Novel method to localize EEG electrodes emplying the optical 3D sensor "Flying Triangulation"

The goal of this thesis is to develop a novel method to localize the centers of EEG electrodes in 3D space by employing an optical sensor based on "Flying Triangulation".

Accurate electrode coordinates enable an activity source analysis of EED data and yield valuable information for presurgical evaluation and plannig of epilepsy surgery. The electrode lcalizations combined with the activies recorded by an EEG sensor, together with magnetoencephalography (MEG) and magnet resoncance imaging (MRI) measurement, yield a nerological examination which can be sued to detect sources in a human brain responsible for epileptic seizures. These regions can then be removed by a surgeon in order to improve the patient's health. Seizures happen when the electrical system of the brain malfunctions. In this work the focus lays on the localization of EED electrode positions.

EEG records spontaneous electrical activity of the brain over a short period of time, detected by multiple electrodes placed on a scalp. The electrodes are usually joined in an EEG cap placed on the patient's head. In neurology, epileptic activity can create abnormalities in an EEG study. The 3D electrode positions are captured to determine the potential activity sources within the brain, in combination with recorded EEG signals.

The state-of-the-art method for digitizing the localizations of the EEG electrodes is based on employing a digitizing pen called "Polhemus". The procedure is as follows: The tip of the Polhemus pen is inserted into the center of each electrode of the EEG cap placed on a patient's head. The position is captured by pressing a button, the Plhemus pen is moved, and positioned at the next electrode center and so on. The order of the electrodes to be localized is fixed and needs to be adhered to. However, the sate-of-the-art method has several dawbacks: Commonly, an EED cap consists of more than 60 electrodes. This makes a tactile capturing of the electrodes a time consuming taks. It usally takes more than 20 minutes to acquire the positions of 68 electrodes. Further, interactive methods are user dependent which means that the positions may differ when captured by different persons or at different times.

For these reasons, we presented a novel EEG electrode localization method which employs an optical 3D sensor based on the measurment principle called "Flying Triangulation". Flying Triangulation enables a freely hand-guided measurement of objects with real-time visualization of the current measurement result. The single 3D views are algned and visualized in real-time, allowing the user to get immediate feedback about non-measured object parts. Further, the sensor has been optimized to show minimal measuerment of uncertainty of ~150 µm. Hence, employing an optical 3D sensor based on Flying Triangulation promises to simplify the localization procedure and to improve the repeatability and accuracy of the localization.

The state-of-the-art method employing the digitizing pen "Polhemus" is compared to the novel method employing a Flying Trinagulation sensor. For this purpose, a 3d-printed head model (made of plaster-like material) of a person wearing an EEG cap has been created from a 3D point cloud obtined with an optical 3D sensor based on the widely applied "Fringe Projection" measurement rpinciple. The electrode positions determined from the 3D point cloud serve as ground truth for the comparison.

The physical heas model was then used to measure the electrode positons with both methods, employing Polhemus and Flying Triangulation sensors. From all trhee resulting data sets (ground truth, Polhemus, Flying Triangulation) the electrode positions were extracted for comparison. The data sets have been registered in order to be in one common coordinate system. The Eculidean distances of the corresponding positions of electrodes and their standard deviation  have been calculated . The standard deviation of Fringe to Polhemus data is σ = 3.39 mm and the deviation of Fringe to Flying Triangulation data is σ = 0.97 mm. Thus, employing the novel method based on Flying Triangulation improved the accuracy of electrode localization by a factor of 3.5

A reason for the large deviation of the Polhemus data to the ground truth might be caused by the movement of the head model during the 20 minute measurement inside the MEG chamber. Capturing the head with Flying Triangulation took less than a minute. While the post-processing steps necessary for the extraction of the electrode positions are still time consuming, they promise to be automatable and to be speeded up.