US 20040167413 A1 Abstract Methods and apparatus for measuring arterial compliance using combined noninvasive arterial tonometry and cuff oscillometry. Some embodiments include a calibration method using an oscillometric signal to calibrate the pressures of tonometric signals in a contralateral arterial site. The times at which two of the three oscillometric blood pressures (systolic pressure, mean pressure, diastolic pressure) are acquired are identified with times of uncalibrated tonometric pressure waveform. These blood pressures are then used to calibrate the tonometric pressure waveform along (optionally) with adjustments for head pressure. For example, a left brachial arterial cuff oscillometric signal is acquired coincidentally with an uncalibrated right radial arterial pressure tonometric signal. The time points of mean arterial pressure and diastolic pressure are determined from the oscillometric signal and identified with coinciding time points on the tonometric signal to produce a calibration. All pressures are then adjusted by the head pressure between the brachial and radial sites.
Claims(43) 1. A method for computerized calculation of one or more arterial-compliance parameters of a patient, the method comprising:
measuring an oscillometric signal and a tonometric arterial signal of the patient; obtaining one or more oscillometric parameters derived from the oscillometric signal; obtaining a sequence of tonometric values that are based on the tonometric signal; receiving the one or more oscillometric parameters and the sequence of tonometric values as inputs into a computer system; calibrating the sequence of tonometric values based on the one or more oscillometric parameters to generate a calibrated tonometric pressure waveform; and processing the calibrated tonometric pressure waveform within the computer system to generate one or more values each corresponding to one of the one or more arterial-compliance parameters. 2. The method of P _{r}(t)=((S _{r}(t)+additive correction_{r})*multiplicative correction_{r}) wherein the additive correctionr and the multiplicative correction
_{r }are calibration constants based at least in part on blood pressure parameters derived from the oscillometric parameters, and each S_{r}(t) is the tonometric signal value at a time t. 3. The method of P _{r}(t)=((S _{r}(t)+additive correction_{r})*multiplicative correction_{r}) wherein the additive correction
_{r }and the multiplicative correction_{r }are calculated as follows: the multiplicative correction
_{r}=(DBP−MBP)/(S _{r}(t _{D})−S _{r}(t _{M})) the additive correction_{r} =MBP/(multiplicative correction_{r})−S _{r}(t _{M}) wherein
each S
_{r}(t) is the tonometric signal value at a time t, MBP is a mean arterial-blood-pressure oscillometric parameter measured near time t
_{M}, and DBP is a diastolic-blood-pressure oscillometric parameter measured near time t
_{D}. 4. The method of _{r}(t) as follows: P _{r}(t)=((S _{r}(t)+additive correction_{r})*multiplicative correction_{r}) wherein the additive correction
_{r }and the multiplicative correction_{r }are calculated as follows: the multiplicative correction
_{r}=(SBP−MBP)/(S _{r}(t _{S})−S_{r}(t _{M})) the additive correction_{r} =MBP/(multiplicative correction_{r})−S_{r}(t _{M}) wherein
each S
_{r}(t) is the tonometric signal value at a time t, MBP is oscillometric mean arterial blood pressure measured near time t
_{M}, and SBP is oscillometric systolic blood pressure measured near time t
_{S}. 5. The method of _{r}(t) as follows: P _{r}(t)=((S _{r}(t)+additive correction_{r})*multiplicative correction_{r}) wherein the additive correction
_{r }and the multiplicative correction_{r }are calculated as follows: the multiplicative correction
_{r}=(SBP−DBP)/(S _{r}(t _{S})−S _{r}(t _{D})), and the additive correction_{r} =DBP/(multiplicative correction_{r})−S _{r}(t _{D}) wherein
each S
_{r}(t) is the tonometric signal value at a time t, SBP is oscillometric systolic blood pressure measured near time t
_{S}, and DBP is oscillometric diastolic blood pressure measured near time t
_{D}. 6. The method of _{r}(t) includes generating the calibrated tonometric pressure waveform P_{r}(t) as follows: P _{r}(t)=((S _{r}(t)−b _{r})(1/a _{r}))+p where a
_{r }and b_{r }are calibration constants based at least in part on blood pressure parameters derived from the oscillometric signal, and p is a hydrostatic pressure head parameter constant. 7. The method of _{r}(t) includes calculating: a _{r}=(S _{r}(t _{D})−S _{r}(t _{M}))/(DBP−MBP), and b _{r} =S _{r}(t _{M})−a _{r} MBP, wherein MBP is oscillometric mean arterial blood pressure measured near time t
_{M}, and
DBP is oscillometric diastolic blood pressure measured near time t
_{D}. 8. The method of _{r}(t) includes calculating: a _{r}=(S _{r}(t _{S})−S _{r}(t _{M}))/(SBP−MBP), and b _{r} =S _{r}(t _{M})−a _{r} MBP), wherein MBP is oscillometric mean arterial blood pressure measured near time t
_{M}, and SBP is oscillometric systolic blood pressure measured near time t
_{S}. 9. The method of _{r}(t) includes calculating: a _{r}=(S _{r}(t _{S})−S _{r}(t _{D}))/(SBP−DBP), and b _{r} =S _{r}(t _{D})−a _{r } DBP, wherein SBP is oscillometric systolic blood pressure measured near time t
_{S}, and
DBP is oscillometric diastolic blood pressure measured near time t
_{D}. 10. The method of 11. The method of 12. The method of 13. The method of calculating a first compliance value based on the calibrated radial pressure waveform; estimating end-effects of the oscillometric signal; and correcting the first compliance value using the estimated end effects. 14. The method of 15. The method of using time points t _{M }and t_{S }from the sequence of values based on the tonometric signal, locating corresponding tonometric signal values shifted to the nearest peak (for t_{S}), nadir (for t_{D}), and calibrating using the formula P _{r}(t)=((S _{r}(t)+additive correction_{r})*multiplicative correction_{r}), using tonometric and oscillometric pressures, P and P _{c}, computing transmural pressure P_{TR}=P−P_{c }at each time point, using P _{c }and n_{c }computing V_{c}, numerically differentiating the data pairs (−V _{c}, P_{TR}) to obtain as a function of P _{TR}. 16. The method of plotting C(P _{TR}) and reporting C(SBP), C(DBP), C(120), C(80), and pressure at C_{max}. 17. The method of using a tonometric signal to calibrate oscillometric pressure. 18. The method of 19. The method of using a tonometric signal to calibrate oscillometric pressure signals in a contralateral arterial site. 20. The method of 21. A system for computerized calculation of one or more vascular-compliance parameters of a patient, the system comprising:
