Use of genetic algorithm for selection of regularization parameters in multiple constraint inverse ECG problem

Area of Science:

• Biomedical Engineering
• Computational Electrophysiology
• Medical Imaging

Background:

• The inverse electrocardiography problem is often ill-posed, requiring regularization techniques.
• Tikhonov regularization is a common method but can introduce bias due to single constraints.
• Employing multiple constraints can improve results but faces challenges in parameter selection.

Purpose of the Study:

• To propose and evaluate a genetic algorithm (GA) based approach for estimating multiple regularization parameters.
• To enhance the performance of Tikhonov regularization in solving the inverse electrocardiography problem.

Main Methods:

• A genetic algorithm (GA) was developed to estimate multiple regularization parameters.
• The proposed method was applied using two and three constraints in the cost function.
• The effectiveness was demonstrated on the inverse electrocardiography problem.

Main Results:

• The GA-based approach successfully estimated multiple regularization parameters.
• The application of multiple constraints, optimized by GA, improved Tikhonov regularization solutions.
• The method showed improved accuracy compared to standard Tikhonov regularization.

Conclusions:

• Genetic algorithms provide an effective means to determine optimal multiple regularization parameters.
• The GA-based multiple constraint approach offers a significant improvement over standard Tikhonov regularization for inverse electrocardiography.
• This method addresses a key limitation in applying multiple constraints for ill-posed inverse problems.