Background: Severe left bundle branch block (LBBB) is treated with cardiac resynchronisation therapy (CRT), however more than 30% treated with CRT are non-responders, likely due to absence of electrical dyssynchrony. In order to improve patient selection, interventricular dyssynchrony can potentially be quantified by solving the inverse electrocardiography (ECG) problem. Existing solutions are based on >100 body surface measurements and patient specific geometries, which limits the clinical applicability. This study aims to determine the electrical activation of the ventricles from standard 12-lead ECG recordings and a generic heart-torso model, and quantify the interventricular electrical dyssynchrony.
Method: Standard 12-lead ECG recordings from 100 normal subjects, 100 LBBB subjects, and 135 subjects admitted to CRT were included. The ventricular myocardial activation was determined in six dipoles. The accuracy and robustness of Gaussian regularisation and Tikhonov regularisation with regularisation parameter λ=0 and λ=30 was compared. Interventricular dyssynchrony was determined as the left-right ventricular uncoupling (LRVU) and compared to QRS duration, QRS area and Q-LV.
Results: Tikhonov λ=0 yielded the most accurate ECG reconstruction in both normal and LBBB subjects. LRVU was higher in LBBB subjects compared to normal subjects. The correlation of LRVU with QRS duration and QRS area was higher in LBBB(r^2 =0.359, 0.791) and CRT subjects(r^2=0.328, 0.630) compared to normal subjects(r^2=0.178, 0.093). Furthermore LRVU was correlated with Q-LV in CRT subjects (r^2=0.441).
Conclusion: Interventricular dyssynchrony can be quantified by solving the inverse ECG problem by standard 12-lead ECG and a generic heart-torso model.