Method and apparatus for non-invasive blood constituent monitoring

[0002] This application claims benefit of Provisional application 60/073,784, filed Feb. 5, 1998.

[0003] The present invention is related to U.S. Pat. Nos. 5,372,136 and 5,499,627 the text and drawings of which are incorporated herein by reference as if reproduced in full.

[0004] The present invention relates to improvements in the systems and methods for non-invasively measuring one or more biologic constituent concentration values. More particularly, the present invention relates to non-invasive spectrophotometric systems and methods for quantitatively and continuously monitoring the hematocrit and other blood parameters.

[0005] Modern medical practice utilizes a number of procedures and indicators to assess a patient's condition. One of these indicators is the patient's hematocrit. Hematocrit (often abbreviated as HCT) is the volume expressed as a percentage of the patient's blood which is occupied by red corpuscles, commonly referred to as red blood cells. The present invention is presented in the context of hematocrit. However, it is to be understood that the teachings of the present invention apply to any desired biologic constituent parameter.

[0006] Medical professionals routinely desire to know the hematocrit of a patient. In order to determine hematocrit using any of the techniques available to date, it is necessary to draw a sample of blood by puncturing a vein or invading a capillary. Then, using widely accepted techniques, the sample of blood is subjected to either high-speed centrifuge, cell counting, ultrasonic, conductometric or photometric methods of evaluating the sample of blood in a fixed container. Prior U.S. Pat. No. 5,372,136 indicates a system and methodology for determining the hematocrit non-invasively, without puncturing or invading the body, spectrophotometrically and continuously in a subject. The present invention relates to improvements upon the above cited system.

[0007] Beyond the above referenced patent, others have suggested various means of noninvasive measurement of hematocrit. Specifically, Mendelson, U.S. Pat. No. 5,277,181; Seeker, U.S. Pat. No. 5,188,108; Gonatas, U.S. Pat. No. 5,528,365; Ishikawa, U.S. Pat. No. 5,522,388; Shiga, U.S. Pat. No. 4,927,264; Tsuchiya, U.S. Pat. Nos. 5,441,054, 5,529,065, 5,517,987 and 5,477,051; and Chance, U.S. Pat. Nos. 5,353,799, 5,402,778, and 5,673,701 have attempted to define means of directly measuring desired biologic constituents such as hematocrit. Even though the various patents indicate the need to utilize multiple wavelengths measured at different detection sites and/or the need to perform differential or ratiometric operations on the detected optical signal, all fail to isolate and resolve the individual and specific scattering and absorption coefficients of the desired constituent. At best they address only bulk attenuation coefficients and/or bulk diffusion constants of the scattering media while-attempting to resolve such constraints as tissue nonhomogeneity. As an example, tissue may be considered to contain a bulk absorptive coefficient due to blood, collagen, water, fibers, bone, fingernail, etc. Hence, in order to determine the absorptive coefficient of the blood itself, the bulk value of the tissue per se must be prorated by the amounts of the above constituents. Secondly, the actual absorptive coefficient of the blood must then be decoupled or isolated from its proration factor as well.

[0008] Thus, it is an object of the present invention to provide an improvement in the systems and methods for the non-invasive (transcutaneous) and continuous determination of the blood Hematocrit in living tissue.

[0009] It is yet another object of the present invention to provide an improvement in the systems and methods for the non-invasive (transcutaneous) and continuous determination of the blood constituents, including glucose, bilirubin, cholesterol, tissue water, etc. in living tissue.

[0010] It is another object of the present invention to provide a system and method and apparatus for the display of both immediate and/or continuous visual information regarding the HCT of the subject.

[0011] It is yet another object of the present invention to provide a repeatable and reliable method and apparatus for the non-invasive determination of hematocrit transcutaneously and in real time even under varying physiological conditions.

[0012] Still another object of the present invention is to provide a method and apparatus for the instantaneous determination of the bulk absorption coefficient of the scattering media.

[0013] These and other objects and advantages of the invention will become more fully apparent from the description in the specification and claims, which follow.

[0014] In one aspect, the present invention accomplishes the transcutaneous, noninvasive, real-time and continuous measurement of the hematocrit and other blood constituents of the patient. That is, the electronic circuitry necessary is included to receive signals from a detector and to generate appropriate signals at various input sites as described in U.S. Pat. No. 5,372,136. Yet another aspect of the present invention is the ability to extract the blood absorption coefficient from the bulk tissue diffusion constant or the bulk absorption coefficient of the scattering media by requiring both physical and mathematical operations.

[0015] FIGS. 1 and 1A show a finger placed into a clam-shell type fixture constituting a receiving means for detector and emitter arrays operating in a transmission mode and the blood conduit which in the figures is the finger.

[0016] FIGS. 1B and 1C are similar to FIG. 1A, but show the detector and emitter arrays operating in a reflectance mode.

[0017] FIG. 1D is a schematic diagram for a mylar base with a detector, emitters and either a strain gage or a pressure transducer for inclusion in the clam-shell fixture.

[0018] FIG. 1E is a schematic diagram for a detector emitter array using a single, moveable emitter in a transmission mode.

[0019] FIG. 2 shows actual patient data plot of ln(i) vs. d.

[0020] FIG. 3 illustrates actual patient data of the (∂i/∂t)/i dependence on d.

[0021] FIG. 4 shows the blood absorption coefficient's dependence on hematocrit.

[0022] FIG. 5 indicates the nonlinear relationship between the V_c/V_𝑓 and pressure.

[0023] FIG. 6 shows the electrical circuit diagram of the piezo film/strain gage transducer means.

[0024] FIG. 7 is the plot of 𝑓(H) vs. measured Hematocrit.

[0025] FIG. 8 shows the instantaneous time derivatives of (∂i/∂t)/i and ∂P/∂t versus time during one cardiac pulse.

