Feasibility of time delay stability analysis of physiological signals acquired during short-time 3T magnetic...

Time delay stability analysis is an established analysis method to quantify interactions ("coupling") between organ systems and has been applied successfully to physiological time series acquired in various fields of research. In the vast majority of literature, this method was applied i) to long-time measurements, e.g. acquired during sleep, and ii) to data with rather high signal-to-noise ratios. We are interested in the potential of applying this analysis to short time series with low signal-to-noise ratios, i.e. acquired during real-time magnetic resonance imaging. Hence, in this work we conduct a feasibility study by computing time delay stability on short 60 s segments and in the presence of gradient-induced noise in physiological time series. To evaluate the feasibility in this setting, we compare two study groups that should show clear differences. One group contains older atrial fibrillation patients (N=10) and the other consists of young healthy controls (N=10), as it is known that in atrial fibrillation coupling between the cardiac and respiratory organ systems is reduced. We measure electrocardiography, pulse oximetry, and breathing belt data in both groups during 3 Tesla brain magnetic resonance imaging. Results show that in 5 of the 6 modality combinations, coupling is reduced. A strong decrease of 14.4% becomes visible in the coupling between pulse and heart rates. Although further studies with larger cohorts are needed, we conclude that our results show the potential of using time delay stability analysis in short time series acquired during magnetic resonance imaging. This could enable coupling analysis between physiological signals and imaging data.