Hermite Polynomial Characterization of Heartbeats with Graphics Processing Units

In this paper we address the massive parallelization of the characterization of heartbeats by means of Graphics Processors. Heartbeats are represented with Hermite polynomials due to the compactness and robustness of this representation. Both the off-line and on-line characterization of QRS complexes are covered, thus, assessing the capabilities of Graphics Processors for these tasks. The results yield that off-line processing with a GPU can be computed 200x faster than a standard CPU, while on-line processing can be 100x faster

keywords: Hermite functions, ECG, QRS, Arrhythmia, GPU, CUDA, Parallelization, Heartbeat Representation