The invention will now be described in more detail. The accompanying drawings, which are schematic illustrations, serve for a better understanding of the features of the present invention. Therein
FIG. 1 is a diagram illustrating the concept of volume-responsiveness by showing a typical graph of cardiac output over preload,
FIG. 2 is a typical plot of arterial pressure readings over time illustrating the effect of breathing on arterial pressure,
FIG. 3 shows a typical power spectrum based on readings of central venous pressure,
FIG. 4 shows a typical power spectrum based on readings of arterial pressure,
FIG. 5 shows the power spectra of heart activity and respiratory effects in double logarithmic scaling wherein the curves have been smoothed to illustrate the general course of the curve,
FIG. 6 shows a flow sheet of determining volume-responsiveness according to one embodiment of the present invention, wherein several options are illustrated using dashed lines,
FIG. 7 illustrates the general setup of an apparatus according to a first embodiment of the present invention,
FIG. 8 illustrates the general setup of an apparatus according to another embodiment of the present invention,
FIG. 9 shows a power spectrum in double logarithmic scaling, wherein the contributions of heart activity and respiratory have not been separated.
In the drawings, the same reference numerals have been used for corresponding features.
Though the embodiments described below focus on determining a parameter usable to characterize volume responsiveness, other physiologic parameters (such as cardiac output or tidal volume) may also (or alternatively) be determined. However, the apparatus and method according to the present invention are particularly advantageous for determining a parameter usable to characterize volume responsiveness. In order to illustrate the above-mentioned fact that an increase of blood volume (or preload) does not necessarily raise cardiac output, FIG. 1 shows a typical graph of cardiac output depending on preload (or blood volume). The graph varies from patient to patient (and depends on the individual patient's current condition). Depending on the local slope, additional volume may greatly increase (curve section marked with solid line) or not increase (curve section marked with dashed line) cardiac output. As the actual course of the curve schematically shown in FIG. 1 is not known beforehand for a specific patient in a specific condition, acquiring parameters helping to assess volume responsiveness can be crucial in clinical practice.
FIG. 7 shows the general setup of an apparatus according to a first embodiment of the present invention. An arterial catheter 1 is equipped with a pressure sensor for measuring arterial blood pressure. The pressure sensor of the catheter 1 is connected, via a pressure transducer 2, to an input channel 3 of a patient monitoring apparatus 4. Beside a proximal port 7 used to acquire the pressure signal, the catheter 1 may comprise one or more other proximal ports 8 to perform additional functions, such as blood temperature measurements or the like. The patient monitoring apparatus 4 is programmed to determine various hemodynamic parameters as described below, and to display the determined parameters (as numeric values, graphically or both) on the display 5. In addition, the determined parameters may be stored at a recording medium and/or printed. For this purpose, the patient monitoring apparatus 4 may comprise various interface ports for connecting peripheral equipment.
This first embodiment described requires a single arterial pressure sensor only. Though the sensor is shown to be invasive, a non-invasive pressure sensor may be implemented instead.
Various implementations of the invasive pressure sensor can be particularly advantageous. Pressure can either be transmitted hydraulically to a proximal catheter port and measured by an external sensor or may be measured directly on-site using a sensor installed at or near the catheter tip. Capacitative sensors, piezo sonsors or optical pressure sensors (e.g. based on Fabry-Perot interferometry) may be used.
As explained above, cardiac, vasculary and pulmonary volumes interact with each other in the patient 6. In particular, cardiac preload is affected by the volume occupied by respiration (either spontaneous or ventilated breathing). Due to recurrent respiratory cycles modulation of blood pressure and flow take place. FIG. 2 shows this modulation in a typical plot of arterial pressure readings over time. Such a modulation can also be observed for central venous pressure or stroke volume.
The patient monitoring apparatus 4 temporarily stores the blood pressure readings read in through the input channel 3 as a pressure curve p(t) over time. As heart rate and breathing cycle differ in frequency (f), the respiratory effect on the pressure curve can be separated from the heart activity. The patient monitoring apparatus 4 thus determines breathing cycle and heart rate from the pressure signal.