[0026] FIG. 9 plots (∂i/∂t)/i versus ∂P/∂t for a given human pulse at d₁, d₂, d₃, and d₄,

[0027] FIG. 10 plots (∂i/∂t)/(∂P/∂t) versus time during a single cardiac pulse cycle.

[0028] FIG. 11 is the circuit diagram of the pressure transducer means.

[0029] FIG. 12 plots α versus X_bat a fixed Hematocrit.

[0030] FIG. 13 gives the patient data of the new transcutaneous Hematocrit method and system plotted versus the measured Hematocrit standard.

[0031] In a preferred embodiment of the invention, measurements are conducted using a modified version of the apparatus described in U.S. Pat. Nos. 5,456,253 and 5,372,136, both of which are incorporated herein as if reproduced in full below. Both of these patents form part of the present disclosure.

[0032] Thus, in a preferred embodiment, hematocrit is measured in living tissue located at some convenient location on the body, such as, an ear lobe, finger tip, nose or other accessible tissue sites. In a preferred embodiment the apparatus and signal manipulations described in U.S. Pat. No. 5,372,136 are utilized to measure various optical parameters that will be described hereafter. The numbered components in FIGS. 1, 1A, 1B, and 1C are similar to the numbers in FIG. 1 of U.S. Pat. No. 5,456,253.

[0033] In the present disclosure, FIG. 1 shows the finger 7 of an individual placed into a clam-shell type fixture 6 wherein the optical and other physical measurements can be easily performed. The clam-shell type holder allows for adaptability to various finger sizes. However, other fixture methods such as FIGS. 1B through 1E, can be used to obtain similar physical data as using the clam-shell fixture.

[0034] Non-invasive, transcutaneous hematocrit measurement using a spectroscopic method is described below:

[0035] I. Introduction

[0036] Earlier spectrophotometric techniques have fallen short of being able to fully characterize the individual blood absorbance coefficients. The following discussion demonstrates the method of decoupling, or isolating from the bulk tissue attenuation parameters (including the convoluted absorptive and scattering parameters) the individual blood absorptive constants. This unique method identifies, isolates and compartmentalizes each of the contributing biologic elements of the tissue media. This decoupling process can either isolate the blood absorbance of interest and/or eliminate the scattering contribution from the bulk media measurement.

[0037] From photon diffusion analysis: $\begin{array}{cc}\frac{{\partial }^2}{\partial {\rho }^2}\Psi \left(\rho \right)-a^2\Psi \left(\rho \right)=-\frac{S\left(\rho \right)}{D}& \left(1\right)\end{array}$

[0038] where, $\begin{array}{cc}D=\frac{1}{3\left(K+S\right)}& \left(2\right)\\ \alpha =\sqrt{3K\left(K+S\right)}& \left(3\right)\\ K=K_b{X}_b+K_s{X}_s+K_w{X}_w& \left(4\right)\\ K_b=\frac{H}{V}\left({\sigma }_{\mathrm{ao}}\mathrm{SAT}+\left(1-\mathrm{SAT}\right){\sigma }_{\mathrm{ar}}\right)+\left(1-H\right)K_p& \left(5\right)\\ S=S_b{X}_b+S_s{X}_s& \left(6\right)\\ S_b=\frac{{\sigma }_{s}H\left(1-H\right)\left(1.4-H\right)}{V}& \left(7\right)\end{array}$

[0039] and where,

[0040] II. Analysis

[0041] The light flux, or intensity, i, is given by $i=D\frac{\partial \Psi }{\partial \rho }.$

[0042] When evaluated at ρ=d, one solution to equation (1) is: $\begin{array}{cc}i=A\frac{e^{\alpha d}}{e^{2\alpha d}-1}& \left(8\right)\end{array}$

[0043] where A is a nontrivial function of the tissue scattering coefficient, S, the distance, d (if small), and the bulk attenuation coefficient, α. If α d>>1, then (8) becomes:

[0044] i=Ae^−αd  (9)

[0045] where $A\approx \frac{\alpha }{\left[d^n·\left(1-e^{-2\alpha d}\right)\right]}\mathrm{or}\left(1/d^2+1/\alpha d\right)\mathrm{for}0<n<2,$

[0046] where n is the power that d is raised to.

[0047] FIG. (2) shows the actual patient data plot of ln(i) vs. d, where α is determined directly from the slope of the line.

[0048] The attenuation coefficient, α, is a bulk term which encompasses the attenuation measurement sensitivity to variations in skin color, presence of bone, callous, blood and water content, etc. In addition, a expresses the optical “path lengthening” effects of both the absorption and scattering characteristics of the tissue. Therefore, since α is a function of HCT and the intensity of the transmitted light can be measured, the HCT can be calculated by manipulation of the preceding relationships.

[0049] Beginning with equation (9), the troublesome and complex tissue function, A, can be eliminated by taking the logarithm of (9) and differentiating with respect to the distance, d. Unfortunately the term X_bis not known but changes with time as a result of a patient's cardiac cycle. Therefore, by differentiating with respect to time, this parameter becomes the time rate of change of blood volume which can be obtained through several methods described below. These time and distance derivatives may be performed in either order.

