A three-dimensional CT-based method for precise localization of right ventricular pacing leads
K Wade, P Fruelund, C Stephansen, S Riahi, T ZarembaAbstract
Background
Right ventricular (RV) pacing site influences electrical and mechanical left ventricular (LV) activation. Current classification of RV lead position relies mainly on fluoroscopy or segmental analysis of CT images. Fluoroscopy has proven inaccurate when compared with CT-based localisation. Segmental CT classification of RV lead position may lead to inappropriate grouping of leads by placing anatomically close leads in separate segments. This may contribute to conflicting results reported in earlier studies.
Purpose
To introduce and validate a three-dimensional coordinate-based CT method for standardised and anatomically precise localization of RV pacing leads.
Methods
We analysed cardiac CT scans from 147 patients with dual-chamber pacemakers using open-source software. Contrast-enhanced scans obtained for clinical indications were used when available and otherwise study-specific cardiac CT scans were performed. Three LV landmarks were identified to define a cylindrical coordinate system aligned with the LV long axis. Each RV lead tip was then described by three spatial coordinates specifying its position within this coordinate system. Scans were mathematically standardised by rescaling the base-to-apex distance and the angle between the ventricular wedges. Reproducibility was assessed in 10 randomly selected scans by two independent observers. Validation was performed by comparative analysis against the traditional segmental approach for RV lead position.
Results
Anatomical landmark localisation was highly reproducible with greater consistency in study protocol CT scans (interobserver variability 1.7 ± 1.1 mm and intraobserver variability 0.9 ± 0.7 mm) than in non-protocol scans (interobserver variability 4.8 ± 3.0 mm and intraobserver variability 3.2 ± 2.0 mm). The coordinate-based method demonstrated substantial agreement with the classical segmental method with a Cohen's kappa coefficient of 0.70 (95 % CI: 0.58 to 0.82). Among 147 analysed lead positions 47 were classified differently by the two methods. Of these discordant cases, 39 (83%) were located within 5 mm of a segmental border. Overall, 108 leads (73%) lay within 5 mm of a segmental boundary.
Conclusion
This CT-based method allows a reproducible and standardised description of RV lead position, offering a robust methodological foundation for future comparative studies. The concentration of leads near segmental borders highlights the methodological value of a continuous spectrum approach for lead localisation.