In particular, the patient monitoring apparatus 4 contains fast Fourier transformation means (FFT) 9 in order to perform a Fourier transformation on the stored pressure curve. The Fourier transform
P(┐)=∫p(t)·e−i┐tdt
of the blood pressure p(t) and the spectral density S(┐), i.e. the Fourier transform of the autocorrelation function, which is also determined by the FFT, are used to determine the contribution of each frequency
f=┐/2π for further evaluation.
A power spectrum, i.e. the spectral density S in dependence of the frequency f, of an arterial blood pressure measurement is shown in FIG. 4. From the spectral density of the pressure signal the components (visualized as “peaks”) caused by the respiratory rate (RR) and higher harmonics, i.e. 2·RR, 3·RR, . . . , can be separated in the power spectrum, as mentioned above. The components caused by the heart rate (HR) and higher harmonics, i.e. 2·HR, 3·HR, . . . , can also be separated in the power spectrum. Therefore, both the power spectrum SR(f) caused by respiration and the spectrum SC(f) caused by heart activity can be determined from the pressure measurements. For illustration, FIG. 5 shows the power spectrum of heart activity SC(f) and respiratory components SR(f) in dependence of the frequency f with logarithmic scaling on both axis. The solid line illustrates the behavior of spectral density of a ventilated patient 6 in comparison with a spontaneously breathing patient 6 (dashed line). For illustration purposes only, the curves have been smoothed. FIG. 9 shows, in comparison, a realistic power spectrum in double logarithmic scale, wherein the curve has not been smoothed. In FIG. 9, cardiac and respiratory contributions to the curve have not been separated.
In particular, the patient monitoring apparatus 4 determines in the spectral density of the pressure signal the magnitudes for the respiration rate and higher harmonics thereof, which leads to the respiratory power spectrum. Likewise, the cardiac power spectrum is determined from the amplitudes in the spectral density at the heart rate and higher harmonics thereof. A width at a predetermined fraction of the maximum of a respective peak (such as the full width at half maximum WR and WC, as depicted in FIG. 4) may or may not be used to separate harmonics and to determine the amount of each.
Integration of the spectral densities over the whole frequency range permits determination of a respiratory power corresponding to respiration and a cardiac power corresponding to heart activity. However, integration over only part of the frequency range will in many cases lead to sufficient approximations or even improve the quality of the results: While the integrals have to run over a suitable range, several frequencies may be suppressed to reduce or eliminate signal disturbances.
The thus determined respiratory and cardiac power values can now be used by the patient monitoring apparatus 4 to calculate the hemodynamic parameters of interest and display the determined parameters on the display 5.
For example, the ratio of respiratory and cardiac power is used as a parameter to assess volume responsiveness (FR) according to
FR=∫S
C(f)df/∫SR(f)df
and cardiac output may be calculated as the cardiac power value divided by an averaged blood pressure.
FIG. 8 shows the general setup of an apparatus according to the second embodiment, wherein two pressure sensors are used. In addition to the arterial pressure measured as described in connection with the above first embodiment, a central venous pressure (CVP) is measured using a pressure sensor in a central venous catheter 14. The pressure sensor of the central venous catheter 14 is connected, via a pressure transducer 10, to a second input channel 11 of the patient monitoring apparatus 4. Beside a proximal port 12 used to acquire the pressure signal, the catheter 14 may comprise one or more other proximal ports 13 to perform additional functions, such as blood temperature measurements, injections or the like. Instead of the central venus catheter 14 a pulmonary catheter (not shown) may be used to provide readings of a pulmonary pressure (PAP or PCWP). Generally, various measurement sites are suitable for providing first and second blood pressure readings.
Though one or both pressure sensors may also be non-invasive, as mentioned in connection with the first embodiment described above, best performance of the system can be achieved with two invasive pressure sensors, as depicted in FIG. 8.
As described above, various implementations of the invasive pressure sensors can be particularly advantageous. Pressure can either be transmitted hydraulically to a proximal catheter port and measured by an external sensor or may be measured directly on-site using a sensor installed at or near the catheter tip. Capacitative sensors, piezo sonsors or optical pressure sensors (e.g. based on Fabry-Perot interferometry) may be used.