[0050] [1] Taking the logarithm of (9) and differentiating with respect to the distance, d, yields: $\begin{array}{cc}\alpha =\frac{\partial \left[\ln \left(i\right)\right]}{\partial d}& \left(10\right)\end{array}$

[0051] Next the derivative of (10) with respect to time, t, gives: $\begin{array}{cc}\frac{\partial \alpha }{\partial t}=\frac{\partial \left(\frac{\partial \left[\ln \left(i\right)\right]}{\partial d}\right)}{\partial t}& \left(11\right)\end{array}$

[0052] [2] Alternatively, first differentiate (9) with respect to time, t, to get: $\begin{array}{cc}\frac{\partial i}{\partial t}=\frac{\partial i}{\partial X_b}\frac{\partial X_b}{\partial t}+\frac{\partial i}{\partial X_s}\frac{\partial X_s}{\partial t}+\frac{\partial i}{\partial X_w}\frac{\partial X_w}{\partial t}& \left(12\right)\end{array}$

[0053] When $\frac{\partial i}{\partial X_s}\frac{\partial X_s}{\partial t}$

[0054] and $\frac{\partial i}{\partial X_w}\frac{\partial X_w}{\partial t}$

[0055] are negligible, and normalizing (12) by i yields: $\begin{array}{cc}\frac{\partial i/\partial t}{i}=\frac{\partial X_b}{\partial t}\left(\frac{\partial \alpha }{\partial X_b}d-\frac{1}{A}\frac{\partial A}{\partial X_b}\right)\mathrm{or},& \left(13\right)\\ \frac{\partial i/\partial t}{i}=\frac{\partial \alpha }{\partial t}\left(d-d_o\right),\mathrm{where}{d}_o\approx \frac{1}{\alpha }-\frac{2d}{e^{2\alpha d}-1}& \text{(13a)}\end{array}$

[0056] FIG. 3 plainly demonstrates the offset term when the various graph lines are extrapolated to d=0. The amount of offset is shown along the y-axis.

[0057] Next differentiate (13) with respect to distance, d, to eliminate that offset term to get: $\begin{array}{cc}\frac{\partial \left(\frac{\partial i/\partial t}{i}\right)}{\partial d}=\frac{\partial X_b}{\partial t}\left(\frac{\partial \alpha }{\partial X_b}\right)=\frac{\partial \alpha }{\partial t}& \left(14\right)\end{array}$

[0058] Equations (3)-(7) are now used to extract the hematocrit from α. Squaring (3) and differentiating with respect to time results in: $\begin{array}{cc}2\alpha \frac{\partial \alpha }{\partial t}=3\left[\frac{\partial K}{\partial t}\left(2K+S\right)+K\frac{\partial S}{\partial t}\right]& \left(15\right)\end{array}$

[0059] Substituting the derivatives of (4) and (6) into (15) and rearranging: $\begin{array}{cc}\begin{array}{c}\frac{\partial \alpha }{\partial t}=\frac{3}{2\alpha }[\left(\frac{\partial X_b}{\partial t}{K}_b+\frac{\partial X_s}{\partial t}{K}_s+\frac{\partial X_w}{\partial t}{K}_w\right)\left(2K+S\right)+\\ K\left(\frac{\partial X_b}{\partial t}{S}_b+\frac{\partial X_s}{\partial t}{S}_s\right)]\end{array}& \left(16\right)\end{array}$

[0060] At 805 nm, $\frac{\partial X_b}{\partial t}{K}_b⪢\frac{\partial X_s}{\partial t}{K}_s,\frac{\partial X_b}{\partial t}{K}_b⪢\frac{\partial X_w}{\partial t}{K}_w,\frac{\partial X_b}{\partial t}{S}_b⪢\frac{\partial X_s}{\partial t}{S}_s$

[0061] and K <<S so that (16) can be simplified to: $\begin{array}{cc}\frac{\partial \alpha }{\partial t}=\frac{3}{2\alpha }\frac{\partial X_b}{\partial t}\left(K_{b}S+{\mathrm{KS}}_b\right)& \left(17\right)\end{array}$

[0062] By using the 805 nm wavelength the red blood cell absorption cross-section constants are equal, σ_αo=σ_ar, and K_pis negligible. The hematocrit can then be determined directly from K_bas (5) simplifies to: $\begin{array}{cc}H=\frac{V}{{\sigma }_a}{K}_b& \text{(17a)}\end{array}$

[0063] FIG. 4 shows the linearity of K_b(H).

[0064] If K_bS>>KS_b, where S is approximately 1.0/mm in human tissue, then solving (17) for K_band substituting into (17a) gives: $\begin{array}{cc}H=\frac{\frac{2V}{3{\sigma }_a}\alpha \frac{\partial \alpha }{\partial t}}{\frac{\partial X_b}{\partial t}S}& \left(18\right)\end{array}$

[0065] To rewrite in terms of measurable intensity, i, (10) and (14) are substituted into (18) to obtain: $\begin{array}{cc}H=\frac{\frac{2V}{3{\sigma }_a}-\frac{\partial \left[\ln \left(i\right)\right]}{\partial d}\frac{\partial \left[\left(\partial i/\partial t\right)/i\right]}{\partial d}}{\frac{\partial X_b}{\partial t}S}& \left(19\right)\end{array}$

[0066] If K_bS is not>>KS_b, then substituting (5) and (7) into (17a) and rearranging terms yields: $\begin{array}{cc}H=\frac{2\alpha }{3}\frac{\partial \alpha }{\partial t}/\frac{\partial X_b}{\partial t}\left[S\frac{{\sigma }_a}{V}+K\frac{{\sigma }_s}{V}\left(1-H\right)\left(1.4-H\right)\right]& \text{(18a)}\end{array}$

[0067] Alternatively from (13a) $\begin{array}{cc}H\approx \frac{\alpha ·\left(\partial i/\partial t/i\right)}{\left(d-d_o\right)·X_b^{\prime }}& \text{(18b)}\end{array}$

[0068] Equation (18a) indicates a small nonlinearity in H may occur based on the magnitude of K for a given individual.

[0069] It should be reiterated that the change in received intensity with time is a result of the change in normalized blood volume resulting from the cardiac cycle itself as blood pulses through the examined tissue. As the intensity of the received light is measured, its time rate of change can be calculated. The change with distance can be determined by placing multiple emitters (such as 1-4 in FIG. 1 A) and/or multiple detectors such that multiple thicknesses of tissue and hence, lengths of tissue are penetrated.

[0070] To examine $\frac{\partial X_b}{\partial t}$

[0071] further, the following can be defined for the illuminated tissue:

[0072] V_b=Volume of blood,

[0073] V_w=Volume of water, and

[0074] V_s=Volume of skin, tissue and other non-water or blood components.

