Noninvasive pulsed infrared spectrophotometer

Claims

We claim:

1. A noninvasive pulsed infrared spectrophotometer for measuring the concentration of at least one predetermined constituent of a patient's blood, comprising:

an infrared source which emits broadband pulses of infrared light including n different wavelengths of at least 2.0 .mu.m, said pulses of infrared light containing energy at each of said n wavelengths being differentially absorbed by said at least one predetermined constituent whereby each predetermined constituent readily absorbs pulses of infrared light at one of said n wavelengths and minimally absorbs pulses of infrared light at another of said n wavelengths, and which directs said pulses of infrared light through an arterial blood vessel of the patient;

at least one infrared detector which detects light at said n wavelengths which has passed through said arterial blood vessel of the patient and has been selectively absorbed by said at least one predetermined constituent and which outputs at least one detection signal;

synchronizing means for synchronizing the application of said pulses of infrared light from said infrared source to said arterial blood vessel of the patient with the systolic and diastolic phases of a cardiac cycle of the patient, said synchronizing means including a cardiac monitor and means responsive to an output of said cardiac monitor for modulating said pulses of infrared light so that said infrared light passes through said arterial blood vessel of the patient only during diastolic and systolic time intervals respectively occurring during the systolic and diastolic phases of said cardiac cycle of the patient; and

means for determining the concentration of said at least one predetermined constituent of the patient's blood from said at least one detection signal.

2. A spectrophotometer as in claim 1, wherein said infrared source comprises one of a modulated laser and a modulated heated element which emits pulses of infrared light in a wavelength range of 2-20 .mu.m.

3. A spectrophotometer as in claim 2, further comprising a dichroic filter adjacent said infrared source which passes infrared energy in a range of approximately 8-12 .mu.m and reflects infrared energy outside of said range back into said infrared source.

4. A spectrophotometer as in claim 2, wherein said diastolic and systolic time intervals each have durations of approximately 0.1-2 msec during the systolic and diastolic phases of said cardiac cycle of the patient.

5. A spectrophotometer as in claim 2, wherein said at least one predetermined constituent includes glucose, said one wavelength is approximately 9.1 .mu.m and said another wavelength is approximately 10.5 .mu.m.

6. A spectrophotometer as in claim 5, further comprising a first bandpass filter disposed between the arterial blood vessel of the patient and said at least one infrared detector, said first bandpass filter passing infrared light in a narrow passband centered at approximately 9.1 .mu.m, and a second bandpass filter disposed between the arterial blood vessel of the patient and said at least one infrared detector, said second bandpass filter passing infrared light in a narrow passband centered at approximately 10.5 .mu.m.

7. A spectrophotometer as in claim 2, wherein said at least one predetermined constituent includes ethyl alcohol, said one wavelength is approximately 3.4 .mu.m and said another wavelength is approximately 4.8 .mu.m.

8. A spectrophotometer as in claim 7, further comprising a first bandpass filter disposed between the arterial blood vessel of the patient and said at least one infrared detector, said first bandpass filter passing infrared light in a narrow passband centered at approximately 3.4 .mu.m, and a second bandpass filter disposed between the arterial blood vessel of the patient and said at least one infrared detector, said second bandpass filter passing infrared light in a narrow passband centered at approximately 4.8 .mu.m.

9. A spectrophotometer as in claim 1, wherein said modulating means comprises a mechanical shutter disposed between said infrared source and said arterial blood vessel of the patient which is synchronized to said systolic and diastolic phases of said cardiac cycle of the patient so as to allow said infrared light to pass therethrough from said infrared source to the skin of the patient only during said systolic and diastolic time intervals respectively occurring during the systolic and diastolic phases of said cardiac cycle of the patient.

10. A spectrophotometer as in claim 9, wherein said cardiac monitor comprises a photoplethysmograph having a pulsed light emitting diode which directs light through a tissue of the patient and a photodetector which detects the light which has passed through said tissue of the patient, and said synchronizing means further comprises processing means for processing a detection output of said photodetector to determine a phase of said cardiac cycle and to control opening and closing of said mechanical shutter in accordance with said phase.

11. A spectrophotometer as in claim 10, wherein said processing means determines from said detection output of said photodetector when to open said mechanical shutter in a next cardiac cycle in accordance with the systolic and diastolic phases detected in a current cardiac cycle.

