MXPA01000468A - Non-invasive glucose monitor - Google Patents

Abstract

This invention is a non-invasive method for determining blood level of an analyte of interest, such as glucose, comprising generating an excitation laser beam (10) (e.g., at a wavelength of 700 nanometers to 900 nanometers);focusing (13) the excitation laser beam into the anterior chamber of an eye of the subject so that aqueous humor in the anterior chamber is illuminated;detecting (32) (preferably confocally (22) detecting) a Raman spectrum from the illuminated aqueous humor;and thendetermining the blood glucose level (or the level of another analyte of interest) for the subject from the Raman spectrum. Preferably, the detecting step is followed by the step of subtracting a confounding fluorescence spectrum from the Raman spectrum to produce a difference spectrum;and determining the blood level of the analyte of interest for the subject from that difference spectrum, preferably using linear or nonlinear multi-variate analysis such as partial least squares analysis. Apparatus for carrying out the foregoing method is also disclosed.

Description

NON INVASIVE GLUCOSE MONITOR This application claims the benefit of the provisional application of E.U.A No. 60 / 092,545, filed July 13, 1998, the description of which is incorporated herein by reference in its entirety.

FIELD OF THE INVENTION The present invention relates to methods and apparatus for the non-invasive monitoring of blood glucose levels by spectographic analysis of the aqueous humor in the anterior chamber of the eye.

BACKGROUND OF THE INVENTION The non-invasive measurement of blood glucose by any method that includes optical spectroscopy techniques has been an elusive target for at least two decades. Blood, tissue and most of the excreted fluids contain numerous substances that confuse the spectral identifications of glucose. On the other hand, the aqueous humor (HA) that fills the anterior chamber of the eye (between the lens and the cornea) contains relatively few molecules capable of interfering with the spectroscopic detection of glucose. These are mainly lactate, ascorbate and urea. This fact and its optically accessible location behind láat ^ MÉiliÉirililliilÉ. the cornea makes the HA an attractive choice as a site in which non-invasive glucose analysis is attempted. Pohjola { Acta Ophthalmologica Suppl, 88, 1-80 (1996)) demonstrates that the ratio of glucose in aqueous humor to plasma glucose in normal euglycemic individuals is related to age and varies from 0.6 to 0.9. In addition, it shows that similar relationships apply in seven humans with constant state hyperglycemia. There is no data on the equilibrium time of glucose in aqueous humor or changes in plasma glucose in humans. 10 Numerous researchers over the years have suggested that the ratio of glucose in aqueous humor to plasma glucose in the normoglycemic rabbit ranges from 0.42 to 1.01 (S. Pohjola, supra; D. Reddy and V. Kinsey, Arch. Ophthalmol, 63, 715-720 (1960), M. Reim et al., Ophthalmologica 154, 39-50 (1967), W. March et al., Diabetes Care 5, 259 (1982)). There is no certainty whether this variability is normal or could be attributed to differences in glucose measurement techniques, collection techniques, sample storage and anesthesia. The ratio of glucose in aqueous humor to glucose in ascending or descending plasma has not been studied previously in rabbits. 20 Cote has reviewed the relative strengths and weaknesses of the optical glucose detection techniques (J. Clin. Engineering 22, 253 (1997)). Raman spectroscopy is potentially attractive because it can distinguish glucose in water solutions that contain various levels of other ^^ | optically active metabolites (S. Wang et al., Applied Optics 32, 925 (1993)). Raman spectroscopy measures the change in wavelength of incident light when molecules disperse it. Any given molecule causes a change in characteristics in the scattered light spectrum, which depends on its intermolecular and intramolecular bonds. This is in distinction by contrast to fluorescence, which is caused by changes in electron energy states, and does not change in relation to the wavelength of the incident light. Wicksted et al, (Appl. Sectroscop 49, 987 (1995)) suggest that the identification of Raman for glucose can be identified in samples of aqueous humor, and Goetz, et al (IEEE Trans Biomed. Eng. 42, 728 (1995)), have shown that higher physiological levels of glucose can be measured with Raman spectroscopy in water solutions. J. Lambert et al., (LEOS Newsletter 12, 19-22 (1998)) suggest that measurement of glucose at 15 physiological levels is possible in water solutions containing other analytes that are commonly found in the aqueous humor. However, when studying solutions containing fluorescent substances, the fluorescence signal can dominate the relatively weak Raman signal shift. This is a potential problem, if Raman spectroscopy is applied to the aqueous humor that contains proteins that fluoresce. U.S. Patent No. 5,243,973 to Tarr et al. Suggests a non-invasive blood glucose measurement system using stimulated Raman spectroscopy. Raman spectroscopy - •• - »• * - '•» «• * -" - "-.«. «* Stimulated stimulation requires the use of a pump and a probe laser beam. It is used to measure Raman light stimulated at a single wavelength after transmission through the anterior chamber of the eye.This is not desirable, since an optical component that makes contact with the eye to direct the beam is required. In addition, the use of a single wavelength can limit the ability to measure glucose at physiological levels within the tissue that contains many other Raman scattering chemicals. Stark suggests a non-invasive glucose measuring device employing wide-band infrared light stimulation. U.S. Patent No. 5,553,617 to Barkenhagen suggests a non-invasive method for measuring the body chemistry of a subject's eye by measuring a spectral response, such as a response of Raman scattering. While it is suggested that the invention can be used for medical applications, such as the determination of sugar in diabetics, no specific details are provided on how this can be done accurately. U.S. Patent No. 5,710.30 for Essenpreis suggests a method for measuring the concentration of glucose in a biological sample, such as the eye (see figure 4 therein) with interferometric measurement methods.

