Name

Baroreflex_Analysis — Baroreflex analysis using multi channel non stationnary signal analysis method

Calling Sequence

results = Baroreflex_Analysis(RR,SBP,freq_sampling,frequency_band ,options)
results = Baroreflex_Analysis(RR,SBP,freq_sampling,frequency_band ,BPfrequencybounds, BPfilterlength ,options)

Parameters

RR

a real vector of length N: The vector of re-sampled (see ResampleRR) interbeat intervals of the R waves in the electrocardiogram (ms).

SBP

a real vector of length N: The vector of systolic blood pressure.

freq_sampling

a real positive scalar, the sampling frequency (in Hz) of the RR and SBPsignals.

frequency_band

a two column array with Nr rows,[fmin fmax] each row specifies a frequency range (in Hz) to be studied. See the ??? section for details.

BPfrequency_bounds

a real row vector of length 2: the band pass filter frequency bounds (in Hz) or an empty matrix. In the latter case BPfrequency_bounds is set to [0.9*min(fmin) max(fmax)*1.1].

BPfrequency_bounds is used to perform a wide range band pass filtering of the given RR and SBP signals.

BPfilterlength
a real scalar with integer value: the order of the FIR Hamming band_pass filter. The default value is 355. If BPfilterlength is set to 0 no wide range band pass filtering is done. See the ??? section for details.
options

a struct with fields:

  • <term>sectionlength</term>: an integer, the signal section length to be used for power and cross spectrum computation. The default value is 256.

  • <term>sectionstep</term>: an integer, the step to be applied for one signal section to the next one. sectionstep must be less or equal to sectionlength. Overlap is sectionlength-sectionstep. The default value is 128.

  • <term>smoothwindowlength</term>: an integer, the length of the smoothing window (hanning). the default value is 5.

  • <term>minimalcoherence</term>: a positive number in ]0 1]. Used to check if RR can be approximated as a linear filtering of SBP. The default value is 0.5.

result
a "RRB typed list with following fields:
  • <term>datatime</term>: a real row vector of length M, the time discretization of the given RR and SBPsignals.
  • <term>RR</term>: a real row vector, the given RR signal.
  • SBP: a real row vector, the given SBP signal.
  • <term>RR_energy</term>: a real M by Nr matrix. Each row contains the smoothed windowed energy of the RR signal in the corresponding frequency range.
  • <term>SBP_energy</term>: a real M by Nr matrix. Each row contains the smoothed windowed energy of the SBP signal in the corresponding frequency range.
  • <term>resulttime</term>: a real row vector of length M, the time discretization of the comptuded signals.
  • <term>RR_fmin</term>: a real M by Nr matrix. Each row contains the frequency around which the dispersion of the RR signal is minimal in the corresponding frequency range.
  • <term>SBP_fmin</term>: a real M by Nr matrix. Each column contains the frequency around which the dispersion of the SBP signal is minimal in the corresponding frequency range.
  • <term>RR_SPSDmin</term>: a real M by Nr matrix. Each column contains the value of the Smoothed Power Spectral Density (SPSD) at the frequency of minimal dispersion for the RR signal in the corresponding frequency range.
  • <term>SBP_SPSDmin</term>: a real M by Nr matrix. Each column contains the value of the Smoothed Power Spectral Density (SPSD) at the frequency of minimal dispersion for the SBP signal in the corresponding frequency range.
  • <term>RR_dispmin</term>: a real M by Nr matrix. Each column contains the value of the minimal dispersion for the RR signal in the corresponding frequency range. Dispersion value is zero when the signal is merely a sinusoide, only one components, whereas is 1 if the spectra is completely spread, so there isn't a principal component.
  • <term>SBP_dispmin</term>: a real M by Nr matrix. Each column contains the value of the minimal dispersion for the SBP signal in the corresponding frequency range. Dispersion value is zero when the signal is merely a sinusoide, only one components, whereas is 1 if the spectra is completely spread, so there isn't a principal component.
  • <term>SBP_RR_gain</term>: a real M by Nr matrix. Each column contains the SBP to RR transfer function gain in the corresponding frequency range.
  • <term>SBP_RR_coherence</term>: a real M by Nr matrix. Each column contains the coherence between SBP and RR signals in the corresponding frequency range.

Description

Baroreflex analysis using multi channel non stationnary signal analysis method

Caution

  • <term>frequency_band</term> This argument is used to reduce the signal band around the effective respiratory frequency. the effective respiratory frequency can be estimated using the Time frequency interactive tool.
  • <term>BPfilterlength</term> The value of this parameter has to establish a compromise between the stiffness of the filter and the sensibility to local variations of the signal.

Examples

Signals acquisition and filtering

    d=read(cardiovascularpath()+"demos/DATA/tilt_divers",-1,5)';//[RR SBP DBP Vt time] 
    freq_sampling=2;
    shift=600;
    RR= d(1,shift:$);
    SBP=d(2,shift:$);
    

Analysis

    clear options
    options.sectionlength=64;
    fbands=[0.04 0.15; //LF
            0.2 0.3];  //HF
    result=Baroreflex_Analysis(RR,SBP,freq_sampling,fbands,options)
    

Drawing

    PlotBaroreflexAnalysis(result,"Tilt")
    

Authors

Alexandro Monti, INRIA
Serge Steer, INRIA

Bibliography

Rapport de recherche INRIA RR-4427:"Short term control of the cardivascular system: modelling and signal analysis" Alessandro Monti, Claire Médigue, Michel Sorine,2002 pp 50-

Used Functions

np_mc_nss_analysis