12. A spectrophotometer as in claim 11, wherein said processing means further determines from said detection output of said photodetector whether systole and diastole actually occurred in the current cardiac cycle at the same time said mechanical shutter was opened in the current cardiac cycle, and if said, forwarding said at least one detection signal to said concentration determining means for determination of the concentration of said at least one predetermined constituent.

13. A spectrophotometer as in claim 12, wherein said processing means includes a memory, said processing means repeating the steps of:

(a) for the current cardiac cycle, digitizing said detection output of said photodetector in sampling time intervals;

(b) selecting a distinctive feature of the digitized detection output of said photodetector for the current cardiac cycle and labelling a time interval of said distinctive feature as a cardiac cycle start time;

(c) labelling all subsequent time intervals in the current cardiac cycle by incrementing time intervals from said cardiac cycle start time until said distinctive feature is encountered in the next cardiac cycle;

(d) determining a peak in said digitized detection output of said photodetector for the current cardiac cycle and storing a time interval label identifying systole in the current cardiac cycle;

(e) determining a minimum in said digitized detection output of said photodetector for the current cardiac cycle and storing a time interval label identifying diastole in the current cardiac cycle;

(f) during the next cardiac cycle, counting a number of time intervals from a cardiac cycle start time of the next cardiac cycle in accordance with the time interval label identifying diastole for the current cardiac cycle and opening said mechanical shutter for a duration of said diastolic time interval, and then counting a number of time intervals from said diastolic time interval in accordance with the time interval label identifying systole for the current cardiac cycle and opening said mechanical shutter for a duration of said systolic time interval;

(g) when said mechanical shutter is open, recording in said memory said at least one detection signal from said at least one infrared detector;

(h) repeating steps (a) through (e) for the next cardiac cycle;

(i) determining whether diastole and systole for the next cardiac cycle actually occurred within a predetermined number of time intervals from when said mechanical shutter was opened in step (f); and

(j) if it is determined in step (i) that diastole and systole for the next cardiac cycle actually occurred within said predetermined number of time intervals from when said mechanical shutter was opened in step (f), passing said recorded at least one detection signal to said concentration determining means for the calculation of the concentration of said at least one predetermined constituent, but if it is determined in step (i) that diastole and systole for the next cardiac cycle did not actually occur within said predetermined number of time intervals from when said mechanical shutter was opened in step (f), erasing said recorded at least one detection signal from said memory.

14. A spectrophotometer as in claim 13, wherein said distinctive feature of the digitized detection output of said photodetector for the current cardiac cycle is a dicrotic notch of the current cardiac cycle.

15. A spectrophotometer as in claim 13, wherein each sampling time interval has a duration of approximately 0.1 to 2.0 ms.

16. A spectrophotometer as in claim 1, wherein said cardiac monitor comprises an electrocardiogram.

17. A spectrophotometer as in claim 1, wherein said modulating means comprises means for electrically modulating said pulses of infrared light so that said infrared light passes through said arterial blood vessel of the patient only during said diastolic and systolic time intervals.

18. A spectrophotometer as in claim 1, wherein said concentration determining means forms a ratio R.sub.1 =(Sys L1-Dias L1)/(Sys L2-Dias L2), where Sys L1 is a detected systolic phase signal at said one wavelength, Dias L1 is a detected diastolic phase signal at said one wavelength, Sys L2 is a detected systolic phase signal at said another wavelength, and Dias L2 is a detected diastolic phase signal at said another wavelength.

19. A spectrophotometer as in claim 18, wherein said concentration determining means calculates the concentration of said at least one predetermined constituent (n) of the patient's blood in accordance with the following equation: ##EQU2## where: C.C..sub.n is the concentration of said at least one predetermined constituent n;

B, C.sub.x,y and D.sub.z are empirically determined calibration coefficients; m is the number of detection and reference wavelengths used;

p is the highest order of polynomial used; and Ln is a natural log function.