[0075] By definition, $\begin{array}{cc}{X}_b=\frac{V_b}{V_b+V_w+V_s}& \left(20\right)\end{array}$

[0076] differentiating (20) with respect to time gives: $\begin{array}{cc}\frac{\partial X_b}{\partial t}=\frac{\left(V_w+V_s\right)\frac{\partial V_b}{\partial t}-V_b\frac{\partial V_w}{\partial t}}{{\left(V_b+V_w+V_s\right)}^2}& \left(21\right)\end{array}$

[0077] Since $\frac{\partial \mathrm{Vw}}{\partial t}⪡\frac{\partial \mathrm{Vb}}{\partial t}$

[0078] and V_b<<V_w+V_s, (21) simplifies to: $\begin{array}{cc}\frac{\partial X_b}{\partial t}=\frac{\frac{\partial V_b}{\partial t}}{V_{\mathrm{total}}}& \left(22\right)\end{array}$

[0079] It is emphasized that α is a function of the bulk absorption and scattering coefficients, K and S, as well as hematocrit, H. Further, that K and S are functions of the fractional volumes of each constituent, X_b, X_s, and X_w, which must be used to prorate the individual absorption and scattering coefficients, K_b, K_s, K_w, S_band S_s. Therefore, the transducer system must be responsive not only to a change in volume (ΔV) due to the influx of the blood, but must also be responsive to the normalized change in volume of blood, normalized to the total volume of the finger (V_𝑓) or tissue being measured, $\left(\frac{\Delta V_f}{V_f}\right).$

[0080] For Reflectance (R) measurements in homogenous tissue:

[0081] R=Ae^−αr, where A≈(1/r²+1/αr), where r is the radial distance, and $\frac{\mathrm{dR}}{R}={\alpha }^{\prime }\left(r/\left({\alpha }^2+\alpha r\right)\right)$

[0082] However, for tissue, which is typically non-homogeneous with a dermal and subcutaneous layer, the reflectance will not be a trivial function but can be described as approximately:

[0083] R=[(C₁+C₂)exp(−C₃·r)]/rⁿ

[0084] Where C₁and C₂are inter-related photon flux densities between the dermal layer 12 and the subcutaneous layer, 12a(see FIGS. 1C and 1E). Likewise, C₃is a strong function of z₁, z₂, α₁, and α₂; i.e., the thickness of the dermis or dermal layer 12, subcutaneous layer 12a, and their respective α's.

[0085] Since C₃′ is a function of the inter-related photon flux densities C₁and/or C₂and if Xb′₁does not equal Xb′₂, then the slope C₃′ will not be nulled out by the Xb′ monitors mentioned. Therefore, Xb₂′ must be greater than Xb₁′. Then the pressure or piezo monitors will compensate correctly. The circular pressure balloon is ideal for not only sensing the change in a pressure, but also providing, a pressure against the dermis causing Xb₁′ to be small. However, recognizing that the penetration depth of the 800 nm light typically extends through dermal layer 12 into the deep tissue, subcutaneous layer 12a, a different wavelength selection is appropriate. Thusly, when the photons only penetrate into the dermal layer 12, C₃′ will only be a function of z₁and α₁. Those selected wavelengths, as mentioned in U.S. Pat. No. 5,372,136, would be the green (570-595 nm) wavelength and 1300 nm wavelength. The green wavelengths are used as the hematocrit bearing wavelength and the 1300 nm wavelength is used as the non-hematocrit bearing, or reference wavelength. That is, for reflectance measurements the green (Gr)-1300 wavelength pair would give the hematocrit information as: $\frac{\Delta \mathrm{Gr}/\mathrm{Gr}}{\Delta 1300/1300}·\frac{{\alpha }_{\mathrm{Gr}}}{{\alpha }_{1300}}=f\left(\mathrm{HCT}\right)$

[0086] III. Methods of $\frac{\partial X_b}{\partial t}$

[0087] measurement

[0088] ∂X_b/∂t can be measured and compensated for through the use of a number of different methods—(a) a pressure transducer, (b) a strain transducer such as piezo electric film or strain gage, (c) a different wavelength of light, such as 1300 nm, which also holds ∂X_b/∂t information, but holds little hematocrit information, or (d) other transducers. The individual methods of obtaining ∂X_b/∂t are addressed below.

[0089] A. Pressure Transducer Measurement of $\frac{\partial X_b}{\partial t}$

[0090] Consider a pressure transducer system 36 with a gas filled bladder 38 surrounding a finger tip 10 of a patient contained within a fixed volume clam shell fixture 6, see FIGS. 1, 1A-1D. The same derivations, equations, and results would apply to any other body appendage or tissue that could be contacted such that a change in the tissue volume would change the pressure of the contacted pressure transducer system. For a finger note:

[0091] V_clam=V_sys+V_𝑓  (23)

[0092] where

[0093] V_clam=Clam-shell fixture volume

[0094] V_sys=Bladder system volume

[0095] V_𝑓=Finger volume

[0096] Also ΔV_𝑓=−ΔV_sys. The system will have a bulk modulus of elasticity, β, such that: $\begin{array}{cc}\frac{\Delta V_{\mathrm{sys}}}{V_{\mathrm{sys}}}=-\frac{\Delta P_{\mathrm{sys}}}{\beta }=-\frac{\Delta V_f}{V_{\mathrm{sys}}}& \left(24\right)\end{array}$

[0097] Substituting (23) into (24) results in: $\begin{array}{cc}\frac{\Delta V_f}{V_f}=\left(\frac{V_{\mathrm{clam}}}{V_f}-1\right)\frac{\Delta P_{\mathrm{sys}}}{\beta }& \left(25\right)\end{array}$

