1. Patent Search
  2. Details

PROCESS AND DEVICE FOR DETERMINING A RESPIRATORY AND/OR CARDIOGENIC SIGNAL

20220330837 · 2022-10-20

    Inventors

    Cpc classification

    International classification

    Abstract

    A process and signal processing unit (5) determine a cardiogenic signal (Sig.sub.kar,est) or a respiratory signal (Sig.sub.res,est) from a sum signal (Sig.sub.Sum), resulting from a superimposition of cardiac activity and breathing of a patient (P). A signal estimating unit (6), which yields a shape parameter as a value of a transmission channel parameter (LF), is generated during a training phase. A sample with a sample element per heartbeat is used. During a use phase, the transmission channel parameter is measured for each heartbeat, a shape parameter value is calculated by the application of the signal estimating unit and is used to calculate an estimated cardiogenic signal segment (Sig.sub.Hz,kar.LF) or an estimated respiratory signal segment. The cardiogenic signal segments are combined into the cardiogenic signal or the respiratory signal segments are combined into the respiratory signal or the cardiogenic signal segments are subtracted from the sum signal.

    Claims

    1. A computer-implemented process for calculating an estimate for a cardiogenic signal and/or a respiratory signal with the use of a signal processing unit, wherein the cardiogenic signal is an indicator for a cardiac activity of a patient and the respiratory signal is an indicator for a for an intrinsic spontaneous breathing and/or a mechanical ventilation of the patient, wherein the process comprises a training phase and a subsequent use phase, the process further comprising the steps of: receiving and processing, with the signal processing unit, at least during the training phase measured values from a sum signal sensor device, which sensor device measures a signal generated in the body of the patient; generating, with the signal processing unit at least in the training phase, depending on a time course of measured values of the sum signal sensor device, a sum signal, which comprises a superimposition of the cardiac activity and the intrinsic spontaneous breathing and/or mechanical ventilation of the patient; detecting during the training phase, with the signal processing unit, a plurality of heartbeats, which the patient performs during the training phase; generating during the training phase, with the signal processing unit, a sample with a plurality of sample elements, wherein each sample element pertains to a respective detected heartbeat, wherein the generation of a sample element for a heartbeat comprises the steps of: determining, with the signal processing unit, a sum signal segment of the sum signal, which sum signal segment pertains to the heartbeat; determining, with the signal processing unit, for at least one shape parameter a shape parameter value which the shape parameter assumes during this the heartbeat by analysis of the sum signal segment, wherein the shape parameter influences a time course of the cardiogenic signal and/or of the respiratory signal; receiving, with the signal processing unit, at least one value of a predefined first transmission channel parameter, which value has been measured during the heartbeat by an additional sensor, or calculating, with the signal processing unit, the value of the first transmission channel parameter by an analysis of the sum signal, wherein the first transmission channel parameter correlates with an effect of an anthropological variable on a transmission channel from a signal source in the body of the patient to the sum signal sensor device; and generating, with the signal processing unit, the sample element for the heartbeat such that the sample element comprises the shape parameter value calculated for the heartbeat and the value of the first transmission channel parameter measured or calculated for the heartbeat; generating, with the signal processing unit during the training phase, with the use of the sample, a signal estimating unit, which unit yields the shape parameter as a function of the first transmission channel parameter; detecting, with the signal processing unit, during the use phase at least one heartbeat which the patient performs during the use phase; during the use phase carrying out the following steps for at least one detected heartbeat: detecting, with the signal processing unit, a characteristic heartbeat time or a heartbeat time period of the heartbeat; receiving a value of the first transmission channel parameter, which value has been measured at the heartbeat, from the additional sensor or generating the sum signal also during the use phase depending on measured values of the sum signal sensor device and calculating a value of the first transmission channel parameter for the heartbeat by an analysis of the sum signal; calculating, a shape parameter value for the shape parameter by applying the signal estimating unit to the value of the first transmission channel parameter measured or calculated for the heartbeat; and calculating, with the use of the calculated shape parameter value, an estimated cardiogenic signal segment and/or an estimated respiratory signal segment for the heartbeat, which segment approximately describes the cardiogenic signal or the respiratory signal, respectively, during the heartbeat; carrying out, with the signal processing unit, at least one of the following three steps during the use phase with the use of the characteristic heartbeat time or the heartbeat time period detected during the use phase; combining the calculated estimated cardiogenic signal segments for the detected heartbeats to the estimated cardiogenic signal; or combining the calculated estimated respiratory signal segments for the detected heartbeats to the estimated respiratory signal or determining the estimated respiratory signal by compensating the cardiac activity by calculation, wherein the step of determining during the use phase the estimated respiratory signal by compensation comprises the following steps with the signal processing unit: generating a sum signal during the use phase depending on measured values of the sum signal sensor device and compensating an influence of the heartbeat on the sum signal generated during the use phase, the compensation is performed by using the estimated cardiogenic signal segment for the heartbeat.

    2. A process in accordance with claim 1, wherein the step of generating the signal estimating unit with the use of the sample comprises the steps performed by the signal processing unit of splitting up the sample elements on the basis of the respective values of the first transmission channel parameter into sample element classes such that the values of the first transmission channel parameter of the sample elements of one sample element class differ from one another by at most one predefined absolute or percentage limit, and calculating for each sample element class a respective reference value range for the first transmission channel parameter and an associated reference signal segment, wherein the signal processing unit combines the sum signal segments of the sample element class to the reference signal segment and wherein the associated reference signal segment acts as the shape parameter, and wherein the signal processing unit generates the signal estimating unit such that the signal estimating unit comprises a library with a plurality of reference signal segments, wherein each reference signal segment is assigned to a reference value range of the first transmission channel parameter, and wherein the step of applying in the use phase the signal estimating unit to a value of the first transmission channel parameter comprises the steps performed by the signal processing unit of determining, depending on the received value of the first transmission channel parameter, at least one reference value range of the first transmission channel parameter and the respective associated reference signal segment and calculating the estimated signal segment depending on the determined reference signal segment.

    3. A process in accordance with claim 2, wherein the signal processing unit calculates during the training phase for each sample element class a respective reference value of the first transmission channel parameter by using the values of the first transmission channel parameter belonging to the sample element class and uses the respective reference value of the first transmission channel parameter as the value range of the first reference transmission channel parameter of the sample element class and wherein the step of applying in the use phase the signal estimating unit to the value of the first transmission channel parameter measured at a heartbeat comprises the steps performed by the signal processing unit of determining in the library a first and a second reference signal segment, which are associated with a first and a second reference value of the first transmission channel parameter as the respective value range of the first transmission channel parameter of the first and second reference signal segment, wherein the first reference transmission channel parameter value is lower than or equal to and the second reference transmission channel parameter value is greater than or equal to the value of the first transmission channel parameter measured at the heartbeat, and calculating the signal segment estimated for the heartbeat by a smoothing between the first determined reference signal segment and the second determined reference signal segment.

    4. A process in accordance with claim 2, wherein at least one reference time course of the sum signal during a heartbeat is predefined or is calculated by the signal processing unit during the training phase and wherein in the step of receiving or calculating for a heartbeat a value of the first transmission channel parameter, the signal processing unit determines the sum signal segment belonging to the heartbeat, calculates a respective agreement value between this sum signal segment and the reference time course and calculates the value of the first transmission channel parameter for the heartbeat with the use of the calculated agreement value, wherein each class of sample elements, which the signal processing unit generates during the use phase, comprises as the reference value of the first transmission channel parameter value range a respective value range of possible agreement values.

    5. A process in accordance with claim 4, wherein the signal processing unit calculates during the training phase the reference time courses with the use of the sum signal segments determined in the training phase by applying a singular value decomposition or a principal component analysis to predefined standardized sum signal segments.

    6. A process in accordance with claim 2, wherein the step of the signal processing unit combining, for a sample element class, the sum signal segments of the class to the reference signal segment comprises the steps performed by the signal processing unit of superimposing the sum signal segments of the sample element class by calculation, so that each sum signal segment pertains to the same sequence of relative sampling time points, generating, for each relative sampling time point by applying a smoothing procedure, a respective fitting curve, which assigns a respective reference signal value to each value range of the first transmission channel parameter belonging to a sample element class, and determining for each value range of the first transmission channel parameter a sequence of the fitting curve values along the relative sampling time points and using the sequence as the reference signal segment for the value range of the first transmission channel parameter.

    7. A process in accordance with claim 1, wherein the step of compensating, by calculation, the influence of the heartbeat on the sum signal during the determination of the estimated respiratory signal comprises the steps performed by the signal processing unit of determining a heartbeat time period of the heartbeat and compensating, by calculation, and by the use of the cardiogenic signal segment estimated for the heartbeat, the influence of the heartbeat on the segment of the sum signal that pertains to the heartbeat time period.

    8. A process in accordance with claim 1, wherein the generation of the sample element for a heartbeat comprises the additional steps that the signal processing unit receives a value of at least one additional predefined transmission channel parameter, which value was measured in the course of the heartbeat and wherein the additional transmission channel parameter correlates with an effect of the anthropological variable or of another anthropological variable on the first transmission channel or on an additional transmission channel guiding to the sum signal sensor device and generates the sample element for the heartbeat such that the sample element additionally comprises the value of the additional transmission channel parameter, which value was measured in the course of the heartbeat, the signal processing unit generates the signal estimating unit such that the signal estimating unit yields for a heartbeat the shape parameter as a function of the first transmission channel parameter and of the additional transmission channel parameter, and wherein the signal processing unit carries out in the use phase for at least one detected heartbeat the additional steps that the signal processing unit receives the respective measured value from the additional sensor or calculates same by an analysis of the sum signal, which measured or calculated values the first transmission channel parameter and the additional transmission channel parameter, respectively, assume at the heartbeat, and calculates a respective value for the shape parameter by applying the signal estimating unit to the respective value of the first and the additional transmission channel parameter measured or calculated at the heartbeat.

    9. A process in accordance with claim 8, wherein the first transmission channel parameter is correlated with a filling level of the lungs of the patient and the additional transmission channel parameter is correlated with a phase during an individual breathing operation and/or ventilating operation.

    10. A process in accordance with claim 1, wherein the first transmission channel parameter depends on a geometry of the body of the patient and the signal processing unit receives and processes both in the training phase and in the use phase a plurality of measured values, which values have been measured by a body geometry sensor, wherein the measured values of the body geometry sensor correlate with the body geometry of the patient, which is current.

    11. A process in accordance with claim 10, wherein the first transmission channel parameter, is a current breathing state and/or ventilating state of the patient and the body geometry sensor comprises a breathing state sensor, which measures the current breathing state and/or ventilating state of the patient.

    12. A process in accordance with claim 11, wherein the breathing state sensor measures at least one of a flow of gas into the body and/or out of the body of the patient, an airway pressure of the patient, a flow of gas out of a mechanical ventilator or into a ventilator, wherein the ventilator is in a fluid connection with the patient and a current position, speed and/or acceleration of at least one reference point on the skin of the patient.

    13. A process in accordance with claim 11, wherein the sum signal sensor device comprises at least one sum signal sensor positioned on the skin of the patient, wherein the signal processing unit receives measured values for a current position relative to a reference point of the sum signal sensor positioned on the skin, wherein the position sensor measures the relative position of the sum signal sensor during both the training phase and the use phase, wherein during the training phase, the signal processing unit generates a functional relationship by means of measured values of the breathing state sensor and of measured values of the position sensor, which functional relationship describes the relative position of the sum signal sensor positioned on the skin as a function of the breathing state and/or ventilating state of the patient, and generates the signal estimating unit such that the signal estimating unit yields for a heartbeat the shape parameter as a function of the measured relative position of the sum signal sensor positioned on the skin, and wherein during the use phase, for at least one detected heartbeat, the signal processing unit receives measured values, which correlate with the current breathing state and/or ventilating state of the patient during the heartbeat, calculates the current relative position of the sum signal sensor by applying the functional relationship to the measured current breathing state and/or ventilating state and calculates the estimated signal segment for the heartbeat by applying the signal estimating unit to the calculated relative position.

    14. A process in accordance with claim 1, wherein the signal processing unit measures the value of the first transmission channel parameter or of an additional transmission channel parameter by the signal processing unit analyzing the received sum signal, which value of the additional transmission channel parameter is measured in the course of the heartbeat and wherein the additional transmission channel parameter correlates with an effect of the anthropological variable or of another anthropological variable on the transmission channel or on an additional transmission channel guiding to the sum signal sensor device.

