Was versteht man unter Spektralanalyse von RR Intervallen?
RR Intervalle sind unterschiedlich lange Zeitintervalle, die rhythmischen
aber auch nicht rhythmischen Schwankungen unterworfen sind. Eine Spektralanalyse transformiert diese rhythmische Schwankungen in frequenzabhängige
Oszillationen.
Was gibt es für Probleme?
Die am meisten eingesetzte Spektralanaysetechnik ist die Fourieranalyse. Diese Analyse
setzt aber äquidistant abgetastetete Messwerte voraus. RR Intervalle sind nicht äquidistant und es sind auch keine Messwerte
im Sinn einer Signalanalyse. Zwischen zwei RR Intervallen gibt es keine weiteren Intervalle. Aus diesem Grund wurden und werden alternative
Analysealgorithmen entwickelt.
Die TRS Lösung dieses Problems
Jedes RR Intervall hat zwei Parameter, den Zeitpunkt des Intervallendes
und die Intervalllänge selbst. Beide Werte sind aber nicht voneinander unabhängig. TRS dreht ebenfalls wie die Fourieranalyse die
Intervalle am Ende um 90 Grad. Diese gedrehten Intervalle werden aber nicht verbunden, sondern mittels einer trigonometrischen
Funktion regressiv behandelt. Diese Regression erfolgt im Sinne einer maximalen Angleichung an die Intervalle, mit anderen Worten,
die Summe der Abweichungsquatrade muss ein Minimum werden (Logo oben links).
Eigenschaften von TRS
* Benutzung realer Zeitpunkte
und Intervalllängen, damit enrfällt die Problematik der
Interpolation (resampling).
* Gleitender Übergang an den Frequenzbandgrenzen.
* Kurze lokale Datensegmente von 15 Sek bis 60 Sek, die zeitlich verschoben werden
können.
* Stationäre oder nicht stationäre
Analysen sind möglich.
* Automatische Erzeugung von Frequenz-Zeit-Diagrammen.
* Oszillationen sind rein physiologischer Art. Damit ergibt
sich die Möglichkeit einer
* Berechnung der Baroreflex Sensitivität.
* Zwei simultane Blutdruckmessungen (z.B. Finometer und Colin)
sind möglich.
* Auswertung kompletter Untersuchungsserien.
* TRS ist eine rein statistische Spektralanalyse
* Fehlerhafte Werte können markiert werden
Gegenwärtig unterstützte Messsysteme:
1. SUEmpathy 100, SUESS Medizintechnik, Aue, Germany
2. Portapres, FMS, The
Netherlands.
3. Finometer, FMS, The Netherlands
4. Task Force Monitor, CNSystems,Medizintechnik AG, Österreich
Beispiele
von TRS Analysen
What does spectral analysis of RR intervals mean?
RR intervals are time differences of neighbouring R waves. The variability of these RR intervals occurs only in the scale itself as a result of rhythmical and non rhythmical oscillations. Using a spectral analysis, these time-varying RR intervals are transformed into frequency-dependent oscillations.
What are the challenges?
The fast Fourier
transformation is the most frequently used method for spectrally evaluation of RR intervals. However, this transformation requires
equidistant sampled measurements.
RR intervals are not equidistantly and they are not values in the sense of signal analysis.
In contrast to a signal which has an infinite number of in-between values, RR intervals do not
consist of such in-between values.
The
TRS solution
Each RR intervals consists of two parameters: distinct points in time characterizing the end of each interval and the
duration of the interval itself. Both parameters are interdependent. To construct a “function of time”, RR intervals are transformed
into a tachogram rotated 90° into the dt axis. This tachogram is used by both TRS and Fourier transformation. Unlike Fourier transformation
which interpolates RR intervals, TRS calculates oscillations by means of statistic regression. All oscillations are detected under
the condition that the variance (sum of squared differences) from the original RR intervals is a minimum. In other word, the sum of the errors has to be a minimum (compare the logo at the top).
Key Features of the TRS technique
TRS uses real points in time of
the end of each RR interval and real interval lengths. Hence,
interpolation is not required (resampling).
There is a smooth transition
by a Gaussian function between neighbouring frequency bands.
TRS uses short local data segments (from 15 s up to 60 s) which can be
shifted beat by beat for temporal resolution.
TRS enables stationary and/or non stationary analyses.
Frequency-time-diagrams are generated
automatically by shifts along the time axis.
TRS oscillations are of pure physiological origin which allows for determination of baroreflex
sensitivity.
Two simultaneously recorded blood pressure measurements can be analysed (e.g. using the
continuous blood pressure devices Finometer and Colin).
TRS supports the evaluation of complete test series.
Incorrect values can be marked
Currently supported measurement systems:
1. SUEmpathy 100, SUESS Medizintechnik, Aue, Germany2. Portapres, FMS, The Netherlands.
3. Finometer, FMS,
The Netherlands
4. Task Force Monitor, CNSystems,Medizintechnik AG, Österreich
Examples
of TRS analyses
The influence of respiration on heart rate, the so called sinus arrhythmia is illustrated in this figure. The time-dependent modulation of respiratory frequency is reflected as high frequency oscillations in RR interval spectrum (time-frequency variation)
Deep metronomic breathing at 0.1 Hz (six breaths per min)
This figure illustrates an example from the EuroBaVar study by Laude et al. (2004)
The patient presents with clinically diagnosed diabetic neuropathy, lying position. Age and sex of this patient are unknown.
In this figure, RR intervals are rotated by 90°, however, the figure is not to scale. The differences in points of time and the RR intervals are originally of equal lengths. This is however a problem of presentation. The generation of these RR intervals (0.15 Hz) was chosen to obtain similar proportions of high- and low frequency oscillations caused by the smooth transition between neighbouring frequency bands.
