Fast interactive simulations of cardiac electrical activity in anatomically accurate heart structures by compressing...

Area of Science:

• Computational Biology
• Biophysics
• Medical Imaging

Background:

• Numerical simulations are crucial for understanding cardiac arrhythmias and guiding patient-specific treatments.
• Current cardiac simulation methods are computationally expensive and challenging to implement.
• Graphics Processing Units (GPUs) offer potential for faster cardiac simulations.

Purpose of the Study:

• To enhance the performance and accuracy of GPU-based cardiac simulations.
• To conserve computational resources and memory during simulations.
• To enable interactive and efficient simulation of cardiac electrical activity on realistic anatomies.

Main Methods:

• Developed a compression algorithm to manage sparsity in Cartesian grids derived from MRI/mCT scans.
• Implemented a discretization scheme incorporating cross-diagonal terms for improved numerical accuracy.
• Utilized Graphics Processing Units (GPUs) for parallel computation.

Main Results:

• Achieved an order of magnitude reduction in memory demand (up to 10x).
• Increased calculation speed by up to 20x, enabling interactive simulations.
• Demonstrated superior performance in simulating slender cardiac tissues.
• Validated simulation results against experiments on explanted human hearts.

Conclusions:

• The developed compression algorithm accelerates cardiac electrical activity simulations on realistic anatomies.
• Improved accuracy, especially in thin tissues, is achieved through the new discretization scheme.
• The method facilitates interactive simulations and paves the way for virtual cohorts and digital twins for personalized therapies.