Fully automated QRS area measurement for predicting response to cardiac resynchronization therapy

BACKGROUND: Cardiac resynchronization therapy (CRT) is an established treatment in patients with heart failure and conduction abnormalities. However, a significant number of patients do not respond to CRT. Currently employed criteria for selection of patients for this therapy (QRS duration and morphology) have several shortcomings. QRS area was recently shown to provide superior association with CRT response. However, its assessment was not fully automated and required the presence of an expert.

OBJECTIVE: Our objective was to develop a fully automated method for the assessment of vector-cardiographic (VCG) QRS area from electrocardiographic (ECG) signals.

METHODS: Pre-implantation ECG recordings (N = 864, 695 left-bundle-branch block, 589 men) in PDF files were converted to allow signal processing. QRS complexes were found and clustered into morphological groups. Signals were converted from 12‑lead ECG to 3‑lead VCG and an average QRS complex was built. QRS area was computed from individual areas in the X, Y and Z leads. Practical usability was evaluated using Kaplan-Meier plots and 5-year follow-up data.

RESULTS: The automatically calculated QRS area values were 123 ± 48 μV.s (mean values and SD), while the manually determined QRS area values were 116 ± 51 ms; the correlation coefficient between the two was r = 0.97. The automated and manual methods showed the same ability to stratify the population (hazard ratios 2.09 vs 2.03, respectively).

CONCLUSION: The presented approach allows the fully automatic and objective assessment of QRS area values.

SIGNIFICANCE: Until this study, assessing QRS area values required an expert, which means both additional costs and a risk of subjectivity. The presented approach eliminates these disadvantages and is publicly available as part of free signal-processing software.