20. A method of noninvasively measuring the concentration of at least one predetermined constituent of a patient's blood, comprising the steps of:

emitting pulses of infrared light at n different wavelengths of at least 2.0 .mu.m, pulses of infrared light at each of said n wavelengths being differentially absorbed by said at least one predetermined constituent, each predetermined constituent readily absorbing pulses of infrared light at one of said n wavelengths and minimally absorbing pulses of infrared light at another of said n wavelengths, and directing said pulses of infrared light through an arterial blood vessel of the patient;

detecting light at said n wavelengths which has passed through said blood vessel of the patient and has been selectively absorbed by said at least one predetermined constituent and outputting at least one detection signal;

synchronizing the direction of said pulses of infrared light through said arterial blood vessel of the patient with the systolic and diastolic phases of a cardiac cycle of the patient; and

determining the concentration of said at least one predetermined constituent of the patient's blood from said at least one detection signal.

21. A method as in claim 20, wherein said synchronizing step includes the step of modulating said pulses of infrared light so that said infrared light passes through said arterial blood vessel of the patient only during diastolic and systolic time intervals respectively occurring during the systolic and diastolic phases of said cardiac cycle of the patient.

22. A method as in claim 21, wherein said synchronizing step includes the steps of directing light through a tissue of the patient, detecting the light which has passed through said tissue of the patient, and processing a detection output of said light detecting step to determine the phase of said cardiac cycle and to control modulation of said pulses of infrared light in said modulating step.

23. A method as in claim 22, wherein said synchronizing step includes repeating the steps of:

(a) for a current cardiac cycle, digitizing said detection output of said infrared light detecting step in sampling time intervals;

(b) selecting a distinctive feature of the digitized detection output of said digitizing step for the current cardiac cycle and labelling a time interval of said distinctive feature as a cardiac cycle start time;

(c) labelling all subsequent time intervals in the current cardiac cycle by incrementing time intervals from said cardiac cycle start time until said distinctive feature is encountered in a next cardiac cycle;

(d) determining a peak in said digitized detection output for the current cardiac cycle and storing a time interval label identifying systole in the current cardiac cycle;

(e) determining a minimum in said digitized detection output for the current cardiac cycle and storing a time interval label identifying diastole in the current cardiac cycle;

(f) during the next cardiac cycle, counting a number of time intervals from a cardiac cycle start time of the next cardiac cycle in accordance with the time interval label identifying diastole for the current cardiac cycle and applying said pulses of infrared light to said arterial blood vessel of the patient for a duration of said diastolic time interval, and then counting a number of time intervals from said diastolic time interval in accordance with the time interval label identifying systole for the current cardiac cycle and applying said pulses of infrared light to said arterial blood vessel of the patient for a duration of said systolic time interval;

(g) when said infrared light is applied to said blood vessel of the patient, recording in a memory said at least one detection signal;

(h) repeating steps (a) through (e) for the next cardiac cycle;

(i) determining whether diastole and systole for the next cardiac cycle actually occurred within a predetermined number of time intervals from when said pulses of infrared light were applied to said blood vessel of the patient in step (f); and

(j) if it is determined in step (i) that diastole and systole for the next cardiac cycle actually occurred within said predetermined number of time intervals from when said pulses of infrared light were applied to said arterial blood vessel of the patient in step (f), calculating the concentration of said at least one predetermined constituent from said recorded at least one detection signal, but if it is determined in step (i) that diastole and systole for the next cardiac cycle did not actually occur within said predetermined number of time intervals from when said pulses of infrared light were applied to said arterial blood vessel of the patient in step (f), erasing said recorded at least one detection signal from said memory.

24. A method as in claim 23, including the further step of repeating steps (a) through (j) until the concentration of said at least one predetermined constituent is calculated from said recorded at least one detection signal a plurality of times, and then averaging the plurality of calculated concentrations to get an average concentration value for said at least one predetermined constituent of the patient's blood.

25. A method as in claim 20, wherein said concentration determining step includes the step of forming a ratio R=(Sys L1-Dias L1)/(Sys L2-Dias L2), where Sys L1 is a detected systolic phase signal at said one wavelength, Dias L1 is a detected diastolic phase signal at said one wavelength, Sys L2 is a detected systolic phase signal at said another wavelength, and Dias L2 is a detected diastolic phase signal at said another wavelength.

26. A method as in claim 25, wherein said concentration determining step includes the step of calculating the concentration of said at least one predetermined constituent (n) of the patient's blood in accordance with the following equation: ##EQU3## where: C.C..sub.n is the concentration of said at least one predetermined constituent n;

B, C.sub.x,y and D.sub.z are empirically determined calibration coefficients; m is the number of detection and reference wavelengths used;

p is the highest order of polynomial used; and Ln is a natural log function.