a first sensor that measures an oscillometric arterial signal; a second sensor that measures a tonometric arterial signal; a first analog-to-digital converter, operatively coupled to the first sensor, that generates a sequence of oscillometric values that are based on the oscillometric signal; a second analog-to-digital converter, operatively coupled to the second sensor, that generates a sequence of tonometric values that are based on the tonometric signal; a computer system, operatively coupled to the first and second analog-to-digital converters, wherein the computer system processes the first and second sequences of values and calibrates the sequence of tonometric values based on the one or more oscillometric parameters to generate one or more values each corresponding to one of the one or more vascular-compliance parameters. 22. The system of _{r}(t) to generate a calibrated tonometric pressure waveform P_{r}(t) as follows: P _{r}(t)=((S _{r}(t)+additive correction_{r})*multiplicative correction_{r}) wherein the additive correction
_{r }and the multiplicative correction_{r }are calibration constants based at least in part on blood pressure parameters derived from the oscillometric signal, and S_{r}(t) is the tonometric signal value at time t. 23. The system of the multiplicative correction
_{r}=(DBP−MBP)/(S _{r}(t _{D})−S _{r}(t _{M})), and the additive correction_{r} =a _{r} MBP−S _{r}(t_{M}), wherein MBP is oscillometric mean arterial blood pressure measured near time t
_{M}, and
DBP is oscillometric diastolic blood pressure measured near time t
_{D}. 24. The system of the multiplicative correction
_{r}=(SBP−MBP)/(S _{r}(t _{S})−S _{r}(t _{M})), and the additive correction_{r} =a _{r} MBP−S _{r}(t _{M}), wherein MBP is oscillometric mean arterial blood pressure measured near time t
_{M}, and SBP is oscillometric systolic blood pressure measured near time t
_{S}. 25. The system of the multiplicative correction
_{r}=(SBP−DBP)/(S _{r}(t _{S})−S _{r}(t _{D})), and the additive correction_{r} =a _{r} DBP−S _{r}(t _{D}), wherein SBP is oscillometric systolic blood pressure measured near time t
_{S}, and
DBP is oscillometric diastolic blood pressure measured near time t
_{D}. 26. The system of _{r}(t) to generate a calibrated tonometric pressure waveform P_{r}(t) as follows: P _{r}(t)=((S _{r}(t)−b _{r})(1/a _{r}))+p where a
_{r }and b_{r }are calibration constants based at least in part on blood pressure parameters derived from the oscillometric signal, and p is a hydrostatic pressure head parameter constant. 27. The system of a _{r}=(S _{r}(t _{D})−S _{r}(t _{M}))/(DBP−MBP), and b _{r} =S _{r}(t _{M})−a _{r} MBP, wherein MBP is oscillometric mean arterial blood pressure measured near time t
_{M}, and
DBP is oscillometric diastolic blood pressure measured near time t
_{D}. 28. The system of a _{r}=(S _{r}(t _{s})−S _{r}(t _{M}))/(SBP−MBP), and b _{r} =S _{r}(t _{M})−a _{r} MBP), wherein MBP is oscillometric mean arterial blood pressure measured near time t
_{M}, and SBP is oscillometric systolic blood pressure measured near time t
_{S}. 29. The system of a _{r}=(S _{r}(t _{S})−S _{r}(t _{D}))/(SBP−DBP), and b _{r}=S_{r}(t _{D})−a _{r} DBP, wherein SBP is oscillometric systolic blood pressure measured near time t_{S}, and DBP is oscillometric diastolic blood pressure measured near time t _{D}. 30. The system of 31. The system of 32. The system of 33. The system of 34. The system of 35. The system of uses time points t _{M }and t_{S }from the sequence of values based on the tonometric signal, and locates corresponding tonometric signal values shifted to the nearest peak (for t_{S}), nadir (for t_{D}), and calibrating using the formula P _{r}(t)=((S _{r}(t)+additive correction_{r})*multiplicative correction_{r}), uses tonometric and oscillometric pressures, P and P _{c}, to compute transmural pressure P_{TR}=P−P_{c }at each time point, uses P _{c }and n_{c }to compute V_{c}, and numerically differentiates the data pairs (−V _{c}, P_{TR}) to obtain as a function of P _{TR}. 36. The system of _{TR}) and reporting C(SBP), C(DBP), C(120), C(80), and pressure at C_{max}. 38. The system of 39. The system of 40. The method of 41. A system for computerized calculation of a vascular compliance parameter of a patient, the system comprising:
a first sensor that measures an oscillometric signal of the patient; a second sensor that measures a tonometric signal of the patient; means for calibrating the tonometric signal based on the oscillometric signal and for calculating a value for the vascular-compliance parameter. 42. The system of means for obtaining a time-correlated dual sequence of digital values that are based on the waveforms monitored by the first and second sensors; and means for processing the input signals to convert the time-correlated dual sequence of digital values to an output signal corresponding to a value of the vascular-compliance parameter. Description [0001] This is a divisional of patent application Ser. No. 10/211,857 filed Aug. 1, 2002. [0002] This invention relates to the field of medical diagnosis, and more specifically, to a method and apparatus for noninvasive measurement of arterial compliance. [0003] High blood pressure and hardening of the arteries can lead to coronary problems. Obtaining various measurements of the vascular system, including compliance of large and small vessels, and systemic resistance, provides physicians with information useful in diagnosing and treating early stages of coronary disease. [0004] U.S. Pat. No. 6,017,313 (“the '313 patent”) to Christopher W. Bratteli et al. (incorporated herein by reference) discloses an apparatus and method for blood pressure pulse waveform contour analysis. Methods of processing an arterial blood pressure waveform to extract clinically useful information on the state of the cardiovascular system are disclosed. In order to obtain the parameters of the modified Windkessel model, the diastolic portion of a subject's blood pressure waveform is scanned over a plurality of ranges and the range that produces the best fit of data and lowest error estimates are selected. [0005] U.S. Pat. No. 5,211,177 (“the '177 patent”) to Charles F. Chesney et al. (incorporated herein by reference) discloses method and apparatus for measuring properties of the human vasculature using an electrical analog model of vascular impedance. These properties include the compliance of large and small vessels, and systemic