[0098] Since ΔV_𝑓ΔV_bthen from (25) we have: $\begin{array}{cc}\frac{\partial X_b}{\partial t}=\left(\frac{V_{\mathrm{clam}}}{V_f}-1\right)\frac{\Delta P_{\mathrm{sys}}}{\beta }& \text{(25a)}\end{array}$

[0099] As stated above, β is a constant of the pressure transducer system. However, an empirical solution for $\left(\frac{V_{\mathrm{clam}}}{V_f}-1\right)$

[0100] was found to have a nonlinear relation to the pressure of the transducer system. For a given clam shell—pressure transducer embodiment a polynomial, F(p), can accurately describe $\left(\frac{V_{\mathrm{clam}}}{V_f}-1\right),$

[0101] see FIG. 5

[0102] B. Strain Transducer (Strain Gage/Piezo Electric Film) Measurement of $\frac{\partial X_b}{\partial t}$

[0103] Again it is assumed that ΔV_b=ΔV_𝑓, and that the finger changes volume only by a change in diameter. A strain gage or piezo electric film is secured tightly around the finger (again any applicable body appendage or tissue would apply) such that a change in diameter would produce a strain in the transducer. Specifically assuming a cylindrical finger: $\begin{array}{cc}\frac{\partial V_b}{\partial t}=\frac{\partial V_f}{\partial t}=\frac{\partial \left(\pi {\mathrm{zr}}^2\right)}{\partial t}=2\pi \mathrm{zr}\frac{\partial r}{\partial t}& \left(26\right)\end{array}$

[0104] Normalizing with respect to V_𝑓, yields: $\begin{array}{cc}\frac{\partial X_b}{\partial t}=-\frac{\frac{\partial V_b}{\partial t}}{V_{\mathrm{total}}}=\frac{2\pi \mathrm{zr}\frac{\partial r}{\partial t}}{\pi {\mathrm{zr}}^2}=\frac{2}{r}\frac{\partial r}{\partial t}& \left(27\right)\end{array}$

[0105] A change in the length of the transducer element is related to a change in finger radius by ΔL=2πΔr, therefore: $\begin{array}{cc}\frac{\partial X_b}{\partial t}=\frac{2\left(\partial L/\partial t\right)}{L_t}=2\gamma \left(t\right)& \left(28\right)\end{array}$

[0106] where $\gamma \left(t\right)=\frac{\partial L/\partial t}{L}$

[0107] is the rate of change in the strain as a function of time. For a strain gage this value can be measured from an appropriate electrical circuit, see FIG. 6, as it is proportional to the rate of change in the gage resistance.

[0108] For a piezo electric film the voltage produced is proportional to the strain, therefore: $\begin{array}{cc}\frac{\partial \mathrm{Xb}}{\partial t}=\frac{2}{g_{31}\tau }\frac{\partial v\left(t\right)}{\partial t}& \left(29\right)\end{array}$

[0109] where, g₃₁is the piezoelectric coefficient for the stretch axis, τ is the film thickness and v(t) is the open-circuit output voltage.

[0110] C. 1300 nm Light Measurement of $\frac{\partial X_b}{\partial t}$

[0111] The selection of the 1300 nm wavelength is based on criteria established in U.S. Pat. No. 5,372,136. The approach here is not to solve for ∂X_b/∂₁and substitute into (19) but to ratiometrically eliminate ∂X_b/∂t. In the case of the 1300 nm reference wavelength, the assumptions following equation (12) are no longer valid; i.e., ΔX_s/Δt and ∂X_w/∂t are not negligible, since water absorption at 1300 nm is so large. Hence, for the 1300 nm equations (13), (14) and (15) would result in: $\begin{array}{cc}\begin{array}{c}{\left(\frac{\partial \alpha }{\partial t}\right)}_{13}=\frac{3}{2\alpha }[\left\{\left(2K+S\right)K_b+{\mathrm{KS}}_b\right\}\frac{\partial X_b}{\partial t}+\\ \left\{\left(2K+S\right)K_s+{\mathrm{KS}}_s\right\}\frac{\partial X_s}{\partial t}+\\ \left\{\left(2K+S\right)K_w\right\}\frac{\partial X_w}{\partial t}]\end{array}& \left(30\right)\end{array}$

[0112] where, α, and the bulk and material specific K, and S are wavelength (λ) dependent. Recalling that, X_b+X_s+X_w=1, by definition, and that: $\begin{array}{cc}-\frac{\partial X_b}{\partial t}-\frac{\partial X_s}{\partial t}=\frac{\partial X_w}{\partial t}& \left(31\right)\end{array}$

[0113] By substituting (31) into (30) and noting that K_w/3≈K_b/3, the following is obtained: $\begin{array}{cc}{\left(\frac{\partial \alpha }{\partial t}\right)}_{13}=\frac{3}{2\alpha }\left[\left\{{\mathrm{KS}}_b\right\}\frac{\partial X_b}{\partial t}+\left\{\left(2K+S\right)\left[K_s-K_w\right]+K_s{S}_s\right\}\frac{\partial X_s}{\partial t}\right]& \left(32\right)\end{array}$

[0114] Since $\frac{\partial X_b}{\partial t}⪢\frac{\partial X_s}{\partial t},$

[0115] (32) becomes: $\begin{array}{cc}{\left(\frac{\partial \alpha }{\partial t}\right)}_{13}=\frac{3}{2{\alpha }_{13}}{\left\{{\mathrm{KS}}_b\right\}}_{13}\frac{\partial X_b}{\partial t}& \left(33\right)\end{array}$

[0116] Therefore, to eliminate $\frac{\partial \mathrm{Xb}}{\partial t}$

[0117] and solve for the hematocrit, (17) is divided by (33) yielding: $\begin{array}{cc}\frac{{\left(\partial \alpha /\partial t\right)}_8}{{\left(\partial \alpha /\partial t\right)}_{13}}=\frac{{\alpha }_{13}}{{\alpha }_8}\frac{K_{b_8}{S}_8}{K_{13}{S}_{b_{13}}}& \left(34\right)\end{array}$

