Parameter choice methods and temporal filtering for the generalized eigensystem method applied to the inverse problem...

Area of Science:

• Biomedical Engineering
• Computational Electrophysiology
• Medical Imaging

Background:

• Estimating cardiac electrical potentials from body surface measurements is crucial for diagnosing heart conditions.
• The generalized eigensystem (GES) method has been proposed for this inverse problem.
• Classical Tikhonov regularization is a common approach for solving inverse problems.

Purpose of the Study:

• To present an alternative formulation of the generalized eigensystem (GES) method.
• To compare the new GES formulation with zero-order Tikhonov regularization.
• To evaluate the effectiveness of incorporating temporal information into these methods.

Main Methods:

• A novel formulation of the generalized eigensystem (GES) method was developed, resembling Tikhonov regularization.
• The new GES method and zero-order Tikhonov regularization were compared using swine experimental data.
• Moving average filtering was applied to incorporate temporal information into the estimates.

Main Results:

• The alternative GES formulation, with a single regularization parameter, was presented.
• GES methods demonstrated superior performance compared to Tikhonov regularization in the swine experiment.
• Temporal filtering via moving average was more effective for GES than for Tikhonov regularization.

Conclusions:

• The revised GES method provides a robust alternative for estimating cardiac electrical potentials.
• Incorporating temporal information significantly enhances the accuracy of GES-based estimations.
• This work advances non-invasive cardiac electrophysiological monitoring techniques.