BACKGROUND OF THE INVENTION

1. Field of the Invention

The present invention relates to an instrument and a method for noninvasively measuring the concentration of glucose, dissolved carbon dioxide, ethyl alcohol or other constituents in a patient's blood. In particular, the present invention relates to an instrument and associated method for monitoring the infrared absorption of such constituents in a patient's blood at long infrared wavelengths where such constituents have strong and distinguishable absorption spectra by passing long wavelength infrared energy through a finger or other vascularized appendage of the patient and measuring the resultant absorption.

2. Brief Description of the Prior Art

Infrared detection techniques have been widely used for the calculation of oxygen saturation and the concentration of other blood constituents. For example, noninvasive pulse oximeters have been used to measure absorption signals at two or more visible and/or near infrared wavelengths and to process the collected data to obtain composite pulsatile flow data of a patient's blood. Sample pulse oximeters of this type are described by Corenman et al. in U.S. Pat. No. 4,934,372; by Edgar, Jr. et al. in U.S. Pat. No. 4,714,080; and by Zelin in U.S. Pat. No. 4,819,752.

Infrared detection techniques have also been used to calculate the concentrations of constituents such as nitrous oxide and carbon dioxide in the expired airstream of a patient. For example, Yelderman et al. describe in U. S. Pat. Nos. 5,081,998 and 5,095,913 techniques for using infrared light to noninvasively measure the absolute concentrations of the constituents of the respiratory airstream of a patient by placing an infrared transmission/detection device on the artificial airway of the patient. These infrared detection techniques and those described above have proven to be quite accurate in the determination of arteriole blood oxygen saturation, the patient's pulse, and the concentrations of carbon dioxide, nitrous oxide and other respiratory constituents.

Spectrophotometric methods have also been used to noninvasively monitor the oxidative metabolism of body organs in vivo using measuring and reference wavelengths in the near infrared region. For example, Jobsis describes in U.S. Pat. Nos. 4,223,680 and 4,281,645 a technique in which infrared wavelengths in the range of 700-1300 nm are used to monitor oxygen sufficiency in an organ such as the brain or heart of a living human or animal. In addition, Wilber describes in U.S. Pat. No. 4,407,290 a technique in which visible and near infrared light emitting diodes and detection circuitry are used to noninvasively measure changes in blood thickness of predetermined blood constituents relative to total change in blood thickness at a test area so as to determine the concentration of such constituents in the blood. Such constituents include hemoglobin and oxyhemoglobin, and the measured concentrations are used to determine the oxygen saturation of the blood. Wilber further suggests at columns 11-12 that such techniques may be extended to the measurement of glucose in the bloodstream; however, Wilber does not tell how to make such measurements, what wavelengths of energy to use, or the form of the mathematics necessary for the calculation of glucose concentration.

Extension of the noninvasive blood constituent measuring techniques described above for use in measuring glucose concentration in the bloodstream is highly desirable. According to the American Diabetes Association, more than 14 million people in the United States have diabetes, though about half of them are not aware of it. Almost 750,000 people per year are diagnosed with diabetes, while approximately 150,000 die from the disease or its complications each year. Since people with diabetes are at risk for blindness, kidney disease, heart disease and stroke, they need to control the disease by closely monitoring their blood glucose levels and carefully controlling the intake of insulin and glucose. Numerous home diagnostic devices have been developed for this purpose.

For example, conventional procedures used to measure glucose levels in the bloodstream include biochemical, electrochemical and spectroscopic techniques. The biochemical techniques measure the glucose oxidase reaction and are widely used in laboratories and in conventional consumer glucose monitoring instruments such as the One Touch.RTM. glucose monitor manufactured by LifeScan, Inc. Although relatively accurate, this technique requires a sample of blood to be withdrawn from the patient and applied to a chemically reactive test strip. The repeated withdrawal of blood samples is less than desirable. The electrochemical techniques, on the other hand, do not require the withdrawal of blood. However, these techniques typically require the surgical implantation of glucose electrodes and cells in the patient for use in providing signals to a regulated insulin reservoir (such as an artificial pancreas). While these techniques show great promise for use in implants and automatic insulin control systems, the associated systems are relatively inaccurate, insensitive and not very selective. Obviously, this technique is quite invasive; nevertheless, it is useful in the case of severe diabetes were the sensor can be implanted together with the electronically regulated insulin reservoir or artificial pancreas to form a complete closed loop system for severely affected diabetics.