    15. A process in accordance with claim 14, wherein the transmission channel parameter or the additional transmission channel parameter is measured by analysis of the sum signal and is an interval between two characteristic times of two consecutive heartbeats or an interval between two signal peaks in a course of a single heartbeat or an difference between a highest value and a lowest value of the sum signal in the course of a single heartbeat.

    16. A process in accordance with claim 1, wherein a standard reference signal segment, which is caused by the cardiac activity in a course of a heartbeat, is predefined, wherein this standard reference signal segment depends on the shape parameter, wherein the generation of the sample element for a heartbeat comprises the steps of the signal processing unit calculating a value for the shape parameter of the standard reference signal segment by analyzing the sum signal segment belonging to the heartbeat and generating the sample element for the heartbeat such that the sample element comprises the value of the shape parameter, which value is calculated for the heartbeat, wherein the signal processing unit generates the signal estimating unit such that the signal estimating unit yields the shape parameter of the standard reference signal segment as a function of the first transmission channel parameter, and wherein the step of calculating in the use phase the estimated signal segment for a detected heartbeat comprises the steps performed by the signal processing unit of calculating the value of the shape parameter of the standard reference segment by applying the signal estimating unit to the value of the first transmission channel parameter, which value was measured at a detected heartbeat, adapting the predefined standard reference signal segment with the use of the calculated value of the shape parameter, and calculating the estimated signal segment for the heartbeat depending on the adapted standard reference signal segment.

    17. A process in accordance with claim 1, wherein the sum signal sensor device comprises at least one sum signal sensor located at a distance from the heart and at least one sum signal sensor located closer to the heart, wherein both sum signal sensors measure a signal generated in the body of the patient, wherein the sum signal sensor located at a distance from the heart has a greater distance from a heart muscle of the patient than the sum signal sensor located closer to the heart, and wherein the signal processing unit generates the sum signal during the training phase by using measured values of the sum signal sensor located at a distance from the heart and detects during the use phase the heartbeat and the characteristic time thereof and/or the heartbeat time period of the heartbeat with measured values of the sum signal sensor located closer to the heart.

    18. A process in accordance with claim 17, wherein the sum signal sensor located at a distance from the heart has a shorter distance from a muscle of breathing muscles of the patient than the sum signal sensor located closer to the heart, wherein during the use phase, the signal processing unit generates the sum signal with the use of measured values of the sum signal sensor located at a distance from the heart without using the measured values of the sum signal sensor located closer to the heart.

    19. A process in accordance with claim 1, wherein the sensor signal device comprises at least one first sum signal sensor and at least one second sum signal sensor; wherein at least during the training phase, the signal processing unit receives measured values from the first sum signal sensor, which first sum signal sensor measures the signal generated in the body of the patient at a first position, and measured values from the second sum signal sensor, which second sum signal sensor measures the signal generated in the body of the patient at a second position different from the first position, and wherein the process comprises the additional steps that during the training phase, the signal processing unit generates a first sum signal depending on measured values of the first sum signal sensor and generates a second sum signal depending one measured values of the second sum signal sensor generates a first sample with the use of the first sum signal and a second sample with the use of the second sum signal and generates a first signal estimating unit with the use of the first sample and a second signal estimating unit with the use of the second sample and wherein during the use phase, for at least one detected heartbeat, the signal processing unit generates a first estimated signal segment for the heartbeat by applying the first signal estimating unit and a second estimated signal segment for the heartbeat by applying the second signal estimating unit and combines the first and second estimated signal segments into an estimated signal segment for the heartbeat.

    20. A process in accordance with claim 19, wherein the generation of the sample element for the detected heartbeat comprises the following steps: the signal processing unit receives measured values from a first parameter sensor, which first parameter sensor measures a first value of the first transmission channel parameter, and receives measured values from a second parameter sensor, which second parameter sensor measures a second value of the first transmission channel parameter or of an additional transmission channel parameter, during the training phase, the signal processing unit generates the first signal estimating unit such that the first signal estimating unit yields the shape parameter as a function of the first transmission channel parameter measured by the first parameter sensor, and generates the second signal estimating unit such that the second signal estimating unit yields the shape parameter as a function of the transmission channel parameter measured by the second parameter sensor, and in the use phase, for the detected heartbeat, the signal processing unit receives a first parameter value, which first parameter value was measured by the first parameter sensor during the heartbeat, receives a second parameter value, which second parameter value was measured by the second parameter sensor during the heartbeat, generates the first estimated signal segment for the heartbeat by applying the first signal estimating unit to the first parameter value and generates the second estimated signal segment for the heartbeat by applying the second signal estimating unit to the second parameter value.

    21. A process in accordance with claim 1, wherein during the training phase, the signal processing unit carries out the steps of generating the sum signal in the time range, transforming for each heartbeat the segment of the sum signal, belonging to the heartbeat, into a sum signal in the frequency range, determining the shape parameter value by an analysis of the sum signal segment transformed into the frequency range, generating each sample element for a heartbeat such that the sample element comprises the shape parameter value determined in the frequency range and the value of the first transmission channel parameter value measured during the heartbeat, and generating the signal estimating unit such that the signal estimating unit describes in the frequency range the shape parameter as a function of the first transmission channel parameter, wherein during the use phase, the signal processing unit carries out for at least one detected heartbeat the steps of calculating an estimated signal segment in the frequency range by applying the signal estimating unit and transforming the estimated signal segment into an estimated signal segment in the time range.

    22. A process in accordance with claim 1, wherein at least one first frequency range is predefined and the process comprises the additional steps of the signal processing unit generating from the measured values of the sum signal sensor device and overall sum signal, determining in the overall sum signal a respective signal component that is in the first frequency range, and determining in the signal component, which is in the first frequency range, a respiratory signal component and/or a cardiogenic signal component and wherein the signal processing unit furthermore takes the action of determining the estimated respiratory signal with the use of the respiratory signal component located in the first frequency range and/or determining the estimated cardiogenic signal with the use of the cardiogenic signal component in the first frequency range.

    23. A process in accordance with claim 22, wherein at least one second frequency range is predefined such that the signal component of the overall sum signal, which signal component is located in the second frequency range, is effected exclusively by the intrinsic spontaneous breathing and/or mechanical ventilation or exclusively by the cardiac activity of the patient wherein the signal processing unit determines the estimated respiratory signal with the use of the respiratory signal component in the first frequency range and of the signal component of the overall sum signal, which said signal component of the overall sum signal is located in the second frequency range and is effected by the breathing/ventilation or determines the estimated cardiogenic signal with the use of the cardiogenic signal component in the first frequency range and the, signal component of the overall sum signal, which said signal component of the overall sum signal is located in the second frequency range, and is effected by the cardiac activity.

    24. A process in accordance with claim 1, wherein a change rule is predefined, which is applicable to a segment of the sum signal belonging to one heartbeat, wherein the predefined change rule depends on the shape parameter and wherein during the use phase the signal processing unit calculates a value for the shape parameter in the predefined change rule by applying the signal estimating unit and in the step of calculating an estimated signal segment for the detected heartbeat, the signal processing unit applies the change rule, which is parameterized with the calculated shape parameter value, to the segment of the sum signal, which segment belongs to the heartbeat, and the signal processing unit calculates the estimated signal segment by applying the parameterized change rule to the sum signal segment.

    25. A process in accordance with claim 1, wherein a subdivision of the heartbeat time period into at least two different heartbeat time period phases, which subdivision is valid for each heartbeat, is predefined, wherein during the training phase and during the use phase the signal processing unit receives a respective value for the first transmission channel parameter and for each heartbeat time period phase, which value was received from the additional sensor during the heartbeat time period phase, wherein during the training phase the signal processing unit generates for each detected heartbeat and for each heartbeat time period phase of the detected heartbeat a respective sample element such that the respective sample element comprises the shape parameter value calculated for the heartbeat time period phase and the value of the first transmission channel parameter, which channel parameter value was measured during the heartbeat time period phase, and wherein during the use phase, for the detected heartbeat, the signal processing unit calculates for each heartbeat time period phase of the detected heartbeat a respective value for the shape parameter by applying the signal estimating unit to the value of the first transmission channel parameter measured during the heartbeat time period phase and calculates the estimated signal segment for the heartbeat with the use of the shape parameter values for the heartbeat time period phases of the detected heartbeat.

    26. A process in accordance with claim 25, wherein during the training phase, the signal processing unit generates for each heartbeat time period phase, with the use of the sample elements generated for the heartbeat time period phase, a respective signal phase estimating unit, which yields the shape parameter as a function of the first transmission channel parameter and is valid for the heartbeat time period phase, and wherein during the use phase, for the detected heartbeat, the signal processing unit calculates for each heartbeat time period phase of the heartbeat a respective estimated signal segment, which approximately describes the cardiogenic signal or the respiratory signal in the course of the heartbeat time period phase of the heartbeat, and calculates the estimated signal segment for the heartbeat with the use of the estimated signal segments for the heartbeat time period phases of the heartbeat and uses the signal phase estimating unit for the heartbeat time period phase for this calculation.

    27. A process in accordance with claim 1, wherein during the use phase, the signal processing unit generates at least one additional sample element which additional sample element relates to a heartbeat detected during the use phase, and wherein the signal estimating unit generated during the training phase is modified or generated again with the use of the additional sample element generated during the use phase.

    28. A process in accordance with claim 1, wherein the patient is ventilated by means of a mechanical ventilator, which carries out ventilation strokes, wherein the ventilation strokes are triggered depending on the determined estimated respiratory signal.

    29. A process in accordance with claim 1, wherein the patient is ventilated by means of a mechanical ventilator and wherein the signal processing unit performs the additional steps in the use phase of receiving a measured ventilator signal, which describes the flow of gas effected by the ventilator between the ventilator and the patient, comparing the ventilator signal with the estimated respiratory signal depending on the comparison result, calculating an assessment of the synchronization between the breathing activity of the patient and the gas flow effected by the ventilator and if this assessment of the synchronization is below a predefined threshold, causing an operating parameter of the ventilator to change automatically and/or an alarm to be outputted.

    30. A signal processing unit for calculating an estimate for a cardiogenic signal and/or a respiratory signal, wherein the cardiogenic signal is an indicator for a cardiac activity of a patient and the respiratory signal is an indicator for an intrinsic spontaneous breathing and/or a mechanical ventilation of the patient, wherein the signal processing unit is configured for carrying out a training phase and a subsequent use phase, wherein the signal processing unit is configured to receive, at least during the training phase, measured values from a sum sensor device, which sum sensor device is configured to measure a signal generated in the body of the patient, and the signal processing unit is configured to process these measured values, wherein the signal processing unit is configured to generate, at least during the training phase, depending on the time course of measured values of the sum signal sensor device, a sum signal, which comprises a superimposition of the cardiac activity and of the intrinsic spontaneous breathing and/or mechanical ventilation of the patient, wherein the signal processing unit is configured to detect during the training phase a plurality of heartbeats, which the patient performs during the training phase, and generate during the training phase a sample with a plurality of sample elements, wherein each sample element pertains to one respective detected heartbeat, wherein the signal processing unit is configured to carry out the following steps during the generation of a sample element for a heartbeat: determine a segment of the sum signal, which sum signal segment belongs to the heartbeat, determine by analyzing the sum signal segment for at least one shape parameter a shape parameter value that the shape parameter assumes at the heartbeat, wherein the shape parameter influences a time course of the cardiogenic signal and/or of the respiratory signal, receive at least one value of a predefined first transmission channel parameter, which value has been measured at the heartbeat by an additional sensor or generate a sum signal during the use phase depending on measured values of the sum signal sensor device and to calculate a value of the first transmission channel parameter by analysis of the sum signal, wherein the first transmission channel parameter correlates with an effect of an anthropological variable on a transmission channel from a signal source in the body of the patient to the sum signal sensor device, and generate the sample element for the heartbeat such that the sample element comprises the shape parameter value and the value of the first transmission channel parameter, which values were measured or calculated at the heartbeat, wherein the signal processing unit is configured to generate with the use of the sample a signal estimating unit, which unit yields the shape parameter as a function of the first transmission channel parameter, wherein the signal processing unit is configured to detect during the use phase at least one heartbeat, which the patient performs during the use phase, wherein the signal processing unit is configured to carry out the following steps for at least one heartbeat detected during the use phase: detecting a characteristic heartbeat time or a heartbeat time period of the heartbeat, receiving a value of the first transmission channel parameter, which value was measured during the heartbeat by the additional sensor, or calculating such a value by an analysis of the sum signal, calculating a value for the shape parameter by applying the signal estimating unit to the value of the first transmission channel parameter, the value being measured or calculated for the heartbeat, and calculating with the use of the calculated shape parameter value an estimated cardiogenic signal segment and/or an estimated respiratory signal segment for the heartbeat, which signal segment describes the cardiogenic signal or the respiratory signal, respectively, in the course of the, wherein the signal processing unit is configured to carry out during the use phase at least one of the following three steps, with the use of the characteristic heartbeat time or heartbeat time period detected during the use phase: combining the calculated estimated cardiogenic signal segments for the detected heartbeats to the estimated cardiogenic signal or combining the calculated estimated respiratory signal segments for the detected heartbeats to the estimated respiratory signal or determining the estimated respiratory signal by compensating the cardiac activity, wherein the signal processing unit is configured to carry out the following steps when determining the estimated respiratory signal by compensating: generating a sum signal depending on measured values of the sum signal sensor device during the use phase as well and compensating by calculation the influence of the heartbeat on the sum signal generated during the use phase, wherein the compensation is performed by using the estimated cardiogenic signal segment for the heartbeat.

