ECG electrode localization using 3D visual reconstruction.

Area of Science:

• Biomedical Engineering
• Medical Imaging
• Electrophysiology

Background:

• Body surface potential maps (BSPMs) offer advanced electrophysiological insights beyond standard ECG.
• Accurate electrode localization is crucial for reliable BSPM generation.
• Current localization methods can be complex or require specialized equipment.

Purpose of the Study:

• To develop and validate AI-based methods for automatic electrode localization in BSPMs.
• To compare the accuracy of a 3D Depth Sensing camera method with a 2D camera method.
• To assess the clinical feasibility of these AI-driven localization techniques.

Main Methods:

• Developed two AI algorithms for electrode detection: one rapid (3D DS camera) and one versatile (2D camera).
• Validated both methods against CT scans and Electromagnetic Tracking Systems (ETS) using a phantom model.
• Tested the methods on 7 healthy volunteers to determine real-world accuracy.

Main Results:

• Both 3D DS and 2D camera methods achieved sub-2mm localization error on the phantom model.
• Volunteer studies showed average 3D Euclidean distances between 2.45 ± 1.32 mm and 5.78 ± 3.09 mm.
• The AI methods demonstrated comparable accuracy to established tracking systems.

Conclusions:

• AI-powered electrode localization using 3D or 2D cameras is highly accurate for BSPMs.
• These methods provide practical and potentially cost-effective alternatives to traditional imaging.
• The findings may increase clinical adoption and utility of BSPMs in diagnosing cardiac conditions.