Automatic Sensor Array Calibration and Coregistration for On-Scalp MEG

Accurate magnetoencephalography (MEG) localization of neural electrical activity requires registration between the MEG sensors and MRI. To reduce the geometric error of the sensor array and realize automatic registration, an array geometry calibration algorithm based on prior knowledge is proposed, which automatically completes the position and orientation calibration of optical pumped atomic magnetometer (OPM)-MEG array with the reference of the design model of MEG helmet. Full-process automatic coregistration based on optical scanning is realized. A position error of 0.23 ± 0.12 mm and orientation error of
$0.34 \pm 0.14 ^{\circ }$
are obtained in the calibration results of 85 sensors on-scalp MEG system. The proposed array geometric calibration (AGC) algorithm is used to quantify the deviation between the manufactured array and the theoretical array for the first time. The results show a mean position deviation of 0.57 mm and an angle deviation of 0.68° in the manufactured array, with the maximum position deviation of 1.45 mm and the maximum orientation deviation of 2.13°. It is indicated that the replacement of the manufactured array by the theoretical array in previous research may introduce significant errors in the MEG source localization results.