resistance. These measurements and others obtained from the model can, in turn, be used to diagnose states of health or disease, and to assess the effectiveness of treatment regimes. For example, see Finkelstein, S. M., Collins, V. R., Cohn, J. N., [0006] U.S. Pat. No. 6,159,166 (“the '166 patent”) to Charles F. Chesney et al., and entitled SENSOR AND METHOD FOR SENSING ARTERIAL PULSE PRESSURE (incorporated herein by reference) discloses a method and apparatus useful for tonometric measuring of an arterial pulse pressure waveform. U.S. Pat. No. 6,331,161 to Charles F. Chesney et al., and entitled METHOD AND APPARATUS FOR FABRICATING A PRESSURE-WAVE SENSOR WITH A LEVELING SUPPORT ELEMENT (incorporated herein by reference) discloses a method and apparatus of a sensor useful for tonometric measuring of an arterial pulse pressure waveform. U.S. Pat. No. 6,132,383 to Charles F. Chesney et al., and entitled APPARATUS AND METHOD FOR HOLDING AND POSITIONING AN ARTERIAL PULSE PRESSURE SENSOR (incorporated herein by reference) discloses a method and a wrist brace and sensor holder useful with any of the above sensors for tonometric measuring of an arterial pulse pressure waveform. [0007] U.S. Pat. No. 5,241,966 (“the '966 patent”) to Stanley M. Finkelstein; et al. issued Sep. 7, 1993 (incorporated herein by reference) discloses an apparatus for measuring stroke volume/cardiac output that includes a transducer for measuring arterial blood pressure waveform, a digitizer for digitizing the analog signal generated by the transducer and a digital signal processor for determining ejection time and heart rate. Processor circuitry determines cardiac output using the ejection time, heart rate, the body surface area and age of the patient. [0008] Compliance is a fundamental property of any pressure-volume system such as the arteries. In the arteries, reduced compliance has been offered as a mechanism whereby the work load on the heart is increased and myocardial perfusion is decreased, thus leading to cardiac disease. It is also thought that disease of the arteries themselves may be evaluated by measurement of arterial compliance. Therefore, arterial compliance is an important cardiovascular parameter. [0009] What is needed is improved noninvasive measurement of arterial compliance. [0010] The present invention provides improved noninvasive measurement of arterial compliance using a combination of noninvasive arterial tonometry and noninvasive cuff oscillometry. Described are a number of improved approaches to estimating systemic vascular resistance and/or compliance. In one embodiment, a computer-controlled pneumatic cuff having a sensor coupled to sense pressure and pressure variations in the cuff is used to obtain oscillometric signals (e.g., from an upper arm of a patient), and simultaneously a contact-pressure sensor (e.g., some type of microphone) is used to obtain tonometric signals (e.g., from the contralateral wrist of the patient and, e.g., a transducer placed on the radial artery), and the oscillometric signals and tonometric signals, which are correlated to one another in time, are combined to calibrate one another and/or obtain information about the patient not available conventionally. In some embodiments, the oscillometric measurement is used to calibrate the tonometric measurement. [0011] One aspect of the present invention provides a method to compute arterial compliance from oscillometric data (2 [0012] Another aspect of the present invention provides a method for improving pressure calibration from oscillometric data (1 [0013] Yet another aspect of the present invention provides a set of methods for building models that compute flow from pressure data (with or without oscillometric data). This can be used in the same way as the cardiac output model described in the '313 patent (i.e., it can be used to improve the compliance estimate). [0014] The present invention provides a number of improvements or refinements to the approaches to waveform analysis set forth in the '177 and the '313 patents. These improvements include an accurate calibration method and apparatus for using an oscillometric signal to calibrate the pressures of tonometric signals in the contralateral arterial site. These improvements also include a simple uncorrected volume arterial compliance curve that is obtained by plotting relative arterial volume under the cuff against brachial arterial transmural pressure. Further improvements account for a bias (i.e., an overestimation of the in vivo compliance curve for the arterial segment due to shear stresses at the ends of the cuff) by providing a correction for the transmural pressure that is based on stress-strain properties of the upper arm. [0015] In some embodiments, the method of the '313 patent is improved with use of the SVR model of the present invention. [0016]FIG. 1A illustrates an example of a vascular-compliance-determining device [0017]FIG. 1B illustrates schematically details of one embodiment of oscillometric sensor [0018]FIG. 1C illustrates schematically details of one embodiment of cuff [0019]FIG. 2A illustrates an oscillometric arterial blood pressure waveform and points of interest therein. [0020]FIG. 2B illustrates a tonometric arterial blood pressure waveform and points of interest therein. [0021]FIG. 3 is an example display of compliance measurements for analysis according to the present invention. [0022]FIG. 4A is a graph of shear force. [0023]FIG. 4B is a schematic diagram of the forces on an arterial cross section. [0024]FIG. 4C is a schematic diagram of the forces on an arterial longitudinal section. [0025]FIG. 4D is a schematic diagram of the forces on an arterial cross section in an arm. [0026]FIG. 5 is an overview of an example process for waveform analysis according to the present invention. [0027]FIG. 6 is a more detailed flowchart of an example process for pulse contour analysis according to the present invention. [0028] In the following detailed description of the preferred embodiments, reference is made to the accompanying drawings that form a part hereof, and in which are shown by way of illustration specific embodiments in which the invention may be practiced. It is understood that other embodiments may be utilized and structural changes may be made without departing from the scope of the present invention. [0029]FIG. 1A illustrates one embodiment of vascular-compliance-determining system [0030]FIG. 1B illustrates details of one embodiment of oscillometric sensor [0031]FIG. 1C illustrates schematically details of one embodiment of