    31. A process according to claim 1, wherein a computer program is provided, which can be executed on the signal processing unit wherein an execution of the program causes, during execution on the signal processing unit when the signal processing unit receives measured values from the sum signal sensor device, the signal processing unit to carry out at least some of the process steps.

    32. A process according to claim 1, wherein a signal sequence is provided, comprising commands, which can be executed on the signal processing unit, wherein an execution of the commands on the signal processing unit causes, when the signal processing unit receives measured values from the sum signal sensor device, the signal processing unit to carry out at least some of the process steps.

    Description

    BRIEF DESCRIPTION OF THE DRAWINGS

    [0113] In the drawings:

    [0114] FIG. 1 is a schematic view showing a plurality of measuring electrodes positioned on a patient and a plurality of additional sensors positioned on and above the patient, wherein the patient is being ventilated by a mechanical ventilator;

    [0115] FIG. 2 is a schematic view showing the determination of the respiratory and cardiogenic signals from the sum signal;

    [0116] FIG. 3 is a schematic view showing how a cardiogenic signal is composed from estimated signal segments for individual heartbeats;

    [0117] FIG. 4 is a schematic view showing how the influence of a transmission channel parameter is taken into consideration in the device according to FIG. 2;

    [0118] FIG. 5 is a schematic view of an embodiment showing how two transmission channel parameters are taken into consideration in the device according to FIG. 4;

    [0119] FIG. 6 is a schematic view showing in an exemplary manner several steps that are carried out during the use phase;

    [0120] FIG. 7 is a graph showing an electrical cardiogenic signal in the course of a single heartbeat;

    [0121] FIG. 8 is a graph showing as an example how sample elements, and from these a signal estimating unit, are generated and how estimated signal segments are generated and are combined into the estimated cardiogenic signal;

    [0122] FIG. 9 is a graph showing a variant of the graph shown in FIG. 8, in which the filling level of the lungs is determined by a pneumatic sensor;

    [0123] FIG. 10 is a view showing how during the training phase in the variant according to FIG. 9 the respective estimated signal segment of a class is formed in the course of a heartbeat from the segments that belong to a heartbeat each and to a lung filling level;

    [0124] FIG. 11 is a graph showing another variant of the graph shown in FIG. 8, in which the lung filling level is determined by an analysis of image sequences;

    [0125] FIG. 12 is a graph showing another variant, in which only signals in a defined frequency range are taken into consideration;

    [0126] FIG. 13 is a view showing how four shape parameter values (averaged maxima) are calculated for the four filling levels of the lungs in the variant according to FIG. 12 during the training phase;

    [0127] FIG. 14 is a graph showing another variant of the graph shown in FIG. 8, in which a singular value decomposition (SVD) applied to signal segments in order to classify the signal segments;

    [0128] FIG. 15 is a view showing how the singular value decomposition is applied during the training phase in the variant according to FIG. 14;

    [0129] FIG. 16 is a view showing how four shape parameter values (averaged signal segments) are calculated during the training phase in the variant according to FIG. 14;

    [0130] FIG. 17 is a graph showing a possible process for calculating a reference signal segment from sum signal segments during the training phase;

    [0131] FIG. 18 to FIG. 23 show a sequence in which different bands are detected after a wavelet transformation.

    DESCRIPTION OF PREFERRED EMBODIMENTS

    [0132] Referring to the drawings, the process according to the present invention is used in one application to automatically control a mechanical ventilator. This ventilator assists the spontaneous breathing of a patient or replaces same completely if the patient is sedated. This work of the ventilator, especially the times and amplitudes of the ventilation strokes, shall be synchronized—if present—with the spontaneous breathing of the patient.

    [0133] FIG. 1 shows schematically [0134] a patient P to be ventilated, [0135] the esophagus Sp of this patient P, [0136] the diaphragm Zw of this patient P, [0137] a ventilator 1 for ventilating the patient P, [0138] a first set 2.1 of measuring electrodes, which is arranged on the chest of the patient P in a position close to the heart and at a distance from the diaphragm, [0139] a second set 2.2 of measuring electrodes, which is arranged on the abdomen of the patient P in position at a distance from the heart and close to the diaphragm, [0140] a pneumatic sensor 3 in front of the mouth of the patient P, which measures the flow Vol′ of gas into and out of the airway, i.e., the volume per unit of time, and optionally the airway pressure P.sub.aw, [0141] optionally a pneumatic sensor 16 in the esophagus Sp of the patient P and [0142] an optional video camera 4, which is directed from the top onto the thoracic region and/or on the abdominal region of the patient P and generates in a contactless manner measured values in the form of images sequences, from which the current lung filling level of the patient P can be determined by image processing.

    [0143] A signal processing unit 5, which preferably belongs to the ventilator 1, generates a sum signal Sig.sub.Sum by processing measured values of the sensors 2.1 and 2.2 and/or of the pneumatic sensor 3 and/or of the optical sensor 4. This sum signal Sig.sub.Sum results from a superimposition of a respiratory signal Sig.sub.res and a cardiogenic signal Sig.sub.kar. The respiratory signal Sig.sub.res describes in this application the intrinsic breathing activity of the patient P. This respiratory signal Sig.sub.res is used for controlling the ventilator 1 and is the wanted signal. The cardiogenic signal Sig.sub.kar is caused by the cardiac activity of the patient P and is in this application an unwanted signal. The spontaneous breathing of the patient P, which is described by the respiratory signal Sig.sub.res, as well as the mechanical ventilation by the ventilator 1, generate an overall breathing and ventilation of the patient P, which is described by an overall signal Sig.sub.ges.

    [0144] FIG. 2 shows schematically and in a simplified form how the respiratory signal Sig.sub.res and the cardiogenic signal Sig.sub.kar are determined from the sum signal Sig.sub.sum. In this example the estimated (representative) cardiogenic signal Sig.sub.kar.est is subtracted from the sum signal Sig.sub.sum, and the difference is used as an estimated (representative) respiratory signal Sig.sub.res,est. Some components essential for the present invention are not shown in FIG. 2. The signal processing unit 5 yields an estimate Sig.sub.res,est for the respiratory signal Sig.sub.res and an estimate Sig.sub.kar,est for the cardiogenic signal Sig.sub.kar. The estimate ideally agrees with the actual signal, i.e., ideally Sig.sub.res=Sig.sub.res,est and Sig.sub.kar=Sig.sub.kar,est. Furthermore, ideally Sig.sub.Sum=Sig.sub.kar+Sig.sub.res=Sig.sub.kar,est+Sig.sub.res,est, i.e., ideally Sig.sub.res=Sig.sub.Sum+Sig.sub.kar,est.

    [0145] The breathing muscles AM of the patient P generate a breathing activity. The heart muscle HM generates a cardiac activity. The intrinsic breathing activity is transmitted in the body of the patient P via a transmission channel Tss to a summation point Σ, where—stated in a simplified manner—the respiratory signal Sig.sub.res appears behind the transmission channel Tss. The cardiogenic signal Sig.sub.kar is transmitted via a transmission channel Tns to the summation point Σ, and the cardiogenic signal Sig.sub.kar appears behind the transmission channel Tns. The transmission channels Tss and Tns thus influence the measured breathing activity and the measured cardiac activity. The signals Sig.sub.res and Sig.sub.kar are superimposed to one another—simply speaking—in this summation point Σ. In addition, a transmission channel Tnn is shown. The subscript s designates the wanted signal, and the subscript n (noise) the unwanted signal.

    [0146] The sensors 2.1 and 2.2 generate respective electrical measured values, as a rule, electrical voltages. A signal processor 13 with an amplifier and with an analog-digital converter processes these electrical measured values. The signal processor 13 preferably carries out, in addition, a baseline filtering, especially in order to compensate electrochemical processes in the measuring electrodes 2.1 and 2.2 on the skin of the patient P and other low-frequency potential differences by calculation. These processed measured values act in the exemplary embodiment as the sum signal or a sum signal Sig.sub.sum. The sensors 2.1 and 2.2 are therefore sum signal sensors forming a sum signal sensor device in the sense of the present invention. The pneumatic sensor 3 and the optical sensor 4 also yield measured values, from which a sum signal is generated in variants of the present invention and another parameter value is generated in other variants.

    [0147] The signal processing unit 5, which preferably forms a part of the ventilator 1, determines from this sum signal Sig.sub.sum the estimate Sig.sub.res.est for the respiratory signal Sig.sub.res being sought. The signal processing unit 5 determines for this an estimate Sig.sub.kar,est for the cardiogenic signal Sig.sub.kar, which acts in this application as an unwanted signal. In other applications, the estimated cardiogenic signal Sig.sub.kar,est is used as a wanted signal, and the respiratory signal Sig.sub.res is an unwanted signal. Or else both signals Sig.sub.res and Sig.sub.kar are wanted signals.

    [0148] FIG. 3 shows the principle of how the influence of the cardiogenic signal Sig.sub.kar on the sum signal Sig.sub.Sum is compensated by calculation during a use phase. Essential components of the present invention are also not shown in FIG. 3.

    [0149] The cardiogenic signal segment Sig.sub.Hz,kar describes an estimated segment of the cardiogenic signal in the course of a single heartbeat. A heartbeat time detector 7 detects the respective time H_Zp(n) of the nth detected heartbeat (n=1, 2, 3, . . . ). This heartbeat time detector 7 detects, for example, the so-called R peak or also the QRS curve in the sum signal Sig.sub.Sum or also in a signal that is obtained exclusively from measured values of the set 2.1 of measuring electrodes that is placed close to the heart, cf. FIG. 7. A reconstructing unit 8 combines these estimated signal segments Sig.sub.Hz,kar into a reconstructed cardiogenic signal Sig.sub.kar,est, which is used as the estimate Sig.sub.kar,est for the cardiogenic signal Sig.sub.kar, with the use of the detected heartbeat times H_Zp(x), H_Zp(x+1), . . . . This reconstructed cardiogenic signal Sig.sub.kar,est is ideally equal to the actual cardiogenic signal Sig.sub.kar, which is generated by the heart muscles HM of the patient P. A compensating unit 9 compensates the influence of the cardiogenic signal Sig.sub.kar on the sum signal Sig.sub.Sum by calculation. For example, the compensating unit 9 subtracts the reconstructed cardiogenic signal Sig.sub.kar,est from the sum signal Sig.sub.Sum. The compensating unit 9 yields in the ideal case the respiratory signal Sig.sub.res being sought, i.e., ideally Sig.sub.res equals Sig.sub.Sum−Sig.sub.kar,est.

    [0150] The respiratory signal Sig.sub.res and/or the cardiogenic signal Sig.sub.kar are influenced by at least one respective anthropological variable in the body of the patient P. A measurable parameter, which acts on at least one above-described transmission channel Tss, Tns and is therefore called transmission channel parameter, correlates with the anthropological variable or with at least one anthropological variable. This influence is not taken into consideration in FIG. 2 and FIG. 3. It will be described below how this is taken into account according to the present invention.

    [0151] FIG. 4 shows as an example an influence on the transmission channel Tns from the breathing muscles AM, which are the signal source for the respiratory signal Sig.sub.res, to the sensor 2.1, 2.2, namely, the lung filling level, LF. The current filling level LF of the lungs of the patient P changes the distance between the breathing muscles AM and the sensor 2.1, 2.2 and hence the length and also the other properties of the transmission channel Tns. The current lung filling level, LF, is correlated with the flow Vol′ of breathing air or of another gas into and out of the airway of the patient P, i.e., with the volume fed or removed per unit of time. The pneumatic sensor 3 in front of the mouth of the patient P is capable of measuring this volume flow Vol′. In the exemplary embodiment, This measured volume flow Vol′ is integrated over time and the run time of gas between the sensor 3 and the mouth as well as between the mouth and the lungs of the patient P as well as optionally the elasticity of the lungs and the resistance offered by the airway of the patient P to the flow of breathing air are additionally taken into consideration. The respective current value is determined in this manner for the transmission channel parameter LF.