VM.VMA? ** ^ M »r, _," _ ^,. US Patent No. 5,666,956 to Buchert et al, suggests that an instrument for the non-invasive measurement of an analyte of the body can be based on naturally-emitted infrared radiation. Despite the above efforts, a non-invasive blood glucose monitor that is commercially adequate and based on a non-invasive analysis of the aqueous humor of the eye has not yet been developed. Difficulties in the development of such a device include the correlation of glucose levels in aqueous humor to blood glucose levels, the difficulty in obtaining accurate measurements, and the need to minimize the damaging effects to the eye caused by excessive exposure. to light in an instrument that the subjects will use on a repetitive basis. Consequently, there is a continuing need to seek new methods for the non-invasive analysis of blood glucose levels. BRIEF DESCRIPTION OF THE INVENTION A first aspect of the present invention is a non-invasive method for determining the blood level of an analyte of interest, such as glucose. The method comprises: generating an excitation laser beam (for example, at a wavelength of 700 to 900 nanometers); ááti ^ ^ ^ ^ i í_ _ _____? _________, E hriWkiÉliÉiÉBla .ir.í- i i. "« a aAt.- .. focusing the excitation laser beam on the anterior chamber of an eye of the subject so that the aqueous humor in the anterior chamber is illuminated; detect (preferably confocal detection) a Raman spectrum from the illuminated aqueous humor; and then determine the level of glucose in the blood (or the level of another analyte of interest) for the subject from the Raman spectrum. Preferably, the detection step is followed by the step to subtract a confusing fluorescence spectrum from the Raman spectrum to produce a difference spectrum; and determining the blood level of the analyte of interest to the subject from that spectrum of difference, preferably by the use of linear or non-linear multivariate analyzes, such as partial least squares or artificial neural network algorithms. A second aspect of the present invention is an apparatus for the non-invasive determination of the blood level of an analyte of interest, such as glucose in a subject. The apparatus comprises: a laser for generating an excitation laser beam (for example at a wavelength of 700 to 900 nanometers); an optical system (for example a confocal optical system) operatively related to said laser for focusing the excitation laser beam in the anterior chamber of an eye of the subject so that the aqueous humor in the anterior chamber is illuminated; -ifkÉltttÉÉBiflhita Umt ^ u ______ m iiiii iiiiéii jJ Jm? a detector operatively related to the optical system and configured to detect a Raman spectrum from the illuminated aqueous humor; preferably a subtraction system, hardware and / or software or other suitable means for extracting a fluorescence spectrum for said aqueous humor from said Raman spectrum to produce a difference spectrum; a processor for determining the blood level of the analyte of interest for said subject from the Raman spectrum (or preferably the difference spectrum). Numerous additional features can be incorporated into the apparatus. The apparatus may include a display screen to visually display the test results on the subject through the same opening when the test is performed. It may include a visual fixation device, also visible through the test aperture, which controls the movement of the eye and at the same time ensures that the laser focus is properly directed in the anterior chamber of the eye. The processor may contain an empirical model of actual testing experience to determine the blood level of the analyte of interest. The apparatus may employ a tuneable laser, a plurality of fixed wavelength lasers, or other means for sliding the Raman spectrum passing a plurality of length detectors. of different waves to obviate the need for a Raman spectrometer in full grid. A line of - jg ^ jj ^ ^ communication connected to the processor to transmit the blood level of the analyte of interest to a remote location. Even other features that may be included in the methods and apparatuses described above are explained below. 5 BRIEF DESCRIPTION OF THE DRAWINGS Figure 1 illustrates that the wide fluorescence spectrum of the aqueous humor may obscure the underlying peaks of interest in the spectrum of Raman. Unlike the fluorescence spectrum, the Raman spectra will change with a modification in the excitation wavelength. The pure spectrum of the aqueous humor of a rabbit is shown taken at two slightly different (higher) wavelengths. The difference spectrum (lower) obtained by subtracting a pure spectrum from the other reveals a resulting bipolar Raman identification. Then linear and non-linear multivariate analyzes can be applied. Figure 2. The glucose concentration in aqueous humor of 16 rabbits was calculated with Raman spectroscopy and compared with the actual glucose concentration measured with a commercial glucometer. The graph shows the concentration of glucose predicted by Raman after subtracting the fluorescence and applying an algorithm of linear partial least squares followed by non-linear retropropagation with a neural network ^^ ¡F «g ^^^^" j3tó2? Íiíííí ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ This resulted in a high degree of correlation (^ = 0.98) of what was predicted with actual glucose concentration. The single application of the partial least squares algorithm resulted in a lower correlation (^ = 0.90). 5 Figure 3. Blood glucose rises easily at variable rates in rabbits after administration of xylazine anesthesia. In the animal (rabbit D, filled diamonds) there was a minor change in blood glucose for unknown reasons. Figure 4. The better adjusted second-order polynomial curves show the relationship between glucose in aqueous humor and plasma glucose, while plasma glucose rises in 9 animals. Aqueous glucose measurements of the first paracentesis of an eye correlate well with simultaneous plasma glucose (dark curve). When the plasma glucose exceeds 200 mg / dL, 15 the relationship is almost linear (dotted line). Glucose in aqueous humor exceeds plasma glucose when plasma glucose is less than 200 mg / dL. The ratio of glucose in aqueous humor to plasma glucose is different when the aqueous humor sample is obtained as a second paracentesis (light curve) suggesting that the initial paracentesis alters the normal glucose homeostasis. Figure 5 schematically illustrates a first embodiment of an apparatus of the invention.

