Related Experiment Video
Updated: Jul 23, 2025
10:17
Real-Time Cardiac Mapping with a Noninvasive Imageless Electrocardiographic Imaging System
Published on: April 11, 2025
714
Improving localization accuracy for non-invasive automated early left ventricular origin localization approach.
Shijie Zhou1,2, Raymond Wang3, Avery Seagren1
1The Department of Chemical, Paper and Biomedical Engineering, Miami University, Oxford, OH, United States.
Frontiers in Physiology
|July 12, 2023
Summary
The K-nearest neighbors algorithm (KNN) significantly improves the accuracy of a non-invasive approach for localizing ventricular arrhythmia origins. This method reduces projection errors, offering a promising tool for clinical use.
Area of Science:
- Cardiology
- Biomedical Engineering
- Medical Imaging
Background:
- A non-invasive method using 12-lead ECG was developed to localize left ventricular activation origins.
- The previous method projected sites onto a generic LV surface using the smallest angle (SA) algorithm.
Purpose of the Study:
- To enhance localization accuracy by employing the K-nearest neighbors (KNN) algorithm.
- To reduce projection errors in non-invasive arrhythmia origin localization.
Main Methods:
- Utilized two datasets: 1012 LV pacing sites and 25 clinical VT exit sites.
- Predicted coordinates from QRS integrals and projected them using KNN versus SA algorithms.
Main Results:
- KNN demonstrated significantly lower mean localization error than SA in both datasets (p < 0.05).
- Bootstrap analysis confirmed KNN's superior predictive accuracy (p < 0.05).
Conclusions:
- KNN significantly reduces projection error, improving non-invasive localization accuracy.
- This enhanced approach shows potential for identifying ventricular arrhythmia origins in clinical settings.