    [0152] FIG. 5 shows how the principle illustrated in FIG. 4, namely, the taking into account of the lung filling level, LF, is applied to the principle illustrated in FIG. 3 in order to compensate the influence of the cardiogenic signal Sig.sub.kar on the sum signal Sig.sub.Sum. A use path Npf and a training path Tpf are shown in FIG. 5 and in figures following it. The use path Npf describes the steps and the components used during the use phase Np, and the training path Tpf describes the steps and the components used during the training phase Tp and during the subsequent adaptation phase Ap, which overlaps with the use phase Np.

    [0153] An additional transmission channel parameter, namely the position Pos of a measuring electrode 2.1 or 2.2 relative to the signal source for the cardiogenic signal Sig.sub.kar, is optionally taken into account in the example shown in FIG. 5. A mechanical sensor 10, for example, an acceleration sensor or a strain gauge, measures the position Pos of measuring electrodes 2.1 or 2.2 relative to a predefined reference point in or at the body of the patient P and hence relative to the heart, i.e., to the signal source HM for the cardiogenic signal Sig.sub.kar. Repeatedly a respective value for the transmission channel parameter LF is derived from the measured values from the sensor 3, and a value for the transmission channel parameter Pos is derived from the measured values from the sensor 10.

    [0154] A third transmission channel parameter, which does not require an additional sensor, especially the length of a heartbeat or also the time period between the two characteristic times H_Zp(x), H_Zp(x+1) of two consecutive heartbeats or the time interval between two signal peaks, e.g., the P peak and the T peak, of the segment Abs.x, Abs.y, . . . of the sum signal Sig.sub.Sum, which occurs in the course of a single heartbeat, is optionally taken into consideration. This time period may remain the same over time or vary over time. A heartbeat time period detector 11 analyzes the sum signal Sig.sub.Sum and the detected heartbeat times H_Zp(x), H_Zp(x+1), . . . and calculates the time interval between two consecutive heartbeat times.

    [0155] A heartbeat time detector 7 detects, in turn, the respective time H_Zp(n) (n=1, 2, . . . ) of each heartbeat. A signal estimating unit (a signal representation unit) 6 receives the measured values from the two sensors 3 and 10 and calculates from these the respective current value, which the transmission channel parameter LF or Pos assumes at the heartbeat time H_Zp(x).

    [0156] Depending on the measured value for the lung filling level, LF, and optionally on the measured value for the relative position Pos during a heartbeat, the signal estimating unit 6 calculates during the use phase Np a respective estimated signal segment Sig.sub.Hz,kar.LF of the cardiogenic signal Sig.sub.kar in the course of this heartbeat for each heartbeat, wherein the estimated signal segment Sig.sub.Hz,kar.LF depends on the lung filling level, LF, at this heartbeat as well as optionally on the position Pos of the measuring electrode 2.1 or 2.2 and/or on the time interval RR between two heartbeats. This signal segment Sig.sub.Hz,kar.LF estimated as a function of at least one transmission channel parameter, varies, as a rule, from one heartbeat to the next. The estimated signal segments Sig.sub.Hz,kar.LF are combined into the estimated cardiogenic signal Sig.sub.kar,est with the use of the heartbeat times.

    [0157] In one embodiment, each estimated signal segment Sig.sub.Hz,kar.LF has the same length. The intermediate space in the estimated signal Sig.sub.kar,est is bridged over by a smoothing. In another embodiment, the respective time period H_Zr(x), H_Zr(x+1), . . . of each heartbeat is measured during the use phase Np, and the estimated signal segment Sig.sub.Hz,kar.LF is adapted to this heartbeat time period by stretching or compression.

    [0158] The signal estimating unit 6 has in one embodiment reading access to a predefined standard reference signal segment Sig.sub.Hz,Ref, which is stored in a library. This segment describes an average segment of the cardiogenic signal Sig.sub.kar in the course of a single heartbeat. This standard reference signal segment Sig.sub.Hz was generated, for example, in advance by measurements on different patients. It contains at least one and preferably a plurality of shape parameters, which change the geometric shape of the reference signal segment Sig.sub.Hz,Ref. The influence of a transmission channel parameter is taken into account indirectly by at least one shape parameter, which will be described farther below.

    [0159] Examples of shape parameters are, cf. FIG. 7: [0160] the duration of the QRS phase, [0161] the QRS amplitude, [0162] the respective amplitude of the Q peak, R peak, S peak and [0163] the time period between the P peak and the T peak.

    [0164] Due to a respective shape parameter value each being introduced into the shape parameter or each shape parameter of this standard reference signal segment Sig.sub.Hz,Ref, a parameterized cardiogenic estimated signal segment Sig.sub.Hz,kar.LF is generated, which describes the estimated cardiac activity in the course of an individual heartbeat and depends in this example on the lung filling level, LF, and optionally on the position Pos. This parameterized standard reference signal segment Sig.sub.Hz,kar.LF is shown in FIG. 5 as the expected signal segment Sig.sub.Hz,kar in the course of an individual heartbeat, as this is shown in FIG. 3.

    [0165] These shape parameter values depend in the example according to FIG. 5 on the current value of the lung filling level, LF, on the other hand. The current lung filling level, LF, is measured in the example according to FIG. 5 by at least one pneumatic sensor 3, and this pneumatic sensor 3 measures the volume flow Vol′ and optionally also the airway pressure P.sub.aw. The shape parameter values optionally depend, in addition, on the position Pos.

    [0166] In one embodiment, the signal estimating unit 6 calculates for each shape parameter of the standard reference signal segment Sig.sub.Hz,Ref and for each detected heartbeat a respective shape parameter value each, which the shape parameter assumes at the heartbeat time H_Zp(x) or during the heartbeat time period H_Zr(x). Using these shape parameter values, the signal processing unit 5 generates during the use phase from the standard reference signal segment Sig.sub.Hz,Ref for each heartbeat an estimated signal segment Sig.sub.Hz,kar.LF, which is adapted to the current value of the lung filling level, LF, and optionally to the current position Pos and/or other transmission channel parameter, and which describes the expected or estimated cardiogenic signal Sig.sub.res in the course of this heartbeat. This is carried out for each heartbeat detected during the use phase Np.

    [0167] In another embodiment, the signal estimating unit 6 determines in a library 12 a stored reference signal segment Sig.sub.Hz,kar,LF.1 or . . . Sig.sub.Hz,kar,LF.4, which segment is associated with this lung filling level, LF.1, . . . , LF.4 and optionally to this position Pos. The signal estimating unit 6 yields the estimated signal segment Sig.sub.Hz,kar,LF for a heartbeat as a function of the determined reference signal segment or of each determined reference signal segment. No standard reference signal segment Sig.sub.Hz,ref is needed after the end of the training phase Tp in this embodiment. The reconstructing unit 8 combines, in both embodiments during the useful segment Np, the estimated cardiogenic signal segments Sig.sub.Hz,kar,LF in the course of a respective heartbeat each into an estimated cardiogenic signal Sig.sub.kar,est and uses for this the heartbeat times H_Zp(x), H_Zp(x+1), . . . , which the time detector 7 has detected. According to the embodiment of the present invention, the reconstructing unit 8 combines the estimated signal segments Sig.sub.Hz,kar,LF adapted to the current filling levels of the lungs, LF, into the reconstructed cardiogenic signal Sig.sub.kar,est. This is preferably repeated continually as soon as a new heartbeat is detected.

    [0168] A plurality of variants of the process according to the present invention will be described below, as it is illustrated by FIG. 4 and FIG. 5. The variants differ by the sensors, from the measured values of which the sum signal Sig.sub.Sum is generated, by the transmission channel parameters taken into consideration and/or by the sensors in order to measure the values of these transmission channel parameters taken into account. In one variant, estimated signal segments are not combined into the cardiogenic signal Sig.sub.kar,est, but into the respiratory signal Sig.sub.res,est.

    [0169] FIG. 6 shows as an example several steps that are carried out during the use phase in order to determine the estimated respiratory signal Sig.sub.res,est. The following steps are shown: [0170] The measuring electrodes 2.1 and 2.2, the pneumatic sensor 3 and/or the optical sensor 4 yield measured values. [0171] The signal processor 13 processes the measured values from the sensors 2.1, 2.2, 3, 4 and yields a sum signal Sig.sub.Sum, [0172] The heartbeat time detector 7 detects the respective heartbeat time H_Zp(n) of the nth detected heartbeat. The heartbeat time detector 7 analyzes for this the sum signal Sig.sub.Sum and/or measured values from the measuring electrode set 2.1 located close to the heart. [0173] The signal estimating unit 6 has reading access to the library 12, in which different reference signal segments Sig.sub.Hz,kar,LF.1, . . . , Sig.sub.Hz,kar,LF.4 for different possible filling levels of the lungs, LF.1, . . . , LF.4 are stored. [0174] The signal estimating unit 6 determines for each heartbeat from the measured heartbeat times H_Zp(x1), H_Zp(x2), . . . and from the measured filling levels of the lungs, LF.1, LF.2 a respective set of shape parameter values FP-W(1), FP-W(2), . . . and herefrom an estimated signal segment Sig.sub.Hz,kar.LF(x1), Sig.sub.Hz,kar,LF(x2), . . . each, for example by introducing the shape parameter values FP-W(1), FP-W(2) into a standard reference signal segment Sig.sub.Hz,Ref. [0175] The reconstructing unit 8 combines these estimated signal segments Sig.sub.Hz,kar.LF(x1), Sig.sub.Hz,kar.LF(x2), . . . into an estimated cardiogenic signal Sig.sub.kar,est. [0176] The heartbeat time period detector 11 optionally measures the respective heartbeat time period H_Zr(x), H_Zr(x+1) of each heartbeat. [0177] The compensating unit 9 compensates by calculation the influence of the respiratory signal Sig.sub.res on the sum signal Sig.sub.Sum, for example, by subtracting the estimated cardiogenic signal Sig.sub.kar,est from the sum signal Sig.sub.Sum and/or the signal segment Sig.sub.HzmkarmLF(x1), Sig.sub.Hz,kar.LF(x2) estimated for this heartbeat from the sum signal Sig.sub.Sum.

    [0178] FIG. 7 shows an exemplary segment of an electrical cardiogenic signal Sig.sub.kar in the course H_Zr(n) of a single heartbeat. The time is plotted on the x axis and the cardiogenic signal in mV is plotted on the y axis. The P peak, the Q peak, the R peak, the S peak and the T peak are shown. The cardiogenic signal Sig.sub.kar and therefore also the sum signal Sig.sub.Sum have approximately the same course for each heartbeat in the range of the P peak through the T peak.

    [0179] The R peak is used in one embodiment as a characteristic time H_Zp(n) of a heartbeat. In addition, the following geometric parameters are illustrated: [0180] The R-R interval RR between the R peaks of two consecutive heartbeats, [0181] the QRS amplitude QRS, i.e., the distance between the highest value and the lowest value during the time period between the Q peak and the S peak, [0182] the P-Q time interval PQ, i.e., the time period between the P peak and the Q peak, and [0183] the S-T time interval ST, i.e., the time interval between the S peak and the T peak.

    [0184] The R-R interval RR correlates with the pulse of the patient P.

    [0185] FIG. 8 shows as an example how the sample elements are generated and used according to a first variant. Shown are [0186] The training phase Tp, during which a sample 14, optionally a library 12 and then an initial signal estimating unit 6 are generated, [0187] the subsequent adaptation phase Ap, during which this signal estimating unit 6 is continually adapted to the sample elements obtained so far during the use phase Np, as well as [0188] the use phase Np, during which the signal estimating unit 6 is used.

    [0189] The adaptation phase Ap overlaps the use phase Np. The time is plotted from left to right on the respective x axis of each signal. The time curves of the following signals are shown: [0190] The sum signal Sig.sub.Sum, [0191] the respective characteristic heartbeat time H_Zp, [0192] the curve of the volume flow Vol′ and [0193] the curve of the lung filling level, LF.

