include relatively less user interaction, can allow confirmation of sECD results, and can be used for localization when the cortical activity cannot be analyzed by the sECD method.
Methods: Data from the UT Southwestern Medical Center MEG and the Children’s Health Dallas Epilepsy Surgery databases were used to identify pediatric patients who had completed MEGs and stereoencephalography between 2/19/2020 and 1/24/2023. This retrospective chart review was approved by the institutional review board (IRB). The patients’ functional MEG data and MR brain imaging were imported and co-registered using Brainstorm. Source localization was achieved using a linear-constraint minimum-variance (LCMV) beamformer. Interictal epileptiform discharges were identified, then power spectrum density maps were created from selected discharges. Each discharge was mapped using the following three frequency groups, 1) +/-1 Hz of the measured frequency of the selected discharge, 2) 20-25 Hz, and 3) a combination of 5-14 Hz and/or 14-50 Hz. Results from each frequency grouping were then averaged to create a single power spectrum density map for that specific frequency group. The average power spectrum density maps were then compared to the patients’ previously obtained results from sECD (Figure 1). Comparisons were then graded based on their level of concordance.
Results: Twenty-nine patients met inclusion and exclusion criteria, with eleven patients thus far having completed the beamforming process and preliminary analysis. The majority of subjects, from each frequency grouping were rated as concordant (1) +/-1 Hz of the measured frequency of the selected discharge – 88%, 2) 20-25 Hz – 73%, and 3) a combination of 5-14 Hz and/or 14-50 Hz – 91%. When individual reviewers’ responses were compared (using a Fisher exact test), no statistically significant difference, in concordance, was noted between the three frequency groups.
Conclusions: Preliminary data suggest that beamforming with power spectrum density mapping offers similar localization to the single equivalent current dipole method. Each of the three frequency band groups offer a similar level of concordance when compared to one another. These results, although limited by a small sample size, support the idea that beamforming with power spectrum density mapping could allow for confirmation of sECD results or could be used when localization with sECD cannot be assessed. Further analysis with a larger sample size are needed to reach more conclusive results.
Funding: No funding needed