The patient monitoring apparatus 4 determines the breathing cycle from the central venous pressure signal. Using the fast Fourier transformator (FFT) 9, the patient monitoring apparatus 4 determines the spectral density, which is shown in an exemplary manner in FIG. 3. In the spectral density derived from this pressure signal, the magnitudes are determined for the respiration rate and higher harmonics thereof, which leads to the respiratory power spectrum. A width at a predetermined fraction of the maximum of a respective peak (such as the full width at half maximum WR and WC, as depicted in FIG. 3) may or may not be used to separate harmonics and to determine the amount of each. Summation/integration over (at least part of) the respiratory power spectrum delivers the respiratory power.
The heart rate is determined from the arterial pressure signal. The cardiac power spectrum is determined from the amplitudes in the spectral density at the heart rate and higher harmonics thereof. A width at a predetermined fraction of the maximum of a respective peak (such as the full width at half maximum WR and WC, as depicted in FIG. 4) may or may not be used to separate harmonics and to determine the amount of each. Cardiac power is determined by summation/integration over (at least part of) the cardiac power spectrum.
Finally, the ratio of respiratory and cardiac power is provided as a measure of volume responsiveness as described above in connection with the first embodiment.
Further improvements of either of above-described embodiments include the following:
The power spectra can be characterized with respect to the kind of breathing (mechanically ventilated or spontaneous). In particular the full width at half maximum WR (see FIGS. 3 and 4) or other width at a predetermined fraction of the maximum of a respective peak, the slope of log SR(f) over log f for f approaching infinity (see FIG. 5), the slope of log Sc(f) over log f for f approaching infinity (see FIG. 5), the limits limf□0S(f), limf□0SR(f), and limf□0SC(f) are of special interest.
Controlled ventilation typically differs from spontaneous breathing by a smaller full width at half maximum WR and a slower decrease of log SR(f) with respect of log f at higher frequencies (see FIG. 5). As sensitivity of the approach according to the present invention may differ depending on spontaneous breathing and controlled ventilation, respectively, the volume responsiveness parameter FR can be corrected by a factor|:
FR=|(WR, aR, . . . )·∫SC(f)df/SR(f)df
wherein aR denotes the slope of log SR(f) over log f at higher frequencies.
Both, aR and aC(the slope of log SC(f) over log f at higher frequencies), can be used to extend the integrations ∫SC(f)df and ∫SR(f)df, respectively. Therefore, calculation of the ratios become more precise. Precision can also be improved by taking into account limits of spectra for f approaching 0. However, for most cases using the spectral density for f=0 is not recommended, as these values consider offsets in the pressure measurements only.
In addition, external input (e.g. by means of an interface to a ventilation machine or by user input determining the kind of current respiration) may be used to support determination of the correction factor |.
For a better understanding of the above, FIG. 6 shows an overview of method steps according to embodiments of the present invention adapted to determine a parameter for volume responsiveness as described above or similar. The method steps may be performed by the above described apparatus or otherwise. Dashed lines show an alternative embodiment including several options added to the basic embodiment (solid lines).
Blood pressure is measured (S1), optionally also at a second site (S1a). As described above, the respiration rate and the contribution of higher harmonics on respiration are determined (S2), as well as the heart rate and the contribution of higher harmonics on heart activity (S3). The respiratory and cardiac power spectra are determined (S4, S5). The ratio between both powers is calculated as the quotient of the integrals over the cardiac power spectrum (enumerator) and the respiratory power spectrum (denominator) (S6). The parameter usable to characterize volume responsiveness is determined (S7) and displayed or otherwise output. Determining this parameter may simply be acquiring the above ratio between cardiac and respiratory power or it may involve a correction factor (or a more complex correction function) using determination of the slopes of logarithmic spectral densities over logarithmic frequency (S6a, S6b), see also FIG. 5, and/or determination of the spectral density for frequency approaching zero (S6c, S6d). Further (or alternatively) correction may be achieved by external control (S7a).