cuff [0032]FIG. 2A illustrates an oscillometric arterial blood pressure waveform [0033]FIG. 2B illustrates a tonometric arterial blood pressure waveform and points of interest therein. [0034]FIG. 3 is an example display of compliance measurements for analysis according to the present invention. [0035]FIGS. 4A, 4B, [0036]FIG. 5 is an overview of an example process for waveform analysis according to the present invention. Method [0037]FIG. 6 is a more detailed flowchart of an example process for pulse contour analysis according to the present invention. This method includes [0038] (a) identifying the physiological parameter to be quantitatively monitored and estimated; [0039] (b) measuring an oscillometric signal and a tonometric physiological signal, which signals are quantitatively dependent on a particular value for the physiological parameter; [0040] (c) obtaining a sequence of values that are based on the oscillometric signal and the tonometric signal; [0041] (d) receiving the sequence of values as input signals to a computer system; and [0042] (e) processing the input signals within the computer system to convert the sequence of values to an output signal corresponding to the particular value of the physiological parameter. [0043] (f) using an oscillometric signal to calibrate tonometric pressure signals in a contralateral arterial site. [0044] In some embodiments, a calibrated radial pressure waveform P [0045] where a [0046] b [0047] p=gh are calibration factors, and where [0048] =density of blood, [0049] g=acceleration to gravity, [0050] h=height difference between the oscillometric and the tonometric measurement sites, and is zero if the patient is supine, [0051] MBP is oscillometric mean arterial blood pressure measured at time t [0052] DBP is oscillometric diastolic blood pressure measured at time t [0053] (g) calculating a first compliance value based on the calibrated radial pressure waveform; [0054] (h) estimating end-effects of the oscillometric signal; and [0055] (i) correcting the first compliance value using the estimated end effects. [0056] As used herein, arterial compliance is defined as the absolute increase in arterial lumen volume for a given increase in arterial lumen transmural pressure, or Compliance= [0057] The static pressure-volume relationship in the arterial lumen is the result of a balance of forces. Forces working to reduce lumen size are atmospheric pressure around the arm, tensions in the tissues surrounding the artery, and tensions in the walls of the artery. Forces working to open the artery are lumen pressure, tensions in the arterial wall working to maintain structure and shape, tethering forces between the walls of the artery and the tissues surrounding the artery, and shear forces that occur when the expansion or compression of one part of the artery pulls on another part. In vivo, the pressure wave travels at many meters per second, and so an artery has the appearance of increasing and decreasing simultaneously along its entire length. This, along with the narrow pressure range and relatively flat compliance range that occurs under physiological conditions, means that shear forces play a negligible role. However, when a short segment of artery is compressed underneath a blood pressure cuff, shear stresses occur in the arterial walls beneath both ends of the cuff. The effects of these shear forces are for the portions of the artery away from the cuff to pull open the portions just beneath the cuff ends. Thus, even under complete occlusion by the cuff, when the cuff is far above systolic pressure, pressure pulsations from the heart are able to expand the tapered region beneath the proximal end of the cuff because the systolic pressure is augmented by shear stresses. The result is that oscillometric signals persist at super-systolic cuff pressures. Using a blood pressure cuff to measure arterial compliance thus produces compliance estimates that are too high because they include shear stresses that are not present without the cuff. Without cuff compression, compliance is primarily a function of arterial-wall stresses and anatomical dimensions. [0058] Cuff oscillometry is currently a widespread technique for obtaining the values of one, two, or three blood pressure points during the cardiac cycle. The technique is most commonly applied to the brachial artery. In this technique, arterial occlusion is obtained by placing a fluid-filled (usually air-filled) compliant chamber such as a cuff or patch around an artery. The pressure in the fluid is mechanically manipulated and measured. Pressure pulsations in the artery create volume displacements which are captured by the cuff. When the volume of the cuff changes, the pressure of the cuff fluid changes. The pressure is measured with a pressure transducer. As the amount of fluid in the cuff is varied between a point producing complete occlusion of the artery, and a point where the cuff and artery become decoupled, a characteristic oscillometric pressure waveform is recorded from the cuff fluid. At high cuff pressures (near arterial occlusion), the artery is in a region of low compliance with the rigid cuff limiting the distensibility of the artery, and the physiologic pressure pulsations produce low amplitude oscillometric signals. As the transmural pressure in the artery approaches the region of maximal compliance, high-amplitude oscillometric signals are produced. As transmural pressures approach their physiologic values, another low compliance region is reached. At this point, the low pressures in the cuff make the cuff very compliant. The combination of low arterial compliance and high cuff compliance again produces low-amplitude oscillometric signals. Pattern-recognition techniques are then used to estimate one or more of the following: systolic pressure, diastolic pressure, and mean pressure. Mean arterial pressure is usually estimated as equal to the cuff pressure near where the oscillations are a maximum. [0059] Cuff oscillometry has been used to estimate elastic properties of the brachial artery such as compliance (Brinton et al., 1997; Brinton et al., 1996; Chio et al., 1989; Chio et al., 1998; Shimazu et al, 1989; see reference list below). The theoretical basis for this capability is that the necessary volume changes are obtained continuously from the oscillometric signal while pressure changes are obtained by identifying patterns within segments of the signal. Unfortunately, continuous pressures are not obtained. One way this may be overcome is to assume a parametric model for the