Statistical Evaluation of Transformation Methods Accuracy on Derived Pathological Vectorcardiographic Leads

Area of Science:

• Cardiology
• Biomedical Engineering
• Signal Processing

Background:

• Vectorcardiography (VCG) offers enhanced spatial insights into cardiac electrical activity compared to standard electrocardiography (ECG).
• VCG exhibits higher sensitivity for detecting myocardial infarction, ischemia, and hypertrophy.
• Clinical VCG measurement is uncommon, necessitating mathematical transformations to derive VCG leads from ECG.

Purpose of the Study:

• To compare the accuracy of mathematical transformation methods for deriving VCG leads from ECG.
• To evaluate the suitability of these derived VCG leads for clinical applications and disease prediction.

Main Methods:

• Four transformation methods were compared: Kors regression, Inverse Dower, QLSV, and Quasi-orthogonal transformation.
• The methods were applied to 30 myocardial infarction patient records from the PTB database.
• Accuracy was assessed using Mean Square Error (MSE) and statistical analysis (Kruskall-Wallis, Dunn test).

Main Results:

• The Kors regression method yielded the most accurate VCG lead transformations overall.
• Statistical analysis confirmed the superiority of the Kors regression method (p<0.05).

Conclusions:

• The Kors regression method is the most accurate for deriving orthogonal VCG leads.
• Accurate VCG lead derivation supports the development of systems for early cardiac disease diagnosis.