Standardization of MEG sensors by the Signal Space Separation method

Abstract

Abstract Different head positions of the subjects hamper the comparison of data from different measurement sessions. Signal space separation (SSS) is a method that, besides its ability to suppress external interference signals, can be used for standardization of MEG data. In order to evaluate the performance of SSS in transforming MEG data between different head positions, three different kinds of steps were followed. The first step included phantom measurements by placing a spherical phantom in seven different positions. One position was used as reference and six other positions were used for representing different head positions. All data were transformed to correspond to the reference position. In addition to visualizing the recorded signals before and after the transformation, amplitude-weighted correlation algorithm was applied to compare the transformed and the reference data. Our results reveal excellent performance of the SSS method with phantom simulations of superficial medium-strength dipoles (100 nAm) or deeper high-strength dipoles (1000 nAm) activation. For the second step of the evaluation, data from two healthy subjects were recorded. In separate measurement runs, the subjects were asked to place their heads in four different positions as well as in one reference location within the sensory array. Auditory evoked fields were recorded. The same evaluation procedure was followed as in phantom measurements resulting in very satisfactory results. The third step included SSS transformation of data from a pharmacological study with auditory and visual stimulation on 13 subjects. Head location in the centre of the sensor array from one session was used as a reference. After the transformation, significantly stronger responses were revealed in some cases over channels around the responding cortical area. The data were generally less noisy, likely due to suppression of external fields by SSS. The SSS transformation seems to substantially promote the reproducibility and comparability of measured MEG waveforms.