Magnetic source imaging in posterior cortex epilepsies

Posterior cortex epilepsies (PCE) are characterized by less satisfying postoperative results than temporal lobe epilepsies and are thus challenging for non-invasive presurgical investigations. The objective of this study was to evaluate the performance of magnetic source imaging (MSI) in PCE, validating the results by the SEEG (stereoelectroencephalography) definition of irritative and epileptogenic zones (IZ and EZ). Fourteen PCE surgery candidates were investigated using MSI and SEEG. LCMV (Linearly Constrained Minimum Variance) and MUSIC algorithms were used. IZ was quantified using a semi-automatic detection of interictal spikes. EZ was quantified using the epileptogenicity index (EI) method that accounts for both the propensity of a brain area to generate rapid discharges and the time for this area to get involved in the seizure. EI values range from 0 (no epileptogenicity) to 1 (maximal epileptogenicity). Levels of concordance between MSI and IZ, MSI and EZ were determined as follows: A = localized on MSI and SEEG for the site of value 1 (IZ and EZ quantification), B = localized on MSI and SEEG for a part of the IZ or a structure involved in the EZ (without the maximal value 1), C = localized on MSI and not SEEG, D = localized on SEEG and not MSI, E = localized on MSI and SEEG, discordant for site. Five PCE cases were characterized by focal IZ, nine by distributed IZ between several distant brain areas. MSI allowed to determinate IZ in 4/5 focal IZ cases. In case of distributed IZ, levels of concordance were A (2 cases), B (4 cases) and D (3 cases). In most distributed cases, MSI allowed to localize only a part of the IZ. Medial temporal involvement in the IZ was frequent (9/12 cases) and not evidenced by MSI. The brain area that demonstrated the maximal value of EI was shown by MSI in four out of five (80 %) focal IZ cases, in two out of nine (22 %) distributed cases. MSI results depend on IZ characteristics. A distributed IZ organization presents difficulties for MSI and highlights the need for further methodological approaches.