volume pressure relationship and then fit this model using the few pressures that can be obtained from the cuff. Most recent efforts have been to obtain better arterial pressure estimates from the cuff pressure, and an empirical compliance curve is assumed. See U.S. Pat. No. 5,579,778 issued Dec. 3, 1996, entitled “Method and apparatus for producing thermodilution cardiac output measurements utilizing a neural network” by P. D. Baker et al., and U.S. Pat. No. 5,339,818 issued Aug. 23, 1994, entitled “Method for determining blood pressure utilizing a neural network” by P. D. Baker et al. [0060] The present invention shows how the few known pressure points from the oscillometric signal can be coupled with a tonometric signal to estimate a person's compliance curve at both physiological and sub-physiological pressures. The latter may be used, for instance, to determine the amount of blood-pressure reduction needed in an individual to reduce their compliance to normal levels. [0061] A balance of forces on the artery produces the following relationship {SEE Derivations 3}:
[0062] where P is lumen pressure, P [0063] Combining these relations one gets
[0064] Under in vivo conditions, or with a very long cuff, P [0065] where A is lumen cross-sectional area, s is the position along a line through the center of the lumen under the cuff, and p
[0066] In a First Embodiment of the invention, both oscillometric and tonometric signals are measured (in one embodiment, both signals are measured substantially simultaneously to obtain a series of value-pairs in order to obtain values that are correlated to one another), and a calibration is performed. In one such embodiment, the oscillometric signal is measured by a pressure sensor coupled to a pneumatic cuff whose average (i.e., constant, or DC) pressure is manipulated by the control computer (wherein the pressure sensor measures the time-varying (i.e., AC) portion), and the tonometric signal is obtained by a microphone sensor and/or sensor holder and wrist brace as described in the above mentioned U.S. patent application Ser. No. 09/045,018, Ser. No. 09/394,301, and Ser. No. 09/045,449 (each of which is incorporated herein by reference). In some embodiments, the tonometric signal is measured at the contralateral radial artery site (e.g., the left wrist) from the cuff measurement site (e.g., the upper right arm). For the Radial Signal (Sr) as measured,
[0067] where P [0068] In some embodiments, two pressure points are needed for calibration. In one such embodiment, MBP (mean arterial blood pressure) and DBP (diastolic blood pressure) are chosen, and respective corresponding times tM and tD (shifted to the appropriate time within a cardiac cycle, e.g., tD is shifted to the nearest pulse minimum) are used: [0069] The radial artery is approximately six inches below the brachial artery. This creates a hydrostatic pressure head that can be accommodated by a further pressure head correction factor of p=gh, where [0070] =density of blood (=1.03 g/cm [0071] g=acceleration to gravity (=980 cm/sec [0072] h=6″15 cm [0073] thus p=1.03 g/cm [0074] 11 mm Hg if sitting (or 0 if supine) [0075] and the calibrated radial pressure waveform P [0076] In another such embodiment, the two pressure points chosen for calibration are MBP (mean arterial blood pressure) and SBP (systolic blood pressure), and respective corresponding times t [0077] In yet another such embodiment, the two pressure points chosen for calibration are SBP (systolic blood pressure) and DBP (diastolic blood pressure), and respective corresponding times ts and tD are used: [0078] In a Second Embodiment of the invention, a first compliance value (herein also called “an uncorrected compliance value”) is calculated as follows:
P _{TR} ΔP−P_{c }
[0079] In some embodiments, it is assumed that the cuff is filled with an ideal gas. Temperature can be eliminated from the ideal gas state equation by assuming adiabatic conditions. According to other researchers (for example, Shimazu, Kawarada):
[0080] (Ideal gas law, before inflation)
[0081] V [0082] R=universal gas constant [0083] =8.314 in SI units [0084] T [0085] P [0086] The computer system controls a cuff/pump system that inflates and deflates the cuff. To keep track of the amount of air in the cuff (n [0087] (inflation)
[0088] (deflation) n _{c}(t)=n _{o} +kt
[0089] (inflation) n _{c}(t)=n _{o}+(n _{max} −n _{o})e ^{−rt }
[0090] (deflation) [0091] Constant r can be obtained for a particular oscillometric device as
[0092] time to 99% complete deflation. [0093] In some of its embodiments, the Second Embodiment uses a method employing the following procedures: [0094] 1. Apply cuff oscillometric system to left upper arm [0095] 2. Apply tonometer system to right radial artery. [0096] 3. Simultaneously, collect oscillometric and tonometer signals. [0097] 4. Using the oscillometric algorithm, identify MBP and SBP from oscillometric signal. These pressure points coincide with time points tM and ts, respectively. [0098] The process [0099] Next, the data is filtered ( [0100] After the marks have been identified, the digitized waveform data is converted to pressure in mm-Hg for analysis using a two-point calibration. In one embodiment, the median systolic and median upstroke values are used together with the systolic and diastolic measurements of blood pressure obtained from the oscillometric cuff pressure measurement unit [0101] 1. Using time points from procedure [0102] 2. Using tonometric and oscillometric pressures, P and P [0103] 3. Using P [0104] 4. Numerically differentiate the data pairs (−V [0105] as a function of P [0106] 5. Plot C(P [0107] Compliance Range: {C(DBP)} to {C(SBP)} [0108] Compliance Range at Normalized Pressure: {C(80)} to {C(120)} [0109] Mean Compliance:
[0110] For example, with SBP=120, DBP=80->