, «* +, AISto? * .- *, »-a« «• > . . . ? . . ».... . . . . . . . . .. .. . . _,. "» _. _. t i jt nite ^ V Figure 6 schematically illustrates a second embodiment of an apparatus of the invention. Figure 7a illustrates schematically the method for sliding the Raman spectra viewed through a particular detector window by changing the excitation frequency. Figure 7b illustrates schematically an apparatus of the invention employing the method illustrated in Figure 7a. Figure 8 illustrates a visual fix display that can be employed in an apparatus of the invention. Figure 9 illustrates a visible reading display that can be used in an apparatus of the invention.

DETAILED DESCRIPTION OF THE PREFERRED MODALITIES The term "glucose" as used herein preferably refers to glucose D. The term "subject" as used herein refers to human and animal subjects, such as dogs, cats and rabbits. Animal subjects can be used in the present invention for veterinary purposes. Human subjects are preferred. The term "processor" as used herein refers to a hardware device, software or device made with __________________ ... ». .1 . .aüm ^ ... a program that runs on a general-purpose computer, or combinations of hardware and software devices. The present invention relates primarily to the determination of blood glucose levels, and is thus explained herein. However, the concentration of blood of other active molecules or analytes by Raman, such as lactate, urea, ascorbate, drugs, steroids and alcohol (in particular ethanol) can also be determined by these techniques. The step of generating an excitation laser beam pulse can be performed with any suitable laser beam source. The excitation laser pulse energy must be low enough to avoid tissue toxicity, but high enough to provide a Raman signal that can be measured from the aqueous humor. In general, the laser beam pulse will be at a wavelength of 700 to 900 nanometers, and more preferably will be at a wavelength of 780 to 860 nanometers to reduce fluorescence, increase tissue penetration, and reduce phototoxicity to the eye The duration of the put will almost always be from 1 to 2 seconds to 30 or 60 seconds (that is, half a minute to a minute). The total energy of the laser pulse will almost always be 200-500 millijoules with instantaneous energy that does not exceed 30 to 50 milliwatts. The optical components of the apparatus used to perform the method of preference are configured so that the energy in the retina of the eye (as well as other areas susceptible to tissue toxicity, such as the lens and cornea) J ^^ of excitation laser beam pulse is not greater than 3000 milliwatts per square centimeter, and more preferably not greater than 2000, 1000 or even 500 milliwatts per square centimeter. Any source laser that provides the desired frequency output 5 can be used to perform the invention. A distributed feedback laser can be used to reduce the size of the instrument. Tunable or multiple fixed frequency lasers can be combined with bandpass filters (Puppels et al., Applied Spectroscopy 47, 1256-67 (1993)) which select the Raman scattering at wavelengths that provide optimal information for the multivariate analysis (this reduces the cost and size of the instrument compared to the use of holographic filters or grids). Any suitable detector can be used to detect a Raman spectrum from an illuminated aqueous humor. A CCD detector or CCD camera is preferred. Preferably, the CCD detector has a high quantum performance in the near infrared range. The high quantum yield can be achieved through any suitable means, such as using a detector with a thin back wall, but thick enough to reduce the interferometer effect. The fluorescence spectrum for the aqueous humor can be subtracted from the Raman spectrum by stimulating the aqueous humor with a second pulse of excitation laser light at a slightly different wavelength of the first pulse (for example, up to two nanometers from the - "-" * - - < a ^ at - "" - ^ ateAMMia first pulse), and then by subtracting one spectrum from the other in a processor in accordance with conventional