Automatic detection of fiducials landmarks toward development of an application for EEG electrodes location (digitization): Occipital structured sensor based-work

TL;DR

This paper presents an algorithm for automatic detection of fiducial landmarks on 3D head scans to facilitate automatic EEG electrode placement, integrating computer vision, MATLAB tools, and 3D spatial analysis.

Contribution

It introduces a novel method combining 3D to 2D projection, custom object detection, and coordinate reprojection for automatic fiducial detection on head scans.

Findings

Successful automatic detection of fiducials on 3D head models.

Integration of MATLAB and FieldTrip Toolbox for feature detection.

Foundation for developing an automatic EEG electrode digitization application.

Abstract

The electrophysiological source imagine reconstruction is sensitive to the head model construction, which depends on the accuracy of the anatomical landmarks locations knowns as fiducials. This work describes how to perform automatic fiducials detection, towards development of an application for automatic electrodes placement (digitization), over a three-dimensional surface of a subject head, scanned with the Occipital Inc. structure sensor ST01. We offer a wide description of the proposed algorithm to explore the three-dimensional object to features detection, by means of: dimensional reduction with perspective projection from 3D to 2D, object detection with custom detectors, robotic control of mouse motion and clicks events and reprojection from 2D to 3D to get spatial coordinates. This is done taking into account the characteristics of the scanner information, the training process of detectors with Computer-Vision Toolbox resources of MATLAB R2018b, the integration of FieldTrip Toolbox and the main properties of several ways to represents pixels; putting together all those things to automatically find the fiducials landmarks to generate the subject's coordinate system. All this result is presented as the initial state of a project focused on developing one application for automatic electrode digitization.