[0111] Maximum C: {C [0112] P=arterial lumen pressure [0113] P [0114] σ [0115] σ [0116] σ [0117] σ [0118] r=lumen radius [0119] h [0120] h [0121] A [0122] The following is Derivations 1-6. [0123] Please refer to drawings of FIGS. 4A, 4B, [0124] The present invention addresses estimating the systemic vascular resistance element of the 3 [0125] Some Basic Fluid Dynamic Preliminaries for the Present Invention: [0126] Two independent state variables are used to determine the properties of an arterial system-pressure and flow. The relationships between these variables are quantities that are frequently of interest when describing a person's arterial system. For example, the ratio of flow to the time rate of pressure change is arterial compliance, the steady-state ratio of mean pressure to mean flow gives vascular resistance, and the pressure rate of change and flow rate of change are related through the inertance. Because blood flows in an axial direction, it is often useful to express flow as the product of cross-sectional area and blood velocity. [0127] Because pressure and flow are both needed to describe system properties, and since it is the unknown dependence between these two quantities that is sought by the present invention for each individual who is assessed, a measurement must be independently made of both quantities. In the 3 [0128] In developing a model for cardiac output for use in the 3 [0129] Before detailing the models, it is useful to recall a deficiency in the 3 [0130] A mass balance through the system leads to the following equations: [0131] where the i's are flows through respective elements. After some minor algebraic manipulations we obtain the following 3 [0132] where D is the differential operator with respect to time. Setting i [0133] To obtain an estimate of R, we again use equation (6). Throughout we have assumed that R is constant, so integrating both sides we obtain ∫P [0134] where the integrals are over a full cardiac cycle. The reason this expression for R is likely to be useful is that ∫ [0135] Heart rate (HR) we can easily obtain from the pressure wave, so we can choose to estimate either stroke volume, or cardiac output. [0136] The Cohn/Finkelstein model for stroke volume is a direct linear parametric model. Here we offer three additional categories of models—an indirect linear parametric model, an indirect nonlinear nonparametric model, and a direct nonlinear nonparametric model. [0137] Indirect Linear Parametric Model [0138] We call this approach linear parametric because it uses ARMAX (AutoRegressive Moving Average with eXogenous input) models [1]. It is indirect because instead of estimating stroke volume directly, ascending aortic cross-sectional area and blood velocity are modeled and combined to estimate stroke volume. [0139] Two ARMAX models are used, one to compute a transfer function from radial arterial pressure to ascending aortic velocity, another to estimate ascending aortic cross-sectional area from radial arterial pressure. A brief review of ARMAX modeling is next presented. [0140] An ARMAX model for y(t) as a function of x(t) is given by [0141] where w is the unknown error, and the vectors a, b, and c are the model parameters. This equation can be rewritten in matrix form as [0142] where the matrix F consists of columns of shifted vectors of y and x, a contains the ‘a’ and ‘b’ parameters, and β contains the ‘c’ parameters. A common way of obtaining estimates of the a, b, and c parameters is to use an iterative two-step ordinary least squares (OLS) algorithm followed by a generalized least squares (GLS) algorithm until convergence is reached [2]. [0143] Taking the Z-transform of the above expression for y[n] and rearranging, we obtain [0144] where B is a polynomial in z [0145] Estimating y[n] by its expected values (i.e. neglecting error) gives the desired result [0146] ARMAX modeling is straightforward except for the selection of the proper order (P, Q, R). A common way of comparing models of different orders is with a bias-adjusted maximum likelihood estimate such as AIC. Several candidate models are fit and the one with the lowest AIC is chosen. [0147] Allen & Murray used ARMAX models to predict photoelectric phlethysmographic blood volumes from isovolumic calibrated photoelectric finger pressures [3]. They used a first-order two-parameter autoregressive (ARX) model (P, Q, R)=(1,1,0) with b[0]=0. The ability of these low-order models to predict volumes from pressures was modest, but suggests that reasonable models might be obtainable with only a few parameters. Allen & Murray related finger volumes to finger pressures, while our needs are to relate aortic variables to radial arterial pressure. Thus the transfer functions that we compute implicitly combine a local pressure to velocity (or area) transfer function with an inter-positional transfer function (aorta to radial artery). [0148] We obtain stroke volume as [0149] where A(t) is ascending aortic cross-sectional area as a function of time, v(t) is ascending aortic blood velocity as a function of time, A[n] and v[n] are the corresponding sampled data values, t is the sampling interval, and N is the number of data points in a single cardiac cycle. Here the integral is approximated with a simple sum, but any appropriate numerical integration could be used to obtain higher precision (e.g. a high-order Newton-Cotes). A[n] and v[n] are obtained as [0150] where P [0151] The following procedure is therefore proposed for any particular patient group of interest: [0152] 1. Using a Doppler ultrasound (US) probe, obtain real-time ascending aortic cross-sectional areas and blood velocities while simultaneously collecting radial arterial tonometric pressures. Obtain an oscillometric cuff mean and diastolic blood pressure reading immediately before the collection. Obtain several cardiac cycles of data for each of several human subjects. [0153] 2. For each subject, estimate the transfer function parameters (a's, b's, and c's) using the two-step ARMAX algorithm described above and selecting the model with the lowest AIC. Obtain both the pressure to velocity transfer function and the pressure to area transfer functions. [0154] 3. Average the transfer function parameters for the several beats and for all individuals. [0155] 4. Using the averaged transfer function parameters, compute the two transfer functions (hv[n], and hA[n]). [0156] 5. In all fLiture windkessel analyses, estimate SV using Equations (15), (16), and (17). [0157] Indirect Nonlinear Nonparametric Model [0158] This method is called indirect because, again, area and velocity are estimated from pressure. It is called nonlinear nonparametric because a neural network is used. This method is similar to the indirect linear parametric model, with two main differences. First, we train a neural network rather than an ARMAX model. Second, because a neural network is being used, additional parameters are easily incorporated. Similar to the ARMAX model, we need to decide upon a model order (Q). In addition, we need to decide which additional variables to include. As with the present PulseWaves implementation, we use gender, age, height, and weight. Data are collected with Doppler ultrasound and arterial tonometry as described above. Any network activation type can be used, but for