Evaluation of Algorithms for ECG Derived Respiration in the Context of Heart Rate Variability Studies
Author
Sohrt-Petersen, Lasse
Term
4. term
Publication year
2014
Submitted on
2014-01-13
Pages
78
Abstract
Variation i hjertefrekvens (HRV) afspejler den autonome regulation, men tolkning af den respirationsrelaterede komponent kræver viden om vejrtrækningen. Da direkte måling kan være besværlig, undersøger denne afhandling, om respiration kan afledes fra EKG (ECG-derived respiration, EDR) til brug i HRV-studier. En række publicerede EDR-metoder blev implementeret og afprøvet på samtidige 5-minutters optagelser af EKG og respiratorisk luftstrøm fra tre raske mænd. For hver metode blev et respiratorisk surrogatsignal udledt fra EKG’et, åndedragsvarigheder blev identificeret i både surrogat- og referencesignalet, og serierne af varigheder blev sammenlignet ved korrelationsanalyse (Pearson). Én metode kunne ikke generere et anvendeligt respirationssignal, mens de øvrige metoder gav surrogater, der korrelerede signifikant med referencesignalet (p<0,001). Resultaterne indikerer, at flere EDR-tilgange kan levere brugbar respirationsinformation under de givne betingelser og motiverer yderligere undersøgelser af EDR i HRV-sammenhæng.
Heart rate variability (HRV) reflects autonomic regulation, but interpreting its respiration-related component requires information on breathing. Because direct respiratory measurements can be cumbersome, this thesis evaluates whether respiration can be derived from the ECG (ECG-derived respiration, EDR) for use in HRV studies. A selection of published EDR methods was implemented and tested on simultaneous 5-minute recordings of ECG and respiratory airflow from three healthy male subjects. For each method, a respiratory surrogate signal was extracted from the ECG, breath durations were identified in both the surrogate and the reference signals, and the resulting duration series were compared using Pearson correlation. One method failed to produce a feasible respiratory signal, whereas the remaining methods yielded surrogates that correlated significantly with the reference (p<0.001). These findings suggest that several EDR approaches can provide useful respiratory information under the given conditions and warrant further exploration of EDR in the context of HRV analysis.
[This summary has been generated with the help of AI directly from the project (PDF)]
Documents