Reliability for non-invasive somatosensory cortex localization: Implications for pre-surgical mapping

In patients with epilepsy or space occupying tumors in cortical regions, surgical resection is often considered as the primary treatment. Pre-surgical neuroimaging can provide a detailed map of pathological and functional cortex, leading to safer surgery. Mapping can be achieved non-invasively using magnetoencephalography (MEG), and is concordant with invasive findings. However, the reliability of MEG mapping between sessions is not well established. The inter-session reliability is an important property in pre-surgical mapping to establish resection margins, but repeated scans are impracticable. The present study sought to quantify the intersession reliability of MEG localization of somatosensory cortex (S1).

Eighteen healthy individuals underwent MEG sessions on 3 consecutive days. Five participants were excluded due to technical issues during one of the three days. Each session included clinical-style S1 localization using electrical stimuli to each median nerve at sub-motor thresholds. The 35 ms peak of the somatosensory evoked field was used for localizing S1 in each session using a single equivalent current dipole model. Intersession reliability was quantified using two methods. Average Euclidean Distance (AED) quantified the difference in localization between each session and the inter-session mean localization. Session Euclidean Distance (SED) quantified the difference in localization between each pair of sessions.

Results showed the AED was 4.8 ± 1.9 mm, whereas the SED was 8.3 ± 3.4mm. While the AED values obtained parallel those reported previously in smaller samples, the SED values were substantially larger.

Clinicians should consider up to an 8mm confidence interval around the estimated location of S1 based on MEG pre-surgical mapping.