    [0194] In this variant, the sum signal Sig.sub.Sum is generated by analyzing electrical measured values of the measuring electrodes 2.1 and 2.2. The volume flow Vol′ is measured, for example, by means of the pneumatic sensor 3, and the current lung filling level, LF, is derived from the respective volume flow Vol′ at a plurality of times. Four classes of filling levels of the lungs, namely, LF.1 (lungs almost empty, lung filling level below a first limit), LF.4 (lungs almost full, lung filling level above a second limit) and two filling levels of the lungs in between, LF.2 and LF.3, are distinguished in the example shown. A different number of classes of filling levels of the lungs, LF, and of other transmission channel parameters may, of course, be distinguished as well. The signal with the time curve that indicates the class to which the current lung filling level, LF, belongs, is designated by LF_cl in FIG. 8.

    [0195] In an embodiment of the example according to FIG. 8, each sample element comprises a segment of the sum signal Sig.sub.Sum in the course of an individual heartbeat, for example, segment Abs.x in the course of the heartbeat with the characteristic heartbeat time H_Zp(x). In addition, each sample element comprises a class each of the lung filling level, LF, for example, class LF.3 for the heartbeat time H_Zp(x). It is illustrated by means of a plurality of arrows in the bottom part of FIG. 8 how four classes LF.1 through LF.4 of sample elements are generated. The corresponding segments of the sum signal Sig.sub.Sum, which belong to the sample elements of one class, are brought to the same length by projecting segments being cut off by calculation, and then being superimposed with the correct time. The arithmetic mean of the segments superimposed with the correct time is formed or these segments are combined in another manner into a reference signal segment, which reference signal segment is assigned to the class of filling levels of the lungs. A computer-accessible library 12 with—in this case four—stored reference signal segments Sig.sub.Hz,kar,LF.1, . . . , Sig.sub.Hz,kar,LF.4 of the cardiogenic signal is generated hereby in the course of a heartbeat. Each reference signal segment SigSig.sub.Hz,kar,LF.1, . . . , Sig.sub.Hz,kar,LF.4 is assigned in the library 12 to a possible lung filling level class LF.1, . . . , LF.4.

    [0196] The class to which the lung filling level, LF(t), belongs at the time t=H_Zp (n) is determined during the use phase Np in one embodiment at the characteristic heartbeat time H_Zp(n) of the nth heartbeat. In one embodiment, the respective reference signal segment which is assigned to this class in the library 12, is used as the estimated signal segment Sig.sub.Hz,kar,LF(n). It describes the segment of the cardiogenic signal in the course of this heartbeat. For example, the reference signal segment Sig.sub.liz,kar,LF.3 for the lung filling level LF.3 is selected for the time H_Zp(y) and is used as an estimated signal segment Sig.sub.Hz,kar,LF(y), and the reference signal segment Sig.sub.Hz,kar,LF.4 for the lung filling level LF.4 is used for the time H_Zp(z) as the estimated signal segment Sig.sub.Hz,kar,LF(z).

    [0197] In another embodiment, the signal processing unit 5 calculates for each class of lung filling levels, in addition to the reference signal segment, a respective reference parameter value each, for example, as a weighted mean value or as a center or median of the transmission channel parameter values (here: lung filling levels) of this class. For example, the relative frequencies of transmission channel parameter values are used as weighting factors. The signal processing unit 5 determines during the use phase Np for each heartbeat the two reference parameter values that are closest to the transmission channel parameter value of this heartbeat and calculates the estimated signal segment for this heartbeat by smoothing, for example, an interpolation or regression.

    [0198] The signal estimating unit 6 consequently yields for each heartbeat time H_Zp(y) an estimated signal segment Sig.sub.Hz,kar,LF(y), which depends on the four possible reference signal segments Sig.sub.Hz,kar,LF.1, . . . Sig.sub.Hz,kar,LF.4. In one embodiment, each estimated signal segment Sig.sub.Hz,kar,LF(y) of the cardiogenic signal is equal to a reference signal segment Sig.sub.Hz,kar,LF.1, . . . , Sig.sub.Hz,kar,LF.4 in the library 12. The estimated signal segment provided depends on which of the four classes LF.1, . . . , LF.4 the lung filling level LF belongs to during this heartbeat.

    [0199] This is carried out after the training phase Tp, i.e., when the signal estimating unit 6 is generated. A respective standard reference signal segment predefined in advance for each detected heartbeat is preferably used for each heartbeat time before the end of the training phase Tp.

    [0200] These estimated signal segments Sig.sub.Hz,kar,LF are combined by the reconstructing unit 8 into the estimated cardiogenic signal Sig.sub.kar,est. This estimated cardiogenic signal Sig.sub.kar,est as well as the estimated respiratory signal Sig.sub.res,est are shown below the curve LF_cl in FIG. 8. The estimated respiratory signal Sig.sub.res,est is generated by the compensating unit 9 subtracting from the measured sum signal Sig.sub.Sum the estimated cardiogenic signal Sig.sub.kar,est generated by combination, i.e., Sig.sub.res,est=Sig.sub.Sum−Sig.sub.kar,est. As can be seen, the estimated respiratory signal Sig.sub.res,est usually assumes the zero value because the heart rate is several times higher than the respiration rate and the cardiogenic signal Sigk.sub.ar is several times stronger in the P-T segment of a heartbeat than the respiratory signal Sig.sub.res. Three breathing processes of the patient P lead to three oscillations Atm.1, Atm.2, Atm.3 of the estimated respiratory signal Sig.sub.res,est shown. FIG. 9 shows a variant of the approach shown in FIG. 8. The coordination between the spontaneous breathing and the heartbeat of the patient P, more precisely the event of whether the exhalation begins shortly before the Q wave of the next heartbeat or not, is used as an additional transmission channel parameter. The signal S_Q shows the time curve of this additional transmission channel parameter. The classes are formed depending on two transmission channel parameters, namely, the lung filling level LF and the exhalation time close to Q (yes/no).

    [0201] In a possible embodiment, this leads with four classes LF.1, . . . , LF.4 for the lung filling level LF and for two classes for the exhalation time (yes and no, i.e., breathing begins and breathing does not begin shortly before the Q wave) to a total of 2×4=8 different classes. By contrast, only four classes are used in the embodiment shown. The possible values for the lung filling level LF are grouped into three classes LF.a, LF.b, LF.c. In connection with the event that the exhalation time is not close to Q, this leads to three classes LQ.a, LQ.b, LQ.c. In addition, a fourth class Q.d is introduced, namely, that the exhalation time is close to Q, regardless of how the lung filling level LF is. Furthermore, the time curve of the belonging to one of these four classes LF.a, LF.b, LF.c, Q.d is shown in FIG. 9, which is designated by LF_Q_cl.

    [0202] The sum signal Sig.sub.Sum is a pressure signal in this variant, which is measured in or in front of the esophagus Sp (esophagus) of the patient P, for example, with a probe or with a balloon in the esophagus Sp. The pressure signal could also be the pressure P.sub.aw at the transition from a hose of the ventilator 1 to the mouth of the patient P, which is measured by the sensor 3. This pneumatic sum signal Sig.sub.Sum results from a superimposition of the pneumatic respiratory signal Sig.sub.res brought about by the breathing activity to a pneumatic cardiogenic signal Sig.sub.kar brought about by the cardiac activity.

    [0203] In the variant shown, the signal processing unit 5 therefore additionally carries out a detrending in the training phase Tp and therefore in the training path Tpf. As a result, the risk that different trends would distort the combination of the sum signal segments arranged in correct time into a reference signal segment is reduced. Both the sum signal Sig.sub.Sum and the processed sum signal Sig.sub.Sum,DT generated by the detrending are shown in FIG. 9.

    [0204] An embodiment of generating the detrending is the following: The signal processing unit determines for each heartbeat the sum signal segment Abs.w, Abs.x belonging to this heartbeat. It calculates a fitted curve, especially a fitted curve through this sum signal segment Abs.w, Abs.x. This fitted curve is generated, for example, by interpolation or as a straight line from the chronologically first to the chronologically last signal value of the sum signal segment Abs.w, Abs.x. The respective fitted curve is subtracted for each heartbeat from the sum signal segment Abs.w, Abs.x. The remaining residue, i.e., the difference, forms the processed sum signal segment Abs_DT.w, Abs_DT.x generated by the detrending. Each sample element comprises such a processed sum signal segment. These segments yield the estimated signal segments Sig.sub.Hz,kar,LD(y), Sig.sub.Hz,kar,LF(z) which are combined into the processed sum signal Sig.sub.Sum,DT.

    [0205] The signal estimating unit 6 yields during the use phase a respective processed sum signal segment Abs_DT.w, Abs_DT.x for each detected heartbeat.

    [0206] The signal estimating unit 6 yields a respective estimated signal segment Sig.sub.Hz,kar,LQ, which is selected among four possible reference segments SigSig.sub.Hz,kar,LQ.a, . . . , SigSig.sub.Hz,kar,Q.d of the cardiogenic signal Sig.sub.kar, during the use phase Np for each heartbeat in one embodiment in the variant according to FIG. 9 as well, and the particular estimated signal segment which the signal estimating unit 6 provides for a heartbeat depends on the lung filling level LF and on the exhalation time during the heartbeat.

    [0207] FIG. 10 shows how the four reference signal segments Sig.sub.Hz,kar,LQ.a, . . . , Sig.sub.Hz,kar,LQ.d of the cardiogenic signal Sig.sub.kar are formed for the four different classes (lung filling levels and Q values) LQ.a, LQ.b, LQ.c, Q.d. The segments of the sum signal Sig.sub.Sum, which are superimposed with correct time, and which belong to the same class, i.e., here to the same lung filling level/Q value LQ.a, LQ.b., LQ.c, Q.d here, are shown in the left column of FIG. 10. The corresponding reference signal segment Sig.sub.Hz,kar,LQ.a, . . . , Sig.sub.Hz,kar,Q.d of the cardiogenic signal is shown in the right column for a class LF.1, . . . , LF.4, which is formed by calculating the arithmetic mean from the signal segments superimposed with correct time for a respective heartbeat. The content of the right column is stored in the library 12.

    [0208] In the variant according to FIG. 11, the sum signal Sig.sub.Sum is determined by an automatic image analysis of image sequences, wherein the video camera 4 is directed towards the thoracic region of the patient P and yields these image sequences. The sum signal Sig.sub.Sum, which is shown in the second row of FIG. 11, is formed from a superimposition of a respiratory signal to a cardiogenic signal in this variant as well. The current lung filling level LF of the patient P is likewise derived from measured values of the pneumatic sensor 3. It is possible to use additionally signals from the video camera 4 to determine the current lung filling level, since these signals show the thoracic region of the patient P, and this region rises and falls depending on the breathing. The topmost row of FIG. 11 shows as a measured value series MWR a sequence of images that have been recorded by the video camera 4. The above-described detrending is applied to the sum signal segments in this variant as well.

    [0209] In the variant according to FIG. 12, the sum signal Sig.sub.Sum is likewise generated from electrical measured values of the measuring electrodes 2.1 and 2.2. The pneumatic sensor 3 likewise measures the volume flow Vol′, and the signal processing unit 5 calculates the current lung filling level LF from a plurality of values for the volume flow Vol′. Four possible lung filling levels LF.1, . . . , LF.4 are distinguished again. No estimated cardiogenic signal Sig.sub.kar,est is calculated in this variant. The estimated respiratory signal Sig.sub.res,est is rather extracted by calculation from the sum signal Sig.sub.Sum in another manner. No reference signal segments are used in this variant. At least two frequency ranges are predefined: A lower-frequency range and a higher-frequency range in the variant shown. For example, a frequency range results from frequencies in which an electrically measured respiratory signal (EMG) can occur, and another frequency range from frequencies, in which an electrically measured cardiogenic signal (ECG) can occur.

    [0210] The sum signal Sig.sub.Sum is broken down in the example shown into a respective signal component per predefined frequency range in both the training phase Tp and the use phase Np. For example, a wavelet transformation or a band filter or a low-pass filter or a high-pass filter is employed. FIG. 12 shows the signal component Sig.sub.Sum,low for the lower frequency range and the signal component Sig.sub.Sum,high for the higher frequency range. The signal component Sig.sub.Sum,low for the lower frequency range is caused essentially, i.e., aside from a negligibly small residue, by the cardiac activity HM of the patient P and is not used for the calculation of the estimated respiratory signal Sig.sub.res,est. The signal component Sig.sub.Sum,high for the higher frequency range results from a superimposition of the respiratory signal Sig.sub.res to a higher-frequency component of the cardiogenic signal Sig.sub.kar.

