Noninvasive myocardial activation time imaging: a novel inverse algorithm applied to clinical ECG mapping data

Area of Science:

• Biomedical Engineering
• Computational Electrophysiology
• Medical Imaging Analysis

Background:

• Linear methods like minimum-norm least-square struggle with electrocardiographic (ECG) mapping data for heart activation time estimation.
• Nonlinear problem formulations requiring regularization are necessary, with Gauss-Newton being a standard but flawed approach.
• Clinical application of Gauss-Newton is hindered by unclear regularization parameter determination.

Purpose of the Study:

• To develop and evaluate a novel iterative algorithm for accurate heart activation time map estimation from ECG data.
• To address the limitations of the standard Gauss-Newton approach, particularly under clinical conditions.
• To compare the performance of the proposed iterative method against the Gauss-Newton approach using simulated and patient data.

Main Methods:

• Developed an iterative algorithm that solves the nonlinear problem through a sequence of regularized linear problems.
• Incorporated an L-curve computation at each iteration to determine the optimal regularization parameter.
• Applied both the novel iterative approach and the standard Gauss-Newton method to simulated ECG mapping data and clinical patient data.

Main Results:

• The proposed iterative algorithm demonstrated excellent numerical and computational performance.
• The novel approach maintained high performance under clinical conditions where the Gauss-Newton method began to fail.
• Both simulated and real patient data showed the superiority of the new iterative method.

Conclusions:

• The novel iterative algorithm offers a robust and accurate solution for estimating heart activation time maps from ECG data.
• This method overcomes the regularization challenges faced by the standard Gauss-Newton approach in clinical settings.
• The findings suggest a significant advancement in ECG-based cardiac mapping techniques.