demonstration purposes, we will describe a logistic two-layer multiple-layered perceptron (MLP) model. The number of inputs is equal to 4+Q+1. There are two outputs (velocity and area). The number of nodes between the two middle layers must be high enough to describe the input-output map, but not so high as to make training or prediction computationally impractical. We always use the same number of nodes in both layers. The number of nodes is between one and two times the number of inputs. Training proceeds with cross-validation until leave-one-out cross validation sum of squared residuals reaches a minimum (rather than until actual sum of squared residuals reaches a minimum). It is because we are unconcerned with the parameter values of the neural network that this approach is called “nonparametric”. [0159] Any nonlinear parameter estimation algorithm may be used, e.g. conjugate gradients, backpropagation, Newton or quasi-Newton methods. Training stops at minimum leave-one-out cross-validation sum-of-squared residuals. The cross-validation sum-of-squares is also used to compare models of different orders. For each beat, we train the two outputs (v[n] and A[n]) to the corresponding pressure, P [0160] The model can be written as ( [0161] Therefore the following procedure is used, in one embodiment, for a particular group of interest: [0162] 1. Collect data as in the indirect linear parametric model above. Because of the generality of neural network models, it may be necessary to obtain data from [0163] 2. Using the method described above, train and select a neural network configuration that predicts v[n] and A[n] from P [0164] 3. In all future windkessel analyses, estimate SV using Equations (15) and (18). [0165] Direct Nonlinear Nonparametric Model [0166] This method is called direct because SV is computed directly. It is called nonlinear nonparametric because a neural network model is employed. [0167] This approach is similar to the indirect nonlinear nonparametric model except that the neural network output is SV, and a fixed set of pressure values are used as inputs. The model is expressed in the following way: [0168] Unlike with the indirect nonlinear approach, this approach is dependent upon the start and end times used for the pressure inputs. We use the foot of each beat for the first pressure point. Various end-points can be tried to obtain the lowest prediction error. The spacing between samples can also be adjusted to obtain simpler models with fewer inputs. Training should otherwise proceed as described above for the indirect nonlinear nonparametric model. [0169] The following procedure is used, in one embodiment, for a particular group of interest: [0170] 1. Using a Fick dilution method, estimate several cardiac outputs for an individual. Simultaneously, collect tonometric pressure recordings at the radial artery that are calibrated to brachial oscillometric cuff blood pressure estimates. [0171] 2. Using the method described above, train and select a neural network configuration that predicts cardiac output from P [0172] 3. In future windkessel analyses, estimate cardiac output using Equation (19). [0173] Conclusion [0174] One aspect of the present invention provides a method for computerized calculation of a variable physiological parameter of a patient. The first method includes (a) identifying the physiological parameter to be quantitatively monitored and estimated, (b) measuring an oscillometric signal and a tonometric physiological signal, which signals are quantitatively dependent on a particular value for the physiological parameter, (c) obtaining a sequence of values that are based on the oscillometric signal and the tonometric signal, (d) receiving the sequence of values as input signals to a computer system, and (e) processing the input signals within the computer system to convert the sequence of values to an output signal corresponding to the particular value of the physiological parameter. [0175] In some embodiments, the first method further includes (f) using an oscillometric signal to calibrate tonometric pressure signals in a contralateral arterial site. In some embodiments, a calibrated radial pressure waveform P [0176] where a [0177] b [0178] p=gh are calibration factors, and where [0179] =density of blood, [0180] g=acceleration to gravity, [0181] h=height difference between the oscillometric and the tonometric measurement sites, and is zero if the patient is supine, [0182] MBP is oscillometric mean arterial blood pressure measured at time t [0183] In some embodiments, the first method further includes (g) calculating a first compliance value based on the calibrated radial pressure waveform, (h) estimating end-effects of the oscillometric signal, and (i) correcting the first compliance value using the estimated end effects. [0184] In some embodiments, the processing function (e) includes estimating a first compliance value using a compliance pressure curve. Some embodiments of the first method further include [0185] Another aspect of the present invention provides a method for computerized calculation of a variable physiological parameter of a patient. This second method includes(a) identifying the physiological parameter to be quantitatively monitored and estimated, (b) coupling at least a first sensor and a second sensor to the patient, the first sensor and the second sensor each being responsive to monitor different time-varying physiological waveforms, which waveforms are quantitatively dependent on a particular value for the physiological parameter, (c) obtaining a time-correlated dual sequence of digital values that are based on the waveforms monitored by the first and second sensors, (d) receiving the sequence of digital values as input signals to a computer system, and (e) processing the input signals within the computer system to convert the time-correlated dual sequence of digital values to an output signal corresponding to a value of the physiological parameter. [0186] In some embodiments of the second method, the physiological parameter is vascular compliance, the first sensor monitors an oscillometric waveform, and the second sensor monitors a tonometric waveform. [0187] In some embodiments of the second method, the physiological parameter is vascular compliance, the first sensor monitors an oscillometric waveform derived from oscillometric signals of a brachial artery site, and the second sensor monitors a tonometric waveform derived from tonometric signals of a contralateral radial artery site. [0188] Yet another aspect of the present