Investigation and modeling of respiratory effects on SCG-signal fiducial points
Author
Hansen, Rolf Oberlin
Term
4. term
Publication year
2019
Submitted on
2019-06-06
Pages
57
Abstract
Seismokardiografi (SCG) er et billigere alternativ til etablerede metoder til registrering af hjertets mekaniske hændelser, men signalets morfologi er endnu ikke fuldt forstået, bl.a. fordi respiration påvirker både timing af hændelser og slagvolumen. Dette speciale undersøger og modellerer, hvordan respiration påvirker amplituder og tidspunkter for SCG‑signalets fiducial‑punkter omkring den første hjertelyd. I stedet for at opdele respiration i to kategorier foreslås en kontinuerlig beskrivelse via respirationsfase. Respiration blev afledt direkte fra SCG og modelleret som en top‑til‑top sinuskurve, der rummer information om både respirationsamplitude og -gradient. Ved at anvende andengradspolynomiel regression til at modellere fiducial‑amplituder som funktion af fase hos 10 forsøgspersoner fremkom gennemgående lave eller ubetydelige justerede r^2‑værdier på grund af høj varians. Der blev dog observeret tendenser i både amplituder og tidsintervaller mellem fiducial‑punkter, men ved forskellige faser, hvilket tyder på en faseforsinkelse mellem respiration og det registrerede SCG‑signal. En sinusformet, kontinuerlig repræsentation af respiration anbefales til fremtidig forskning, da simple todelte klassifikationer ikke synes egnede til at kortlægge SCG‑morfologien.
Seismocardiography (SCG) offers a lower-cost alternative to established methods for recording mechanical cardiac events, but its signal morphology remains insufficiently understood, partly because respiration alters event timing and stroke volume. This thesis investigates and models how respiration affects the amplitudes and timings of SCG fiducial points around the first heart sound. Instead of classifying respiration into two categories, a continuous phase representation is proposed. Respiration was derived directly from SCG and modeled as a peak-to-peak sine wave that captures both respiratory amplitude and gradient. Using second-degree polynomial regression to model fiducial amplitudes as a function of phase in 10 subjects yielded consistently low or negligible adjusted r^2 values due to high variance. Nonetheless, tendencies were observed in both amplitudes and time intervals between fiducial points, but at different phases, suggesting a phase delay between respiration and the recorded SCG. A sinusoidal, continuous representation of respiration is recommended for future research, as simple two-class schemes appear inadequate for mapping SCG morphology.
[This summary has been generated with the help of AI directly from the project (PDF)]
Keywords
Documents