[0118] Since S₈and K₁₃are well behaved and known (let K₁₃/S₈=G) in human tissue and the ratio $\frac{K_{\mathrm{b8}}}{S_{\mathrm{b13}}}$

[0119] is a function of H, then rearranging (34) gives: $\begin{array}{cc}f\left(H\right)=\frac{K_{b_8}^{\prime }}{S_{b_{13}}}=\frac{{\alpha }_8}{{\alpha }_{13}}\frac{{\left(\frac{\partial \alpha }{\partial t}\right)}_8}{{\left(\frac{\partial \alpha }{\partial t}\right)}_{13}}G& \left(35\right)\end{array}$

[0120] Where $\frac{\partial \alpha }{\partial t}$

[0121] can be measured using (11) or (14). See FIG. 7 for 𝑓(H).

[0122] D. Other ∂X_b/∂t measurements such as doppler, ultrasonic, electrical conductivity, magnetic permeability and other techniques have similar derivations. The important consideration is that ∂X_b/∂t is a normalized time varying quantity.

[0123] IV. Analytical implementation

[0124] If hematocrit is constant over a given time interval, averaging can eliminate system noise whose frequency components have corresponding periods much shorter than the interval. In addition, by observing the data variance during the interval it may be determined that the data is invalid. In the present system, the data acquisition rate is approximately 1000 data samples per second. This means that within a typical human pulse about 1000 samples of data are available for appropriate numerical analysis, averaging and qualification. Recognizing that both the intensity of light and the pressure in the transducer system are changing in time during the influx of blood is of great importance. Since the parametric relationship of ∂α/∂t as a function of ∂P/∂t (where P is pressure) during the cardiac cycle should be linear, a multiplicity of data points facilitate qualification of the signal for accuracy and linearity. Whereas, prior techniques involving only the peak and valley values of the cardiac cycle require numerous pulses to qualify the data set. See FIGS. 8, 9 and 10.

[0125] FIG. 8 shows di/dt/i as well as dP/dt verses time during the cardiac pulse—it is a pulse showing≈200+data samples during the pulse.

[0126] FIG. 9 shows (di/dt)/i vs dP/dt showing that within one cardiac pulse 200 plus data samples are link related, i.e. trace up out of the “0” origin up to a maximum value and then back down toward the origin again.

[0127] FIG. 10 shows dα/dt/dP/dt versus time during one single cardiac pulse with 200 plus samples of data from time 15-45 giving a value of about 4.5 thousandths. The data can then be averaged, as if 200+individuals pulse (max-min) values were actually taken as present day oxymeters do.

[0128] A. Homogeneity

[0129] Since the above derivations are based on the assumption of tissue homogeneity (i.e.,∂X_b1/∂t=∂X_b2/∂t, A₁=A₂, ∂A₁/∂X_b=∂A₂/∂X_b, α₁=α₂, etc.), high-speed, single-pulse, multiple parameter sampling allows for mathematical qualification of homogeneity, by requiring linearity of ln(i) vs. d and (∂i/∂t)/i vs. d. Under these constraints and when qualified as homogeneous, (∂α/∂t)/(∂P/∂t) also may be assumed to be linear over the entire pulse contour. Finally, both α and ∂α/∂t must also be linear, further assuring homogeneity in X_b, and in ∂X_b/∂t.

[0130] B. Circuitry

[0131] See U.S. Pat. No. 5,372,136 for the operational circuitry description, which allows for high speed sampling of the optical intensities. See FIGS. 6 and 10 for similar circuitry considerations for sampling of pressure, peizo, and strain-gage measurements.

[0132] The circuitry shown and discussed in U.S. Pat. No. 5,372,136 is programmable by conventional techniques to solve and implement the equations and calculations presented in this application. FIG. 6 shows a piezo transducer circuit having a transducer 50 connected to a series of operational amplifiers, resistors and capacitors in accordance with the figure. The circuit terminates in an analog output 52 for connection to the “E” connection shown in the middle left side of FIG. 9D in U.S. Pat. No. 5,372,136. FIG. 11, on the other hand, shows a pressure transducer circuit having a pressure transducer made 62 connected to a series of operational amplifiers, a capacitor, resistors and variable resistors as shown in the figure. The circuit terminates in an analog output also connected to the aforementioned “E” connection.

[0133] Referring more specifically to FIG. 6, a crystal oscillator is connected to ground and to the non-inverting input of a first operational amplifier, which may be an LM158. The non-inverting input of the first operational amplifier is connected to ground by a 0.047 μF capacitor C3. The first operational amplifier's feedback path to its inverting input includes a 470 K resistor R8. The first operational amplifier is suitably biased at the junction of a 220 Ω resistor R7 and a 150 μF capacitor C4 that are connected between VCC and ground.

[0134] A second operational amplifier, which may also be an LM 158, receives the output of the first operational amplifier at its inverting input via a 10 KΩ resistor R5. The second operational amplifier's non-inverting input is connected to several locations:

[0135] to a voltage VB51, which may be 4.096 volts, through a 10 KΩ resistor R2;

[0136] to a middle node of a voltage divider, the voltages divider extending between the non-inverting input of the first operational amplifier via a 10 MΩ resistor R4 to the middle node, and via a 10 KΩ resistor R1 to ground;

[0137] to the inverting input of the first operational amplifier via a 10 KΩ resistor R9; and

[0138] to ground via a 220 μF capacitor C5.

[0139] The second operational amplifier's feedback path to its inverting input includes a parallel arrangement of a 0.1 μF capacitor C2 and a 47 KΩ resistor R6. The second operational amplifier drives the A/D output 52 via a 10 KΩ resistor R3, the output connected to ground via a 1 μF capacitor C1.

[0140] Of course, the particular choice, arrangement and values of components shown in FIG. 6 may be varied while still remaining within the scope of the invention.