invention provides a system for computerized calculation of a variable physiological parameter of a patient. The system includes a first and a second sensor. The first sensor measures an oscillometric physiological signal that is quantitatively dependent on a particular value for the physiological parameter. The second sensor measures a tonometric physiological signal that is quantitatively dependent on the particular value for the physiological parameter. A first analog-to-digital converter, operatively coupled to the first sensor, generates a first sequence of digital values that are based on the oscillometric signal. A second analog-to-digital converter, operatively coupled to the second sensor, generates a second sequence of digital values that are based on the tonometric signal. A computer system, operatively coupled to the first and second analog-to-digital converters, processes the first and second sequences of values to generate an output signal corresponding to the particular value of the physiological parameter. [0189] In some embodiments, the first sensor senses the oscillometric signal from one side of a patient, the second sensor senses the tonometric signal from a contralateral arterial site, and the computer uses the oscillometric signal to calibrate tonometric pressure signals in the contralateral arterial site. [0190] In some embodiments, the computer derives a calibrated radial pressure waveform P [0191] where a [0192] b [0193] p=gh are calibration factors, and where [0194] =density of blood, [0195] g=acceleration to gravity, [0196] h=height difference between the oscillometric and the tonometric measurement sites, and is zero if the patient is supine, [0197] MBP is oscillometric mean arterial blood pressure measured at time t [0198] DBP is oscillometric diastolic blood pressure measured at time t [0199] In some embodiments, the computer system further calculates a first compliance value based on the calibrated radial pressure waveform, estimates end-effects of the oscillometric signal, and corrects the first compliance value using the estimated end effects. [0200] In some embodiments, the computer system further estimates a first compliance value using a compliance pressure curve. [0201] In some embodiments, the computer system further uses a tonometric signal to calibrate oscillometric pressure signals in a contralateral arterial site. [0202] In some embodiments, the computer system further estimates end-effects of oscillometric sensor apparatus on the oscillometric signal. [0203] A. Brinton, T. J., Cotter, B., Kailasam, M. T., Brown, D. L., Chio, S. S., O'Connor, D. T., MeMaria, A. N., [0204] B. Brinton, T. J., Kailasam, M. T., Wu, R. A., Cervenka, J. H., Chio, S. S., Parmer, R. J., DeMaria, A. N., O'Connor, D. T., [0205] C. Chio, S. S., U.S. Pat. No. 4,880,013 (1989). [0206] D. Chio, S. S., U.S. Pat. No. 5,836,884 (1998). [0207] E. Shimazu, H., Kawarada, A., Ito, H., [0208] Another aspect of the invention is to improve the accuracy of the estimate of stroke volume using one of three methods. A first method includes measuring a relevant population of persons to determine empirically the hA and hv parameters for Equations 16 and 17 respectively, and then using Equations 16 and 17 to derive arterial cross section area A[n] and velocity v[n] respectively. These A[n] and v[n] parameters are then input into Equation 15 to determine an improved estimate of stroke volume. A second method further obtains age, height, weight and sex information for each person in the population used to obtain data to empirically obtain weighting factors, and Equation 18 is then used to obtain the A[n] and v[n] parameters, which are then input into Equation 15 to determine an improved estimate of stroke volume. A third method also obtains age, height, weight and sex information for each person in the population used to obtain data to empirically obtain weighting factors, and Equation 19 is then used to directly obtain an improved estimate of stroke volume. These improved estimates of stroke volume are then used in the methods described above to obtain improved estimates of compliance values. [0209] On the following pages are explanations of the derivations 1-6 described above. [0210] The following is Derivation 1: [0211] Please refer to drawings of FIGS. 4A and 4B. [0212] Here we derive a general expression for a shear force.
[0213] The following is Derivation 2: [0214] Using Derivation 1, we derive expressions for the rate of change of shear force along the length of the artery. We do so for both the wall and tissue layers. We also define the geometric coefficients, Ψ.
[0215] If r [0216] The following is Derivation 3: [0217] Using Derivation 2 and FIGS. 4C and 4D, we write a fotce balance, and derive the relationship between various pressues in te arn and caff. [0218] Force Balance:
[0219] where [0220] F=Pr [0221] F [0222] F [0223] δlF [0224] and x is eiher t or w. [0225] As δl→0,
P+P _{sh} =P _{c} +P _{c,ex} =P _{t }
[0226] where
[0227] μ [0228] The following is Derivation 4: [0229] Here we comnputc simplified expressions for the terms that appear in the geometric coefticients in Dervation 2.
[0230] The followixng is Derivation 5: [0231] Here we uses Derivations 2 and 4 to estimate the value of the geonmelic coefficients based on the assumptions that the ratio h [0232] Ψ [0233] When we measure
[0234] the volume change is due not only to P V _{tot} =∫A(x,t)dx,V _{TR} =LA(0,t)
[0235] where L is the cuff length. [0236] Letting r [0237] We can find a model for A(x) by solving the nonlinear ODE from Derivation 3.
[0238] Assume that wall and tissue properties are sirnilar and use Derivation 5. Also use μ [0239] Substituting α [0240] Solving for A(x) and centering around x [0241] Next we evaluate each of the terms in &kappa:
[0242] Substituting back into κ; we get
[0243] To estimate r [0244] In one embodiment, the Third Embodiment uses the following procedure: [0245] 1. Calculate P, P [0246] 2. Calculate r [0247] 3. Calculate β using P [0248] 4. Calculate S by numerically differentiating logβ [0249] 5. Calculate D [0250] 6. Calculate κ from D [0251] 7. Calculate C [0252] It is to be understood that the above description is intended to be illustrative, and not restrictive. Many other embodiments will be apparent to those of skill in the art upon reviewing the above description. The scope of the invention should, therefore, be determined with reference to the appended claims, along with the full scope of equivalents to which such claims are entitled. Referenced by
Classifications
Rotate |