ECG electrode localization using 3D visual reconstruction

Figures

• 

  FIGURE 1. (A) RealSense D415 camera (left) mounted on an articulated arm to take photos from 12 fixed locations around the torso (right) (B) BioSemi strip electrodes fixed to a phantom torso and (C) on a healthy male volunteer.

• 

  FIGURE 2. Flowchart outlining the 3D electrode localization and labelling process.

• 

  FIGURE 3. Reconstruction of the torso using a 2D or 3D camera: (A) Acquisition of RGB and depth images; (B) Generated 3D point cloud from a single acquisition of the torso, denoised using DBSCAN, shown in true color (top) and in yellow (bottom); (C) Final 3D torso point cloud resulting from the alignment of N acquisitions of the torso using the ICP-based algorithm, displayed in true color (top) and with distinct colors representing each view of the torso used in the alignment (bottom).

• 

  FIGURE 4. ECG Electrode localization: (A) Example of YOLOv8 detection and classification of ECG electrodes in RGB image, where black is for class A, red for B, orange for C, and yellow for D; (B) Extraction of electrode locations as a 3D point cloud; (C) Alignment of all electrode point clouds; (D) Final 3D reconstruction of ECG electrodes; (E) Automated identification of the initial and secondary ECG electrodes in the RGB image; (F) Systematic Labeling and Arrangement of 3D ECG Electrode Positions.

• 

  FIGURE 5. Flowchart outlining the algorithm to automatically label each ECG electrode channel.

• 

  FIGURE 6. Gold standard methods for ECG electrodes localization process using the CT-scan (top) or an Electromagnetic Tracking System (bottom): (A) 3D acquisition of the phantom via CT-scan; (B) Reconstruction of the phantom mesh using MusiCardio software; (C) Final 3D reconstruction of ECG electrodes from CT-scan; (D) Electromagnetic Tracking System and its electromagnetic field generation space; (E) Localization of electrodes on the phantom with the Electromagnetic Tracking System; (F) Final 3D reconstruction of ECG electrodes from the Electromagnetic Tracking System.

• 

  FIGURE 7. Phantom Data Results: Euclidean distance (mm) between electrodes positions obtained using the CT-scan, electromagnetic tracking system (ETS), 3D camera and 2D camera in 3D (left), and for X, Y, and Z axes (right).

• 

  FIGURE 8. Clinical Data Results: Euclidean distance (mm) between electrode positions obtained using the electromagnetic tracking system (ETS), 3D and 2D camera in 3D (left), and for the X, Y, and Z axes (right). Data obtained from 7 healthy male volunteers (P1 to P7). Yellow rectangles indicate the maximum movement (across X, Y, and Z axes) of two fixed catheters for each patient, as presented in Table 2.

• 

  FIGURE 9. Location of two ETS marker on the torso of healthy volunteers.

• 

  FIGURE 10. Euclidean distances measured across the chest, abdomen, and torso sides in seven healthy volunteers, for ETS and the 3D camera (left), and ETS with the 2D camera (right).

• 

  FIGURE 11. Comparison of preparation, acquisition, processing, and total times between 2D, 3D, ETS, and CT-scan methods.