Magnetoencephalographic validation parameters for clinical evaluation of interictal epileptic activity

The authors demonstrate that the confidence volume (the spatial volume that encompasses the 95% probability of source localization) of the single equivalent current dipole is helpful invalidating magnetoencephalographic epileptic spike mapping. Such mapping involves distinguishing spikes from other neuronal events. The usual criteria for validating dipole fit reliability involve four parameters-correlation coefficient (R greater than or equal to 0.98), goodness of fit (greater than or equal to 0.95), root mean square magnetic field value (> 400 fT), and dipole moment (Q value > 200 nAm)-but other parameters (direction of dipole moment, location of dipole, and confidence volume) can be considered. In 21 patients with epilepsy, the average correlation coefficient for 608 epileptic spikes was 0.99; average goodness of fit, 0.98; average root mean square, 1,198 fT; and the average Q value, 370 nAm. The mean average confidence volume was 0.30 +/- 0.27 cm(3). Correlation coefficient values for quiet brain activity were less than 0.90; goodness of fit values, less than 0.85; and confidence volumes were large (> 5 cm(3)); and for noise runs (no subject) they were even larger (> 100 cm(3)), although correlation coefficient values were more than 0.80 and goodness of fit values were more than 0.85. Confidence volumes for noise data are large-for background brain activity even larger-but confidence volumes for epileptic spikes are small. Confidence volume, in conjunction with other parameters, may be a robust parameter for spike selection.