[Wolff-Parkinson-White syndrome : Comparison of different algorithms]

Area of Science:

• Cardiology
• Medical Imaging
• Artificial Intelligence

Background:

• Wolff-Parkinson-White (WPW) syndrome involves accessory pathways bypassing normal cardiac conduction.
• Accurate preprocedural localization of these pathways is crucial for effective ablation, complication reduction, and minimizing radiation exposure.

Purpose of the Study:

• To systematically review and analyze ECG-based algorithms for accessory pathway localization in WPW syndrome.
• To compare the diagnostic performance of classical, modern rule-based, and deep learning (DL) approaches.

Main Methods:

• Systematic review of ECG-based algorithms for accessory pathway localization.
• Analysis of rule-based algorithms (e.g., EASY-WPW, SMART-WPW) and DL models.
• Comparison of accuracy, sensitivity, and specificity against classical methods.

Main Results:

• Classical algorithms showed variable accuracy (72%-92%).
• Modern rule-based algorithms (EASY-WPW, SMART-WPW) achieved high accuracy (93%-97%) with excellent sensitivity and specificity (>90%).
• DL approaches demonstrated 84% accuracy (AUROC 0.92), outperforming classical algorithms, and enabling automated analysis with reduced variability.

Conclusions:

• Validated ECG algorithms and DL models are valuable for preinterventional planning in WPW syndrome.
• Modern rule-based algorithms provide excellent diagnostic accuracy exceeding 90% sensitivity and specificity.
• AI integration and multimodal strategies are expected to further enhance accessory pathway localization accuracy.