Spatial localization of EEG electrodes using 3D scanning

Abstract

Objective. A reliable reconstruction of neural activity using high-density electroencephalography (EEG) requires an accurate spatial localization of EEG electrodes aligned to the structural magnetic resonance (MR) image of an individual’s head. Current technologies for electrode positioning, such as electromagnetic digitization, are yet characterized by non-negligible localization and co-registration errors. In this study, we propose an automated method for spatial localization of EEG electrodes using 3D scanning, a non-invasive and easy-to-use technology with potential applications in clinical settings. Approach. Our method consists of three main steps: (1) the 3D scan is ambient light-corrected and spatially aligned to the head surface extracted from the anatomical MR image; (2) electrode positions are identified by segmenting the 3D scan based on predefined colour and topological properties; (3) electrode labelling is performed by aligning an EEG montage template to the electrode positions. The performance of the method was assessed on data collected in eight participants wearing high-density EEG caps with 128 sensors, from three different manufacturers. We estimated the co-registration error using the distance between the MR-based head shape and the closest 3D scan points. Also, we quantified the positioning error using the distance between the detected electrode positions and the corresponding locations manually selected on the 3D scan data. Main results. For all participants and EEG caps, we obtained a median error of co-registration below 3.0 mm and of spatial localization below 1.4 mm. The method based on 3D scanning data was significantly more precise compared to the electromagnetic digitization technique, and the total time required for obtaining electrode positions was reduced by about half. Significance. We have introduced a method to automatically detect EEG electrodes based on 3D scanning information. We believe that our work can contribute to a more effective, reliable and widespread use of high-density EEG as brain imaging tool.

More Reliable EEG Electrode Digitizing Methods Can Reduce Source Estimation Uncertainty, But Current Methods Already Accurately Identify Brodmann Areas

“…Both the structured-light and infrared 3D scanning methods were more reliable than digitizing with the ultrasound method. Furthermore, our reliability results for the two 3D scanners align well with a recent study that showed that an infrared 3D scan could automatically digitize electrode locations on three different EEG caps and achieve good reliability after additional post-processing [39]. Of the two 3D scanners we tested, the less expensive infrared 3D scanner was more reliable, had higher validity, and resulted in less dipole uncertainty, compared to the structured-light 3D scanner.…”

“…Recently, common EEG analysis toolboxes such as EEGLAB [37] and FieldTrip [38] support using 3D scanners to digitize the electrode locations. Studies suggest that 3D scanners can improve digitization accuracy and significantly reduce digitization time [39]. Using other camerabased systems such as time-of-flight scanners and virtual reality headsets were also reported to provide comparable digitization reliability as the ultrasound or electromagnetic digitizing methods, while reducing the time spent for digitizing the EEG electrodes [27], [40], [41].…”

SPOT3D: Spatial positioning toolbox for head markers using 3D scans

“…The GUI in SPOT3D enables the user to correct imprecisions in the spatial registration of the individual 3D scan to MR space, as well as in the identification of head marker positions. This enhances the positioning accuracy and the overall usability of the 3D scanning technique in comparison with our previous automated method 17 .…”

“…Thanks to the colour information and high resolution of the 3D scan, surface markers on the individual’s scalp can be localized with low positioning errors 4,16 . In our most recent work, we introduced an automated method for accurate spatial registration of a 3D scan to the structural MR image, as well as for detecting and labelling EEG sensors 17 .…”

The interaction between endogenous GABA, functional connectivity and behavioral flexibility is critically altered with advanced age

“…For the individual geometrical description of the head (mesh), the anatomical image was segmented into 12 tissue classes (skin, eyes, muscle, fat, spongy bone, compact bone, cortical gray matter, cerebellar gray matter, cortical white matter, cerebellar white matter, cerebrospinal fluid and brain stem), based on the MIMA model [87] using SPM12 (http://www.fil.ion.ucl.ac.uk/spm/software/spm12/) as described previously [88–90]. The EEG electrode positions were rigidly co-registered to the individual head surface (skin contour) by projecting the electrode coordinates in the native space through a rigid-body transformation, based on: (i) the estimation of anatomical landmarks (nasion, left/right peri-auricular points), (ii) the alignment of the electrode positions on the head surface through Iterative-Closest Point registration, and (iii) the projection of the electrodes onto the surface choosing the smallest Euclidean distance [91]. Conductivity values for each tissue class were grounded on previous findings [92, 93].…”