    [0211] The respective maximum and the respective minimum are detected in the course of a heartbeat in the signal component Sig.sub.Sum,high during the training phase Tp. For example, two maxima Max.1 and Max.8 are shown. The same is carried out for the minima. For example, a minimum Min.1 is shown. These maxima are divided into four classes of maxima in the respective heartbeat depending on the particular lung filling level LF.1, . . . , LF.4. FIG. 13 shows in the left column (sample 14) by means of four histograms the maxima of these four classes. Each rectangle corresponds to a class. The value of the maximum, i.e., a value in mV, the frequency of this maximum in a class of lung filling levels LF.1, . . . , LF.4, is plotted on the x axis of a histogram. A characteristic value, for example, an arithmetic mean or a median or maxima, is calculated for each class of maxima. Especially the two mean values or medians Max_MW.LF.1 and Max_MW.LF.2 for the two classes, which belong to the lung filling levels LF.1 and LF.2, respectively, are shown in FIG. 13. It is shown in the right column in FIG. 13 (library 12) how a respective averaged maximum, i.e., an arithmetic mean or median or maxima, is assigned as a shape parameter value to each class LF.1, . . . , LF.4 of lung filling levels. These are stored in the library 12. Furthermore, an averaged minimum, which was determined in a corresponding manner, is associated with each class. The two shape parameter values are used to parameterize a change rule (calculation rule), which will be described below.

    [0212] The signal estimating unit 6 determines the sum signal segment Abs.x, the higher-frequency signal component segment and the respective lung filling level for each heartbeat in the use phase. The signal estimating unit 6 determines a respective averaged maximum as well as a respective averaged minimum, for which the signal estimating unit 6 uses the measured lung filling level LF at this heartbeat as well as the maxima and minima determined in the library 12. The signal estimating unit 6 cuts off by calculation the components that are above the averaged maxima or below the averaged minima in the segment of the higher-frequency signal component Sig.sub.Sum,high that belongs to this heartbeat. These components originate with certainty essentially from the cardiogenic signal Sig.sub.kar and contain no respiratory components that is to be taken into account. The cutting off is illustrated in FIG. 12 on the basis of the two averaged maxima Max_MW.LF.1 and Max_MW.LF.2 stored in the library 12. The remaining components, i.e., the components of the higher-frequency signal component Sig.sub.Sum,high that are located between the weighted minimum and the weighted maximum, originate from the respiratory signal Sig.sub.res and are preferably smoothed by calculation. The gaps formed due to the cutting off are set, for example, at zero, or they are interpolated in a suitable manner between the remaining components. A respective signal component Sig.sub.Hz,res,LF(y), Sig.sub.Hz,res,LF(z), . . . , which describes the estimated respiratory signal in the course of this heartbeat, is generated in this manner for each heartbeat. The reconstructing unit 8 combines these signal segments Sig.sub.Hz,res,LF(y), Sig.sub.Hz,res,LF(z), lung filling level into the estimated respiratory signal Sig.sub.res,est.

    [0213] Averaged maxima and averaged minima are used in this example as shape parameter values of a class of transmission channel parameter values (here: lung filling level end LF.1, . . . , LF.4). These shape parameter values are used in this variant to parameterize a predefined change rule. The parameterized change rule changes a respective segment Abs.x, Abs.y of the sum signal Sig.sub.Sum—in this variant: a segment of the higher-frequency signal component Sig.sub.Sum,high. The change comprises in this variant the step of cutting off signal components above the maxima and below the minima.

    [0214] It is also possible to use additional or other arithmetic shape parameters and hence other change rules, e.g., averaged first and/or second derivatives. It is also possible to use weighting factors and/or a “soft threshold.” A segment of the sum signal Sig.sub.Sum, which segment belongs to a heartbeat, or of a signal component in the segments in which the slope of the sum signal Sig.sub.Sum is below a predefined limit, is stretched in another embodiment. Due to the variant shown in FIG. 12 and FIG. 13, an estimated respiratory signal Sig.sub.res,est, for which a higher-frequency signal component Sig.sub.Sum,high is used, is calculated. The described process can also be applied to calculate an estimated cardiogenic signal Sig.sub.kar,est. The process is applied correspondingly to the lower-frequency signal component Sig.sub.Sum,low for this application. A respective estimated signal segment Sig.sub.Hz,kar.LF of the cardiogenic signal Sig.sub.kar,est is preferably calculated for each heartbeat. The segment of the lower-frequency signal component Sig.sub.Sum,low, which segment belongs to this heartbeat, as well as the areas of the higher-frequency signal component Sig.sub.Sum,high that are above the averaged maximum or below the averaged minimum for this heartbeat are combined for this purpose into the signal segment Sig.sub.Hz,kar,LF for a heartbeat. The reconstructing unit 8 combines these estimated signal segments Sig.sub.Hz,kar,LF into the estimated respiratory signal Sig.sub.kar,est.

    [0215] In a preferred application of the variant that is shown by FIG. 12 and FIG. 13, two frequency ranges are predefined, namely, a frequency range from f1 to f2 for the ECG signal (cardiogenic signal) and a frequency range from f3 to f4 for the EMG signal (respiratory signal). Now: f1<f3<f2<f4, i.e., the two frequency ranges overlap in the range from f3 to f2. The sum signal Sig.sub.Sum is divided by calculation into three signal components, namely, a signal component for the frequency range from f1 to f3, a signal component for the overlapping frequency range from f3 to f2, and a signal component for the frequency range from f2 to f4. The lower-frequency signal component in the range from f1 to f3 is essentially a cardiogenic signal, i.e., the respiratory component in the lower-frequency signal component may be ignored. The high-frequency signal component in the range from f2 to f4 is essentially a respiratory signal, and the medium-frequency signal component in the range from f3 to f2 results from a superimposition of the respiratory signal to the cardiogenic signal, which superimposition is to be taken into account. The process just described is carried out only for this overlapping frequency range from f3 to e, i.e., especially the two signal components Sig.sub.Sum,high and Sig.sub.Sum,low are formed. The estimated respiratory signal Sig.sub.res,est is combined from the component in the high-frequency range from f2 to f4 as well as from the respiratory signal obtained as just described in the overlapping frequency range from f3 to f2. The estimated cardiogenic signal Sig.sub.kar,est is correspondingly combined from the component in the lower-frequency range from f1 to f3 as well as from the cardiogenic signal obtained as just described in the overlapping frequency range from f3 to f2.

    [0216] In the embodiments just described, the signal processing unit 5 receives a plurality of measured values from at least one sensor, wherein this sensor is not a sum signal sensor 1, 2.1, 2.2, 3, 4, and it generates by signal processing the transmission channel parameter value or each transmission channel parameter value from these measured values. It is also possible that the signal processing unit 5 calculates the value of at least one transmission channel parameter and measures it by the calculation by the signal processing unit 5 analyzing the sum signal Sig.sub.Sum. Another sensor for the transmission channel parameter is thus unnecessary for this transmission channel parameter.

    [0217] Possible transmission channel parameters, which can be measured by calculation and without a separate physical sensor, are shown in FIG. 7, namely, [0218] the R-R interval RR, [0219] the QRS amplitude QRS, [0220] the P-Q time interval PQ, [0221] the P-T time interval and [0222] the S-T time interval ST.

    [0223] FIG. 14 through FIG. 16 show another variant, in which no additional physical sensor is needed to measure a transmission channel parameter. The basic idea of this variant is that at least one reference curve, preferably two or three reference curves, are determined before the beginning of the training phase Tp or else during the training phase Tp. The signal processing unit 5 calculates in the use phase Np a respective individual agreement value, i.e., a value for the agreement between the sum signal segment and the reference curve, for each sum signal segment Abs.x, Abs.y, . . . and each reference curve. Each sum signal segment Abs.x, Abs.y, . . . is preferably standardized in advance. The signal processing unit 5 calculates from the individual agreement values an overall agreement value. This overall agreement value acts in this variant as the transmission channel parameter or a transmission channel parameter. As in the variants described above, the signal processing unit 5 has reading access to a library 12, in which a respective reference signal segment is stored for each class of transmission channel parameter values, in this additional variant as well. In this case, each class is a range of possible overall agreement values. Depending on the calculated overall agreement values between a sum signal segment for a heartbeat and the reference curves V.1, V.2, . . . used, the signal processing unit 5 selects in the use phase Np for each heartbeat at least one respective reference signal segment from the library 12 and uses it as the estimated signal segment Sig.sub.Hz,kar,ÜM for this heartbeat or yields an estimated signal segment Sig.sub.Hz,kar,ÜM depending on the selected reference signal segments. The signal processing unit 5 combines the estimated signal segments Sig.sub.Hz,kar,ÜM provided in this manner with the use of the heartbeat times into the estimated cardiogenic signal Sig.sub.kar,est or compensates the influence of the cardiac activity on the sum signal and uses the provided estimated signal segments and the heartbeat times for the compensation.

    [0224] An embodiment of this variant will be explained below with reference to FIG. 14 through FIG. 16. The sum signal Sig.sub.Sum is likewise divided into sum signal segments Abs.x, Abs.y, . . . , namely, into one signal segment for each heartbeat. These sum signal segments may have different lengths. By the signal processing unit cutting off parts of the sum signal segments when needed, it generates a sample, in which the sample elements comprise segments of equal length of the sum signal Sig.sub.Sum. The relative times of the five peaks (P peak through T peak, see FIG. 7) of these signal segments differ from one another as little as possible. These equal-length signal segments, arranged with the correct time, will hereinafter be called standardized signal segments and are designated by Abs_std.x, Abs_std.y, . . . in FIG. 15.

    [0225] These standardized signal segments Abs_std.x, Abs_std.y, . . . are arranged in a matrix M. Each row of this matrix represents a heartbeat and each column a scanning time. The signal processing unit applies to the set of these standardized signal segments in a first part Tpf.1 of the training path Tpf a singular value decomposition (SVD) or also a principal component analysis (PCA). This step yields a plurality of reference curves in a decreasing order, wherein the order depends decreasingly on an agreement value. The first reference curve V.1 agrees with the standardized signal segments most strongly, etc. The three most important reference curves V.1 through V.3 are shown in FIG. 15 in a decreasing order from top to bottom. The standardized signal segments can be reconstructed again from these reference curves.

    [0226] The reference curves V.1, V.2 are predefined in an alternative embodiment.

    [0227] The signal processing unit 5 classifies next the standardized sum signal segments Abs_std.x, Abs_std.y, i.e., in a second part Tpf.2 of the training path Tpf. Only the two most important reference curves V.1 and V.2 are used for this in the example being shown. It is also possible to use more than two reference curves. The signal processing unit 5 calculates for each sum signal segment Abs_std.x, Abs_std.y, . . . a respective value each for the agreement between this standardized sum signal segment and the reference curve V.1, V.2 used. For example, it calculates the scalar product between the standardized sum signal segment Abs_std.x, Abs_std.y, . . . and the reference curve V.1, V.2. The time course ÜM.1 of the individual agreement value for the first reference curve V.1 and the time course ÜM.2 of the individual agreement value for the second reference curve V.2 are shown in FIG. 14. The signal processing unit 5 then classifies each standardized sum signal segment on the basis of the two calculated individual agreement values. Two classes each of individual agreement values are used in the example shown per reference curve V.1, V.2, so that the standardized reference curves are groups into a total of 2*2=4 groups. These are called ÜM.a, . . . , ÜM.d. Furthermore, the time curve ÜM_cl of this classification is shown in FIG. 14.

    [0228] FIG. 16 shows in the left column (sample 14) the standardized sum signal segments Abs_std.x, Abs_std.y, . . . , which are divided into the four classes ÜM.a, ÜM.dThe signal processing unit 6 aggregates the standardized signal segments Abs std.x, Abs std.ye, . . . of a class ÜM.a, . . . , ÜM.d into a respective reference signal segment Sig.sub.Hz,kar,ÜM.a, . . . , Sig.sub.Hz,kar,ÜM.d each per class, for example, by forming for each relative scanning time the mean value or the median over the standardized signal segments Abs_std.x, Abs_std.y of this class. The library 12 with four reference signal segments Sig.sub.Hz,kar,ÜM.a, . . . , Sig.sub.Hz,kar,ÜM.d in this case is shown in the right column. The signal processing unit 5 generates in the use phase Np a standardized sum signal segment for each detected heartbeat from the corresponding sum signal segment Abs.x, Abs.y, . . . and calculates the respective individual agreement value between this standardized sum signal segment and each reference curve V.1, V.2, . . . , for example, as a scalar product. These two (or three) individual agreement values are summed up by the signal processing unit 5 into a preferably two-dimensional overall agreement value. Depending on this overall agreement value ÜM.a, . . . , ÜM.d, the signal processing unit 5 selects in the library 12 a standardized reference signal segment Sig.sub.Hz,kar,ÜM.a, . . . , Sig.sub.Hz,kar,ÜM.d and uses it as an estimated signal segment Sig.sub.Hz,kar,ÜM(y), Sig.sub.Hz,kar,ÜM(z), . . . . The signal processing unit 5 combines the selected estimated signal segments Sig.sub.Hz,kar,ÜM with the use of the detected heartbeat times H_Zp(1), H_Zp(2), . . . into the estimated cardiogenic signal Sig.sub.kar,est. The signal processing unit preferably interpolates two estimated signal segments located adjacent in time in the signal Sig.sub.kar,est in order to fill a gap.

