GPGPU accelerated cardiac arrhythmia simulations.

Area of Science:

• Computational biology
• Cardiac electrophysiology
• High-performance computing

Background:

• Cardiac models are crucial for understanding arrhythmia mechanisms and guiding clinical therapy.
• Current computational models are complex, requiring efficient solutions for real-time application.
• Accelerating simulations is essential for clinical utility.

Purpose of the Study:

• To investigate the feasibility of using low-cost Graphics Processing Units (GPUs) to accelerate cardiac arrhythmia simulations.
• To evaluate the performance of GPU implementations compared to traditional Central Processing Unit (CPU) methods.

Main Methods:

• Porting a two-dimensional monodomain cardiac electrophysiology model to various GPU platforms.
• Simulating electrical activity under point stimulation and rotor activity conditions.
• Comparing simulation speeds between GPU and CPU implementations.

Main Results:

• GPU implementations achieved significant speedups: 18X for point stimulation and 12X for rotor activity.
• The number of concurrently launchable threads was identified as a critical optimization factor.
• Demonstrated the potential of GPGPU hardware for accelerating complex cardiac simulations.

Conclusions:

• Low-cost GPGPU hardware offers a viable solution for accelerating cardiac arrhythmia simulations.
• Optimized GPU implementations can significantly reduce computation time.
• This approach holds promise for enabling real-time clinical applications in arrhythmia management.