Combination of diagnostic classifications from ECG and VCG computer interpretations

Area of Science:

• Cardiology
• Medical Imaging
• Computational Medicine

Background:

• The Common Standards for Quantitative Electrocardiography (CSE) study highlighted the superiority of combined expert or program classifications over individual ones.
• Previous research suggests that combining diagnostic tools can enhance accuracy in electrocardiographic (ECG) interpretation.

Purpose of the Study:

• To evaluate if combining ECG and vectorcardiographic (VCG) interpretations yields better diagnostic results than using either method alone.
• To explore the efficacy of combining ECG with a reconstructed VCG (rVCG) and single-lead ECG (sECG) interpretations.

Main Methods:

• Utilized the CSE diagnostic database (n=1,220) for performance testing.
• Compared the diagnostic accuracy of ECG alone, VCG alone, and combined ECG+VCG, ECG+rVCG, and sECG interpretations.
• Investigated the use of a simulated VCG (rVCG) derived from the 12-lead ECG and combined interpretation of single beats (sECG).

Main Results:

• Combined interpretations (ECG+VCG, ECG+rVCG, sECG) showed significantly higher performance (74.2%, 73.6%, 71.2%) than individual ECG (69.8%) or VCG (70.2%) interpretations (p < 0.001).
• The combination of ECG with VCG or rVCG demonstrated the highest diagnostic accuracy.
• No significant difference in performance was found between the sECG combination and VCG alone.

Conclusions:

• Combining ECG and VCG interpretations significantly enhances diagnostic classification accuracy.
• The use of a reconstructed VCG (rVCG) offers a viable alternative when a simultaneous VCG recording is unavailable.
• Combined interpretation strategies represent a valuable advancement in quantitative electrocardiography.