Spectroscopic glucose monitoring techniques using infrared light are presently believed to be the most accurate and are the subject of the present application. Unlike the noninvasive oxygen saturation measurement techniques described above, prior art spectroscopic glucose monitoring techniques have typically used extra-corporeal "flow through" cells that allow continuous measurements using infrared light. Indeed, attenuated total internal reflection (ATR) cells have been employed in the long wavelength infrared to measure the glucose content of extracted blood samples. However, such techniques also require samples of blood to be taken from the patient and are thus undesirable for widespread consumer use.

Laser Raman Spectroscopy is another spectroscopic technique which uses a visible spectrum range stimulus and the visible red spectrum for measurement. As with ATR cells, extra-corporeal blood is also used with Raman technology to make the glucose measurements. However, the Raman technique is based upon the principle that over the entire visible spectrum range whole blood has a high absorption due to haemoglobin and other chromophores which produce a high fluorescence background making detection of bands that are not resonance amplified very difficult. Sub-nanosecond laser pulses are used to overcome some of these problems; however, this technology is quite complex and expensive.

Another spectroscopic technique offers a non-invasive solution to the problem of measuring glucose in the bloodstream. According to this technique, near infrared spectroscopy, light is passed through a finger or suitable appendage for measuring glucose levels in vivo. Unfortunately, this technique suffers from two sources of inaccuracy: tissue interference and lack of specificity. Moreover, while the near infrared wavelengths used are easily and economically generated by light emitting diodes (LEDS) and solid state lasers, they are not in a range specifically absorbed by glucose. This lack of "fingerprint" absorbance and interference from tissue pigment and condition render the technique useless for accurate concentration determination but possibly acceptable for trending if stability can be maintained. Samples of prior art patents describing such spectroscopic techniques are described below.

Kaiser describes in Swiss Patent No. 612,271 a technique in which an infrared laser is used as the radiation source for measuring glucose concentration in a measuring cell. The measuring cell consists of an ATR measuring prism which is wetted by the patient's blood and an ATR reference prism which is wetted with a comparison solution. CO.sub.2 laser radiation is led through the measuring cell and gathered before striking a signal processing device. A chopper placed before the measuring cell allows two voltages to be obtained corresponding to the signal from the sample and the reference prisms. Due to absorption corresponding to the concentration of the substance measured in the blood, the difference between the resulting voltages is proportional to the concentration. Unfortunately, the infrared laser used by Kaiser has the undesirable side-effect of heating the blood, which may be harmful to the patient, and also does not overcome the effects of tissue absorption. Although Kaiser suggests that heating of the blood may be prevented by using extra-corporeal cuvettes of venous blood and high blood flow rates, Kaiser does not describe a noninvasive technique for measuring glucose concentration which overcomes the effects of tissue absorption or other sources of error which are present in the portion of the infrared spectrum were Kaiser makes his measurements.

March in U.S. Pat. No. 3,958,560 describes a "noninvasive" automatic glucose sensor system which senses the rotation of polarized infrared light which has passed through the cornea of the eye. March's glucose sensor fits over the eyeball between the eyelid and the cornea and measures glucose as a function of the amount of radiation detected at the detector on one side of the patient's cornea. Unfortunately, while such a technique does not require the withdrawal of blood and is thus "noninvasive", the sensor may cause considerable discomfort to the patient because of the need to place it on the patient's eye. A more accurate and less intrusive system is desired.

Hutchinson describes in U.S. Pat. No. 5,009,230 a personal glucose monitor which also uses polarized infrared light to noninvasively detect glucose concentrations in the patient's bloodstream. The amount of rotation imparted on the polarized light beam is measured as it passes through a vascularized portion of the body for measuring the glucose concentration in that portion of the body. Although the monitor described by Hutchinson need not be mounted on the patient's eye, the accuracy of the measurement is limited by the relatively minimal absorption of glucose in the 940-1000 nm range used by Hutchinson.