[0141] Referring now to FIG. 10, first through fourth operational amplifiers, which may be LM348s, are illustrated. The operational amplifiers are powered and biased by voltages VCC and VEE.

[0142] The first operational amplifier's non-inverting input is set to a value determined by the tap setting of a 1 KΩ adjustable resistor R2 that extends between VCC and VEE. The DAC input drives the first operational amplifier's inverting input via a 1 KΩ resistor R1. The first operational amplifier's feedback path includes a 50 KΩ adjustable resistor R4. The first operational amplifier drives the second operational amplifier's inverting input through an 11 KΩ resistor R3. The feedback path to the inverting input of the second operational amplifier includes a 100 Ω resisitor R5.

[0143] A transducer 62, which may include a Motorola MPX20100P, has opposite terminals that drive the non-inverting inputs of the second and third operational amplifiers, respectively. The other two opposite terminals of the transducer are connected to VCC and ground, respectively.

[0144] The second operational amplifier drives the inverting input of the third operational amplifier via a 750 Ω resistor R6. The third operational amplifier's feedback path to its inverting input includes a parallel arrangement of a 93.1 KΩ resistor R10 and a 0.001 μF capacitor C1.

[0145] The third operational amplifier drives the non-inverting input of the fourth operational amplifier via a 1 KΩ resistor R7. The inverting input of the fourth operational amplifier is connected to ground via a 1 KΩ resistor R8. The feedback path to the inverting input of the fourth operational amplifier includes a 50 KΩ adjustable resistor R9. The fourth operational amplifier drives the output of the FIG. 11 circuit.

[0146] Of course, the particular choice, arrangement and values of components shown in FIG. 13 may be varied while still remaining within the scope of the invention.

[0147] C. Preferred Embodiment

[0148] Physical embodiments as shown in FIG. 1 include the optical array, pressure transducer/balloon system and clam-shell fixture. Requisites of the preferred embodiment include a holder for the finger (or other tissue) such as seen in FIGS. 1 and 1A and 1B. This clam-shell fixture not only secures the tissue but also the optical array, and transducer system.

[0149] FIG. 1D is a schematic diagram for a mylar base member 38 that is shaped generally like a cross. As oriented in FIG. 1D, vertically extending portion 52 crosses with a horizontally extending portion 54 to yield top leg 56, bottom leg 58, and side legs 60, 62. In use, a finger 7 lies along the longitudinally extending portion 52 with the finger tip placed on the top leg 56 to properly cover the arrangement of LED's 32 and photodetector 34, which are arranged like those on FIGS. 1A-1C. A piezoelectric pressure transducer or strain gage 66 spans the horizontally extending portion 54 from near the tip of side leg 60 to the tip of side leg 62. In this orientation, the transducer or gage may be wrapped around the finger 7 for use in measurements.

[0150] The optical array 30, seen in FIG. 1D, shows the arrangement of multiple LED's 32 spaced at known separation distances from the detector 34. This array provides for the instantaneous distance, or “d”, derivative, by the transmission mode shown in FIG. 1A or in reflectance modes shown in FIGS. 1B and 1C. However, as shown in FIG. 1E, a single LED 42 swept across the finger 7 or tissue surface 9 with a stepper motor 44 would provide a d derivative as would a cantilevered clam-shell with an angular measurement device. In any case, d must be known and/or fixed. Also the detectors and emitters may be placed anywhere about the finger.

[0151] The pressure/balloon, strain gage, or peizo transducer system incorporated within the clam-shell fixture (see Section III, A, B, C and FIG. 1A) provides the contact surface area needed to define the ∂X_b/∂t.

[0152] High-speed sampling provides for a closer approximation of the instantaneous time, t, derivative, ∂/∂t, as opposed to peak-valley values, see FIG. 8. Therefore, the above embodiments allow for the direct measurement of In (i) at d₁, d₂, d₃and d₄cotemporaneously, thereby determining the actual a of the sampled tissue. Likewise (∂i/∂t)/i can be directly measured at d₁, d₂, d₃and d₄, cotemporaneously during the pulse which determines the instantaneous ∂α/αt.

[0153] The above mentioned optical array can be utilized transmissively and/or reflectively provided the separation distance between the detector and first emitter (d₁) is greater than 3 mm.

[0154] D. Choice of Non-Ionizing Wavelengths

[0155] Since hematocrit is an example of the desired biological constituent concentration value of interest, selection criteria of the preferred wavelength must include an understanding of equation (5). That is, a wavelength whose coefficients K_s, K_w, K_pare small compared to K_band which are also insensitive to oxygen saturation status must be selected. Such wavelengths include 805 nm, 590 nm, 569 nm and other isobestic wavelengths with negligible water absorption. While non-isobestic wavelengths, with small water absorption, could function, a second wavelength is needed to null out the oxygen saturation effects.

[0156] If the desired biologic constituent value of interest is the blood glucose, bilirubin, cholesterol or other parameters, then a second wavelength must be chosen. The first wavelength, 805 nm, is used to measure the hematocrit, H, after which a K_p805(the absorbance of plasma at λ=805 nm) can be determined. Then, knowing the H, a second wavelength, 570 nm, is chosen where K_p570is less than K_p805. Similarly, if the first wavelength used to measure the H and the reference glucose, K_p(glucose) is 570 nm, the second wavelength, 1060 nm, is chosen where K_p570is much less than K_p1060. In the case of bilirubin, the first wavelength used to measure the H and the reference bilirubin, K_p(bilirubin), is 570 nm, the second wavelength, 440 nm, is then chosen when K_p570is much less than K_p440. The selection of these above mentioned wavelengths therefore assures uniqueness for the measurement of the desired biologic constituent.