    [0229] In a number of the versions just shown each sample element comprises a respective sum signal segment or a processed sum signal segment. Depending on the calculated value or calculated values of the transmission channel parameters used, the signal processing unit 5 combines in the training phase Tp the sample elements into classes. The signal processing unit 5 generates for each class a respective reference signal segment, e.g., the four reference signal segments Sig.sub.Hz,kar,LF.1, . . . , Sig.sub.Hz,kar,LF.4 or Sig.sub.Hz,kar,ÜM.a, . . . , SigSig.sub.Hz,kar,ÜM.d. Different processes are possible for combining the sum signal segments of a class of sample elements into a reference signal segment, which will then be stored in the library 12. FIG. 17 shows such a process as an example.

    [0230] Time, more precisely, a plurality of relative scanning times, are plotted om the x axis. “Relative” means relative to the beginning of the signal segment. The transmission channel parameter used or a transmission channel parameter used, in this example the R-R interval RR between the R peaks of two consecutive heartbeats, is plotted on the y axis. This process can just as well be used for other transmission channel parameters with two numbers as the parameter values and also for a plurality of transmission channel parameters. The value range of the transmission channel parameter plotted on the y axis is divided in this example into more than ten classes, and in the extreme case up to the accuracy of the machine, i.e., one class per number that can be displayed on the signal processing unit 5 used. The signal value, i.e., the value of the sum signal at this scanning time and at this transmission channel parameter value is plotted on the z axis. The sum signal segments of the sample elements were standardized in advance, so that the standardized sum signal segments Abs_std.x, Abs_std.y have all the same length and the R peaks have the same relative scanning time. These sum signal segments are represented one on top of another with the correct time in the view shown in FIG. 17. All R peaks are located at the relative scanning time T_R.

    [0231] The signal processing unit 5 calculates in the training phase Tp a fitted curve, which extends in the y-z plane, for each scanning time (x axis) by smoothing. This is illustrated in FIG. 17 for the relative scanning time T_R for the R peak. The signal values which the standardized sum signal segments assume at this scanning time T R yield a point cloud in the y-z plane at the x value T_R. The signal processing unit 5 generates by smoothing over this point cloud a fitted curve, e.g., the fitted curve Ak(T_R) for the scanning time T_R. This is carried out for each scanning time. As a result, a sequence of fitted curves is generated along the x axis. The signal processing unit 5 receives or calculates in the use phase Np the respective value of the transmission channel parameter or each transmission channel parameter for each detected heartbeat at this heartbeat. The transmission channel parameter is an R-R interval in the example shown in FIG. 17. The signal processing unit 5 determines the corresponding class, into which the transmission channel parameter value falls. Each possible transmission channel parameter value forms a class of its own in the extreme case (precision of the machine). The signal processing unit 5 determines for each relative scanning time in the course of this heartbeat the value which the fitted curve, which is associated with this relative scanning time, assumes in this class. This determination yields a signal value. The sequence of the signal values for this class and for the sequence of scanning times is used as the estimated signal segment for this detected heartbeat. Geometrically speaking, the corresponding class specifies a plane, which is at right angles to they axis. The points of intersections of the fitted curve with this perpendicular plane yield the estimated signal segment.

    [0232] FIG. 18 through FIG. 23 show another variant, in which the cardiogenic signal is determined from a sum signal and a wavelet transformation is applied.

    [0233] The time curve of the input signal E_Sig.sub.Sum, which is generated from electrical measured values of the measuring electrodes 2.1 and 2.2 and results from a superimposition of the heartbeat activity and the breathing activity of the patient P, is shown in the topmost row in FIG. 18. The measured value in mV is plotted on they axis. The sum signal Sig.sub.Sum can be generated from this by a corresponding measured value processing.

    [0234] In the row H_Zp under the input signal E_Sig.sub.Sum, the respective beginning, on the one hand, as well as the respective QRS segment of each heartbeat, for example, the beginning Anf_Zp(x) and the QRS segment H_Zp(x) of the xth heartbeat, are shown. The respective QRS segment acts in one embodiment as the characteristic heartbeat time.

    [0235] The sum signal Sig.sub.Sum is subjected to a wavelet transformation, and different frequency ranges are predefined. The wavelet transformation yields a respective signal component for each predefined frequency range. Three signal components A through C are calculated in the example being shown, and more than three signal components are preferably calculated. A respective other process, which will be described below, is carried out for each signal component A through C.

    [0236] The EMG power (power of the respiratory signal), which is illustrated in FIG. 18, is used as the transmission channel parameter for the signal component A. The influence of the cardiogenic signal Sig.sub.kar is compensated for this by calculation in the sum signal Sig.sub.Sum, for which purpose, for example, a standard signal segment (standard template) is used, which is valid for each heartbeat, or one of the variants described farther above is used. The compensation yields an estimated respiratory signal Sig.sub.res,est, which may still have a relatively great deviation from the actual respiratory signal Sig.sub.res. An envelope, which has exclusively positive signal values, is calculated from the estimated respiratory signal, for example, by calculation of the effective value (root mean square). For example, three classes EMG_Pow.sub.1 (low), EMG_Pow.sub.2 (medium) and EMG_Pow.sub.3 (high) of EMG powers are distinguished. The third row EMG_Pow shows in which segments the current EMG power belongs to which of these three classes.

    [0237] A respective limit each is determined for each class in the training phase, i.e., a total of three limits Max_Pow.sub.1 (for EMG_Pow.sub.1), Max_Pow.sub.2 (for EMG_Pow.sub.2) and Max_Pow.sub.3 (for EMG_Pow.sub.3) are determined. The row shows the application in the use phase. The cardiogenic component in the signal component A shall be determined. Designated by Sig.sub.Sum,A in the signal component A, the values whose respective absolute value is above the respective limit Max_Pow.sub.1, Max_Pow.sub.2, Max_Pow.sub.3 are used as values belonging to the cardiogenic component. Which threshold value it is depends on the current EMG power. The other signal values are set at zero by calculation.

    [0238] FIG. 19 shows the approach for the signal value B, which is designated by Sig.sub.Sum,B. The approach likewise uses the EMG power and differs from the approach for the signal component A as follows: Instead of forming a plurality of classes of EMG powers and then determining a limit for each class, a limit Max_Pow(t), which is variable over time, is calculated. To use the cardiogenic component in the signal component B, a signal value Sig.sub.Sum,B(t) above the limit Max_Pow(t) is used for this time t.

    [0239] FIG. 20 and FIG. 21 show an approach for the signal component C, which is designated by Sig.sub.Sum,C. The lung filling level LF is used as the transmission channel parameter. Three classes of lung filling levels, namely, LF.1, LF.2 and LF.3, are used in this example. The time course of the lung filling level and the respective class are shown in the upper row of FIG. 20. A respective smoothed envelope Sig.sub.Sum,LF.n is shown in the middle row of FIG. 20 for each heartbeat depending on the respective class LF.n.

    [0240] The signal power is calculated from the signal component, e.g., by calculating the effective value (root mean square). This calculation yields a time curve of the signal power. A respective power curve segment is calculated for each heartbeat. Depending on the lung filling level LF.1 or LF.2 or LF.3 at this heartbeat, a power curve segment Sig.sub.Hz,Pow.LF.1 or Sig.sub.Hz,Pow.LF.2 or Sig.sub.Hz,Pow.LF.3 is calculated hereby at this heartbeat.

    [0241] The power curve segments for a lung filling level class LF.12 or LF.2 or LF.3 are placed one on top of another with the correct time. The segments of one class placed one on top of another are combined, for example, averaged. As a result, a standard power curve segment is formed for each class. The three standard power curve segments Sig.sub.Hz,Pow,LF.1 and Sig.sub.Hz,Pow,LF.2 and Sig.sub.Hz,Pow,LF.3 calculated in this manner are shown in the lower row of FIG. 20. Three limits Max_Pow.LF.1, Max_Pow.LF.2 and Max_Pow.LF.3, which are variable over time, are calculated from these three standard power curve segments for the three classes LF.1, LF.2, LF.3. In one embodiment, the standard power curve segment of one class is scaled and bracketed, for example, by the median of the standard power curve segment being calculated: Median_Pow.LF.n=median(Sig.sub.Hz,Pow,LF.n).

    [0242] The limit Max_Pow.LF.n is then calculated depending on this median, for example, according to the formula


    Max Pow.LF.n=min(α*Median Pow.LF.n,β+γ*Median Pow LF.x/Sig.sub.Hz,Pow,LF.n).

    Where, α, β and γ are predefined constants, for example, α=6, β=0.01 and γ=0.05.

    [0243] These limits Max_Pow.LF.1, Max_Pow.LF.2 and Max_Pow.LF.3 are the result of the training phase Tp in this approach.

    [0244] Only the values of the signal component C that belong to the cardiogenic signal, which are above the limit for the respective lung filling level class, are used again in the use phase Np. FIG. 21 again shows, in the upper row, the three limits for the three classes of lung filling level. The signal component C, likewise designated by Sig.sub.Sum,C, is shown in the second row.

    [0245] Depending on the respective lung filling level class LF.1 or LF.2 or LF.3, the respective limit Max_Pow.LF.1 or Max_Pow.LF.2 or Max_Pow.LF.3 is entered.

    [0246] The respective cardiogenic component in the three signal components A, B and C are combined into an estimated cardiogenic signal Sig.sub.kar,est. This estimated cardiogenic signal Sig.sub.kar,est is shown in the third row. The difference from the sum signal Sig.sub.Sum and from the estimated cardiogenic signal Sig.sub.kar,est yields the estimated respiratory signal Sig.sub.res,est, which is shown in the fourth row.

    [0247] It is possible to use an additional transmission channel parameter, namely, the instantaneous EMG power, as this was explained for the signal component B in reference to FIG. 19.

    [0248] FIG. 22 (training phase) and FIG. 23 (use phase) show a variant of the process for the signal component C. The lung filling level LF is likewise used again as the transmission channel parameter, and three different classes LF.1, LF.2, LF.3 of lung filling levels are likewise distinguished. The time course of these classes LF.1, LF.2, LF.3 is illustrated in the topmost row of FIG. 22.

    [0249] Two characteristic heartbeat times, namely, the maximum value of the P peak and the maximum value of the QRS area, are detected for each heartbeat in the signal component C, likewise designated by Sig.sub.Sum,C. These terms were explained with reference to FIG. 7. Three maximum P values Max_P(x), Max_P(y) and Max_P(z) as well as three maximum QRS values Max_QRS(x), Max_QRS(y) and Max_QRS(z) for three heartbeats x, y, z are shown as an example in FIG. 22.

    [0250] Two histograms, namely, a histogram Hist_P for the maximum P values and a histogram Hist_QRS for the maximum QRS values, are calculated from these maximum values. The signal value is shown on the x axis and the percentage frequency on the y axis.

    [0251] Using these two histograms Hist_P and Hist_QRS, three limits, which are variable over time, are again calculated for the three classes LF.1, LF.2, LF.3. These limits are designated by Max_PQRS.LF.1, Max_PQRS.LF.2 and Max_PQRS.LF.3.

    [0252] A mean value Mean_QRS.LF.x for the class LF.n is calculated by averaging over all maxima Max_QRS(x) of the QRS segments of all heartbeats, which belong to the class LF.n, arithmetically or in another manner. A mean value Mean_P.LF.x, in which averaging is carried out over all maxima Max_P(x) of the P peaks of all heartbeats, which belong to the class LF.n, is correspondingly calculated for the class LF.n. These six mean values are shown in FIG. 22.