Dahne et al. in U.S. Pat. No. 4,655,225 describe a spectrophotometric technique for detecting the presence of glucose using specially selected bands in the near infrared region between 1100 and 2500 nm. Dahne et al. found that by applying light at wavelengths in the 1000-2500 nm range acceptable combinations of sufficient penetration depth to reach the tissues of interest with sufficient sensitivity may be obtained for ascertaining glucose concentration variations without the risk of overheating tissues.

Mendelson et al. in U.S. Pat. No. 5,137,023 also found that wavelengths in the near infrared range are useful for noninvasively measuring the concentration of an analyte such as glucose using pulsatile photoplethysmography. In particular, Mendelson et al. describe a glucose measuring instrument which uses the principles of transmission and reflection photoplethysmography, whereby glucose measurement is made by analyzing either the differences or the ratio of two different near infrared radiation sources that are either transmitted through an appendage or reflected from a tissue surface before and after blood volume change occurs in the systolic and diastolic phases of the cardiac cycle. The technique of photoplethysmography can thus be used to adjust the light intensity to account for errors introduced by excessive tissue absorptions. However, despite the assertions by Dahne et al. and Mendelson et al., the wavelengths in the near infrared (below 2500 nm) are not strongly absorbed by glucose yet are susceptible to interference from other compounds in the blood and thus cannot yield sufficiently accurate measurements.

Rosenthal et al. in U.S. Pat. No. 5,028,787 disclose a noninvasive blood glucose monitor which also uses infrared energy in the near infrared range (600-1100 nm) to measure glucose. However, as with the above-mentioned devices, these wavelengths are not in the primary absorption range of glucose and, accordingly, the absorption at these wavelengths is relatively weak. A more accurate glucose measuring technique which monitors glucose absorption in its primary absorption range is desired.

As with other molecules, glucose more readily absorbs infrared light at certain frequencies because of the characteristic and essential invariate absorption wavelengths of its covalent bonds. For example, as described by Hendrickson et al. in Organic Chemistry, 3rd Edition, McGraw-Hill Book Company, Chapter 7, Section 7-5, pages 256-264, C--C, C--N, C--O and other single carbon bonds have characteristic absorption wavelengths in the 6.5-15 micron range. Due to the presence of such bonds in glucose, infrared absorption by glucose is particularly distinctive in the far infrared. Despite these characteristics, few have suggested measuring glucose concentration in the middle to far infrared range, likely due to the strong tissue absorption that would attenuate signals in that range.

In one known example of such teachings, Mueller describes in WO 81/00622 a method and device for determining the concentration of metabolites in blood using spectroscopic techniques for wavelengths in the far infrared range. In particular, Mueller teaches the feasibility of measuring glucose in extra-corporeal blood samples using a 9.1 .mu.m absorption wavelength and a 10.5 .mu.m reference wavelength for stabilizing the absorption reading. However, Mueller does not describe how such wavelengths maybe used in vivo to measure glucose concentration noninvasively while overcoming the above-mentioned tissue absorption problems. Mueller also does not suggest synchronizing such determinations to the systolic and diastolic phases of the heart for minimizing tissue absorption errors.

Accordingly, it is desired to extend the techniques used in noninvasive pulse oximeters and the like to obtain absorption signals from pulsing arterial blood which can be used for accurate measurements of the concentration of glucose, ethyl alcohol and other blood constituents while overcoming the problems caused by interference from tissues and the like. In particular, a noninvasive blood constituent measuring device is desired which uses long wavelength infrared energy for better absorption characteristics and improved signal to noise ratios while also synchronizing the pulses of long wavelength infrared energy with the cardiac cycle so that very accurate in vivo measurements of the concentrations of such constituents in the arterioles may be made. A method and device for this purpose is described herein.

SUMMARY OF THE INVENTION

The above-mentioned limitations in prior art glucose and other blood constituent measuring devices are overcome by providing an instrument which noninvasively measures the concentration of glucose and other blood constituents in a patient's blood by monitoring the infrared absorption of the blood constituent in the blood at long infrared wavelengths were such blood constituents have strong and readily distinguishable absorption spectra. Preferably, the long wavelength infrared energy is passed through a finger or other vascularized appendage and the measurement is made without injury, venipuncture or inconvenience to the patient.

Since the patient's tissue, water and bone are also strong and variable absorbers of long wavelength infrared energy, the signal to noise ratio is such a system could cause serious errors in the blood constituent concent