Abstract
In studies with near-infrared spectroscopy, the recorded signals contain information on the temporal interbeat intervals of the heart. If this cardiac information is needed exclusively and could directly be extracted, an additional electrocardiography device would be unnecessary. The aim was to estimate these intervals from signals measured with near-infrared spectroscopy with two novel approaches. In one approach, we model the heartbeat oscillations in these signals with a Fourier series where the coefficients and the fundamental frequency can continuously change over time. The time-dependent model parameters are estimated and used to calculate the interbeat intervals. The second approach uses empirical mode decomposition. The signal measured with near-infrared spectroscopy is empirically decomposed into a set of oscillatory components. The sum of a specific subset of them is an estimate of the pure heartbeat signal in which the diastolic peaks and consequential interbeat intervals are detected. We show in simultaneous electrocardiography and near-infrared spectroscopy measurements on 11 subjects (8 men and 3 woman with mean age 32.8 ± 8.1 yr), that the interbeat intervals (and the consequential pulse rate variability measures), estimated using the proposed approaches, are in high agreement with their correspondents from electrocardiography.
Trajkovic, I