    [0253] A predefined limit is used at the beginning of the use phase Np. As soon as a sufficient number of heartbeats are detected, two different limits are used for each class LF.1, LF.2, LF.3, namely, [0254] a limit according to the calculation rule in the time range of the P-wave of a heartbeat


    α1−β1*Mean_P.LF.x and [0255] a limit according to the calculation rule


    α2−β2*Mean_QRS.LF.x [0256] in the time range of the QRS segment of a heartbeat.

    [0257] The four predefined constants have, for example, the values α1=0.05, β1=0.5, α2=0.025 and β2=0.05.

    [0258] FIG. 23 shows again how the three limits Max_PQRS.LF.1, Max_PQRS.LF.2 and Max_PQRS.LF.3, which are variable over time, are used in order to calculate the estimated cardiogenic signal Sig.sub.kar,est and then the estimated respiratory signal Sig.sub.res,est.

    [0259] While specific embodiments of the invention have been shown and described in detail to illustrate the application of the principles of the invention, it will be understood that the invention may be embodied otherwise without departing from such principles.

    LIST OF REFERENCE CHARACTERS

    [0260] 1 Ventilator; it assists the breathing activity of the patient P; it comprises the signal processing unit 5 [0261] 2.1 Set of measuring electrodes located close to the heart and at a distance from the diaphragm on the chest of the patient P; it acts as a set of sum signal sensors [0262] 2.2 Set of measuring electrodes located close to the heart and at a distance from the diaphragm on the abdomen of the patient P; it acts as a set of sum signal sensors [0263] 3 Pressure sensor in front of the mouth of the patient P; it acts as a set of sum signal sensors [0264] 4 Video camera, which is directed towards the thoracic area of the patient P; it generates the measured value series MWR [0265] 5 Signal processing unit; it generates from the sum signal; Sig.sub.Sum the estimated respiratory signal Sig.sub.res,est and/or the estimated cardiogenic signal Sig.sub.kar,est; it comprises the signal processor 13, the heartbeat time detector 7, the reconstructing unit 8 and the compensating unit 9 [0266] 6 Signal estimating unit; it yields the shape parameter value or each shape parameter value and the expected course Sig.sub.Hz,kar.LF of the cardiogenic signal or the expected course of the respiratory signal Sig.sub.Hz,res.LF in the course of a single heartbeat depending on the measured values of the transmission channel parameter value or of each transmission channel parameter value; it has reading access to the library 12 [0267] 7 Heartbeat time detector in the signal processing unit 5; it detects the respective time H_Zp(n) of each heartbeat [0268] 8 Reconstructing unit in the signal processing unit 5; it combines the estimated signal segments Sig.sub.Hz,kar into the reconstructed (estimated) cardiogenic signal Sig.sub.kar,est [0269] 9 Compensating unit; it compensates by calculation the influence of the respiratory signal Sig.sub.res on the sum signal Sig.sub.Sum [0270] 10 Mechanical sensor, which measures a value for the position Pos [0271] 11 Heartbeat time period detector; it measures the time period between the two characteristic times H_Zp(x), H_Zp(x+1) of two consecutive heartbeats and/or measures the respective heartbeat time period H_Zr(x), H_Zr(x+1) of each heartbeat [0272] 12 Library with an estimated signal segment Sig.sub.Hz,kar.LF per class, which describes the estimated cardiogenic signal Sig.sub.Hz,kar.LF.1, . . . in the course of a respective heartbeat each [0273] 13 Signal processor; it processes the electrical signals from the measuring electrodes 2.1 and 2.2 and/or from the pneumatic sensor 3 and/or from the optical sensor 4; it comprises an amplifier and an analog-digital converter; it performs a baseline removing in one embodiment [0274] 14 Sample with sample elements, which are classified according to the transmission channel parameter and comprise each a signal segment in the course of a heartbeat [0275] 16 Sensor in the esophagus Sp [0276] Abs.w, Abs.x, Segment of the sum signal Sig.sub.Sum in the course of the heartbeat with the [0277] Abs.y, Abs.z characteristic time H_Zp(w) and H_Zp(x) and H_Zp(y) and H_Zp(z) [0278] Abs_DT.w, Processed sum signal segment generated by detrending [0279] Abs_DT.x, [0280] Abs_DT.y [0281] Abs_std.x, Corrected signal segments for a respective heartbeat each; they are all of [0282] Abs std.y equal length and are aligned with the correct time [0283] Ak(T) Fitting curve for the relative scanning time T [0284] AM Breathing muscles of the patient P; they are a source of the respiratory signal Sig.sub.res [0285] Ap Adaptation phase, in which the signal estimating unit 6 is adapted to the current sample elements; overlapped with the use phase Np [0286] Atm.1, Atm.2, . . . Oscillations caused by the breathing activity of the patient P in the estimated respiratory signal Sig.sub.res,est [0287] FP-W(1), FP—Set of shape parameter values for a heartbeat [0288] W(2), . . . [0289] H_Zp(n) Time of the nth heartbeat (n=1, 2, . . . ) detected by the heartbeat time detector 7 [0290] Hist_P Histogram for the maxima of the P peaks [0291] Hist_QRS Histogram for the maxima of the QRS peaks [0292] HM Heart muscle of the patient P; it is the source of the cardiogenic signal Sig.sub.kar [0293] H_Zp(x) Characteristic heartbeat time of the xth heartbeat [0294] H_Zr(x) Heartbeat time period of the xth heartbeat [0295] LF Current lung filling level of the patient P; it is correlated with the volume flow Vol′; it is a transmission channel parameter [0296] LF.1, . . . , LF.4 Classes of lung filling levels, to which a respective reference signal segment Sig.sub.Hz,kar.LF.1, . . . , Sig.sub.Hz,kar.LF.4 each is assigned in one embodiment in the library 12 and to which a set of shape parameter values are assigned in another embodiment; each class is used to estimate the cardiogenic signal Sig.sub.Hz,kar.LF or the respiratory signal Sig.sub.Hz,res.LF in the course of an individual heartbeat [0297] LQ.a, LQ.b, LQ.c, Exemplary division into classes: It comprises three classes for the lung [0298] Q.d filling level LF and one class for the event that the exhalation time is before the Q wave [0299] Max.1, . . . Maximum, which occurs in the course of a heartbeat in the signal component Sig.sub.Sum,high for the higher frequency range [0300] MWR Measured value series with an image sequence, which is recorded by the video camera 4; it yields in one variant the sum signal used [0301] Max_MW.LF.1, Averaged maxima of all segments of the signal component Sigslim.hieh, [0302] Max_MW.LF.2 which belong to the lung filling level LF.1, LF.2, . . . ; stored in the library [0303] Max_P(x) Maximum of the P peak of the xth heartbeat [0304] Mean_P.LF.n Mean value over all maxima Max_P(x) of the heartbeats, at which the lung filling level belongs to class LF.n [0305] Max_Pow.LF.1, Limits for detecting in the signal component C (Sig.sub.Sum,C) the cardiogenic [0306] Max_Pow.LF.2, component; they are calculated in the use phase Np depending on the [0307] Max_Pow.LF.3 respective EMP power for the three classes LF.1, LF.2, LF.3 [0308] Max_PQRS.LF.1 Limits for the three classes LF.1, LF.2, LF.3 for detecting the cardiogenic [0309] Max_PQRS.LF.2 component in the signal component C (Sig.sub.Sum.C); they are calculated [0310] Max_PQRS.LF.3 during the use phase Np as a function of the two histograms Hist_QRS and Hist_P [0311] Max QRS(x) Maximum of the QRS segment of the xth heartbeat [0312] Mean_QRS.LF.n Mean value of the all maxima Max_QRS(x) of the heartbeats at which the lung filling level belongs to class LF.n [0313] Np Use phase; it follows the training phase Tp; it overlaps with the adaptation phase Ap [0314] Npf Useful path; it describes the steps and components during the use phase Np [0315] P Patient, whose intrinsic breathing activity is assisted by the ventilator 1; he is measured by the measuring electrodes 2.1 and 2.2, by the pneumatic sensor 3 and by the video camera 4 [0316] Pos Position of a measuring electrode 2.1, 2.2 relative to the heart of the patient P, measured by sensor 10; it acts as an additional transmission channel parameter [0317] Sig.sub.ges Overall signal for the breathing and ventilation of the patient P; it is generated by a superimposition of the intrinsic breathing activity of the patient P and the mechanical ventilation by the ventilator 1 [0318] Sig.sub.kar,est Reconstructed (estimated) cardiogenic signal, combined from the estimated cardiogenic signal segments Sig.sub.Hz,kar with the use of the heartbeat times H_Zp(n) [0319] Sig.sub.Hz,kar Estimated signal segment: Segment of the cardiogenic signal in the course of an individual heartbeat, provided by the signal estimating unit 6 [0320] Sig.sub.Hz,kar.LF Estimated cardiogenic signal segment; it is the segment of the estimated cardiogenic signal Sig.sub.kar,est in the course of a single heartbeat which is adapted to the current value LF.1, . . . , LF.4 of the transmission channel parameter or each transmission channel parameter (here: lung filling level LF); provided by the signal estimating unit 6 [0321] Sig.sub.Hz,kar,LF.1, . . . Cardiogenic reference signal segments stored in teeth library 12 for the [0322] Sig.sub.Hz,kar,LF.4 four classes LF.1, . . . , LF.4 of the lung filling level LF [0323] Sig.sub.Hz,kar,.Math.M.a, Cardiogenic reference signal segments stored in the library 12 for the four classes ÜM.a, . . . , ÜM.d of agreement values with the reference [0324] Sig.sub.Hz,kar,ÜM.d curves V.1, V.2 [0325] Sig.sub.Hz,kar,ÜM Estimated cardiogenic signal segment, provided by the signal estimating unit 6 as a function of the overall agreement value [0326] Sig.sub.kar Cardiogenic signal; it describes the cardiac activity of the patient P [0327] Sig.sub.kar,est Estimate for the cardiogenic signal Sig.sub.kar, generated by the signal processing unit 5 [0328] Sig.sub.Hz,Ref Predefined standard reference signal segment, average cardiogenic signal segment in the course of an individual heartbeat [0329] Sig.sub.res Respiratory signal; it describes the intrinsic breathing activity of the patient P [0330] Sig.sub.res,est Estimate generated by the signal processing unit 5 for the respiratory signal Sig.sub.res [0331] Sig.sub.Hz,res,LF Estimated respiratory signal segment, the segment of the estimated respiratory signal in the course of an individual heartbeat, which is adapted to the current value LF.1, . . . , LF.4 of the transmission channel parameter or of each transmission channel parameter (here: lung filling level LF, provided by the signal estimating unit 6 as a function of at least one transmission channel parameter value [0332] Sig.sub.Hz,res,LF.1, . . . , Respiratory reference signal segments stored in the library 12 for the four [0333] Sig.sub.Hz,res,LF.4 classes LF.1, LF.4 of the lung filling level LF [0334] Sig.sub.Sum Sum signal, measured by the sum signal sensors 2.1, 2.2, 3 or 4; it is a superimposition of the respiratory signal Sig.sub.res and of the cardiogenic signal Sig.sub.kar [0335] Sig.sub.Sum,high Component in the sum signal Sig.sub.Sum, which is in the higher frequency range [0336] Sig.sub.Sum,low Component in the sum signal Sig.sub.Sum, which is in the lower frequency range

    [0337] S_Q Signal, which describes another transmission channel parameter, namely, whether the exhalation by the patient begins shortly before the Q wave or not [0338] Sp Esophagus of the patient P [0339] Tnn Additional transmission channel for the cardiogenic signal Sig.sub.kar; it begins in the heart muscle [0340] Tns Transmission channel for the cardiogenic signal Sig.sub.kar; it leads from the heart muscle to the sensor 2.1, 2.2 [0341] Tss Transmission channel for the respiratory signal Sig.sub.res; it leads from the breathing muscles to the sensor 2.1, 2.2 [0342] Tp Training phase; it precedes the adaptation phase Ap [0343] Tpf Training path; it describes the steps and components during the training phase Tp and the subsequent adaptation phase Ap [0344] T_R Relative scanning time, on which the R peak falls [0345] ÜM.1, ÜM.2, . . . Overall agreement value; it depends on the agreement between a sum signal segment and a reference curve V.1, V.2 [0346] Vol′ Volume flow of breathing air into and out of the airway Aw; it is correlated with the lung filling level LF, it is a transmission channel parameter, which is correlated with an anthropological size (hcrc: lung filling level LF), which influences the transmission channel Tns [0347] V.1, . . . , V.3 Reference curves, generated by singular value decomposition (SVD) from the standardized sum signal segments Abs_std.x, Abs_std.y, . . . [0348] Zw Diaphragm of the patient P