[0157] Additionally for glucose determination, recall that the 1300 nm wavelength is not hematocrit or hemoglobin dependent but will be glucose sensitive. This is primarily due to the dependence of the scattering coefficient on the difference between the index of refraction of pure water and glucose, i.e.: recall

[0158] S_b8=H (1−H) σ_s8(from equation 7) where:

[0159] σ_s8=8π²η₀²(η′₈−1)²·b^v/λ²

[0160] where η′₈=index of refraction of the RBC hemoglobin at 800 nm relative to plasma η₀(the plasma index of refraction), and,

[0161] S_b13=H (1−H)σ_s13(also from equation 7) where:

[0162] σ_s13=8π²η₀²(η′₁₃−1)²·b^v/λ²

[0163] η′₁₃=the index of refraction of glucose at 1300 nm relative to η₀.

[0164] Therefore, the 8 13 ratio has both hematocrit and glucose information. Whereas the α₈·α′₈/ΔP (equation 18a) ratio has only hematocrit information. Therefore the differential combination of those ratios will be a strong function of glucose only.

[0165] E. Improved Accuracy Pulse Oximeter Device

[0166] The accuracy of present day pulse oximeters suffers from 4 major problems: tissue perfusion (low X_band low ∂X_b/∂t), d dependence (varying finger sizes), tissue nonhomogeneity (the tissue penetration depth for 660 nm light is not the same as for 940 nm light), and H dependence (see equation (5)).

[0167] All of the above mentioned deficiencies in pulse oximetry can be eliminated by understanding equation (13). Equation (13) indicates an “offset term”, $-\frac{1}{A}\frac{\partial A}{\partial X_b}.$

[0168] Hence, while merely dividing (Δi/i)_λ1by (Δi/i)_λ2mitigates the effect of ∂X_b/∂t, the d's do not completely cancel, thereby yielding the above mentioned problems. To improve pulse oximeter accuracy, a derivative is needed as in (14), which eliminates the “offset term”. Hence, the ratio of (∂α/∂t)₈₀₅/(∂α/tt)₆₆₀results in no H, d, or X_bdependence and the use of the multiple LED array and high-speed sampling as mentioned in section IV qualifies the tissue as homogeneous.

[0169] V. A Simplified Two Step Approach

[0170] (A) Determination of H and X_b

[0171] The bulk attenuation coefficient, α, can be easily measured with the optical array, at 805 nm, utilizing equation (10) and as described in Section IV(C). Notice that at 805 nm, α is a strong function of H and X_bsince K_s8K_w8, K_p8are small, see FIG. 12.

[0172] Therefore, by knowing X_bitself, H can be determined. X_bitself can be determined using a strain gage in the following two step approach. Step One, measure the strain gage resistance when the finger is made bloodless, by squeezing finger, such as with a stepper motor. Step Two, measure the strain gage resistance when the finger is blood filled, for example by suction. Mathematically, at 805 nm and when K_s, K_p, K_w, are small, equation (3) is approximated by:

[0173] α²≈3KS   (36)

[0174] or

[0175] α²≈3[K_bX_b][S_bX_b+S₃X₃]  (37)

[0176] Substituting (5) and (7) into (37) yields: $\begin{array}{cc}0=3\left[H\frac{{\sigma }_a}{V}{X}_b\right]\left[H\left(1-H\right)\left(1.4-H\right)\frac{{\sigma }_s}{V}{X}_b+S_s{X}_s\right]-{\alpha }^2& \left(38\right)\end{array}$

[0177] With X_band α measured and known, and with the σ's and S_s, X_sapproximately constant, H can be solved with a quadratic formula or a polynomial fit. The strain gage determination of X_bis as follows:

[0178] Let V_o=the volume of a bloodless finger. Let V_𝑓=the volume of blood filled finger, and again considering the finger as a cylinder:

[0179] V_o=πr²z=V_s+V_w  (39)

[0180] V_𝑓=πR²z=V_b+V_s+V_w  (40)

[0181] and $\begin{array}{cc}\frac{V_o}{V_f}={\left(\frac{r}{R}\right)}^2& \left(41\right)\end{array}$

[0182] From equation (20) $\begin{array}{cc}{X}_b=\frac{V_f-V_o}{V_f}=1-\frac{V_o}{V_f}& \left(42\right)\end{array}$

[0183] Substituting (41) into (42): $\begin{array}{cc}{X}_b=1-{\left(\frac{r}{R}\right)}^2& \left(43\right)\end{array}$

[0184] Where the strain gage resistances are proportional to the radius, r and R, of the finger.

[0185] (B) Determination of tissue water content X_w

[0186] Choosing the wavelength of 1300 nm, where K_sand K_ware significant, the tissue water content, X_w, can be determined. Recall that 1−X_b−X_w=X_sand substituting into (3) yields:

[0187] α₁₃²=3({K_b−K_s}X_b+{K_w−K_s}X_w+K_s) ([{K_b−K_s}+{S_b−S_s}]X_b+[{K_w−K_s}−S_s]X_w+(K_s+S_s))   (44)

[0188] With α_l3′, X_band H determined and because K_b, K_s, K_w, S_b, and S_sare known coefficient values at 1300 nm, X_wis solved with either a quadratic formula or a polynomial fit.

[0189] FIG. 13 demonstrates preliminary results with 30 patients the application of the method and apparatus and the application of Equation 19 on numerous patients with a correlation of r=0.96. As implied throughout, those skilled in the art will also appreciate that the methods for determining blood hematocrit values within the scope of the present invention may be adapted for determining other non-hematocrit biologic constituent values such as glucose, bilirubin, cholesterol, tissue water, etc.

[0190] The present invention may be embodied in other specific forms without departing from its spirit or essential characteristics. While the foregoing described embodiments are to be considered in all respects only as illustrative of the claimed invention, they are not intended to restrict the scope of the claims. The scope of the invention is, therefore, indicated by the following appended claims rather than by the foregoing description. All changes within the meaning and range of equivalency of the claims are to be embraced within their scope.