techniques. These techniques are known. See, for example, Funfschilling and Williams, Applied Spectroscopy 30, 443 (1976); Baraga et al., Applied Spectroscopy 46, 187 (1992); Wicksted et al., Applied Spectroscopy 49, 987 (1995)). In the alternative, the fluorescence spectrum can be subtracted through the use of software or other processing techniques. In this way, the term "subtraction" as used herein is intended to include techniques such as filtration. Although not essential, other spectra or signals of potential interference, such as Raman scattering, lens, iris or cornea, can also be filtered or subtracted through a hardware and / or software processor. Almost always, water is not important for Raman spectroscopy, but a water spectrum can be subtracted if desired. The blood glucose level for the subject is then determined from the difference spectrum by an empirical base model, formula or real subject matrix or sample test in a hardware and / or software processor. The model can be obtained through linear multivariate analysis (for example, partial least squares) or non-linear analysis (for example, artificial neural networks). It is important that the concatenated series of Raman spectra used for the multivariate analyzes include samples of the aqueous humor with a wide range of concentrations of the Raman dispersion metabolites. The major Raman dispersion metabolites (or "Raman active compounds") (glucose, lactate, urea, ascorbate and any exogenous compound or drug present) preferably should not vary colinear to one another among the samples in the concatenated series. The model can be produced with spectral samples that are obtained from one or more previous subjects, can be produced with spectra samples that are obtained from the subject from which the current blood level of the analyte of interest is being determined (in which case, that subject would be required to provide a blood sample to determine the blood concentration of the analyte of interest), or both. Almost always, the concatenated series will require at least 20, 25 or 30 samples of Raman spectra (and samples of the corresponding blood levels of the analyte of interest) with substantial variability within the samples at the levels of the major dispersion metabolites of Raman (for example, glucose, ascorbate, lactate, urea and any exogenous drug or compound present). The concentration of the analyte of interest (in the blood or aqueous humor) must vary by at least a factor of 2, 5, 10 or 20 or more from the sample with the lowest concentration to the sample with the highest concentration . This may require the development of the model that uses subjects with multiple different diseases (eg, kidney failure, diabetes, seizures, mitochondrial myopathies, sickle cell disease, heart failure, blood clots, etc.). For human applications, the model can even be determined with samples of spectra obtained from animals, particularly primates. The humor sample t ^^^^^^^^^^^ ___ ^^^^^^^^^^? _ ^? ^? ___ ^^^^^ tt ^^^? _________________ t_l? _tt? ¿^ t ^? _ The watery can be a natural, human or animal sample, or it can be a sample of manmade or substituted aqueous humor created to simulate natural samples, where the blood level is calculated from a prior knowledge of the relationship between blood levels and aqueous humor levels for the analyte of interest. For example, where the analyte of interest is glucose, the concatenated series of samples for the empirical model should comprise at least 20, 25 or 30 specimens of aqueous humor spectra, with the corresponding blood glucose levels, where the active compounds from Primary Raman in aqueous humor samples (glucose, ascorbate, lactate, urea and preferably any exogenous compound, such as drugs) have a substantial non-colinear variation between samples, with blood glucose concentration or in aqueous humor that varies from 0 to 50 mg / dL to 800 or 1,000 mg / dL, and with a difference of at least 100, 200, 300 or 400 mg / dL in said concentration between the sample with the lowest concentration and the sample with the highest concentration. A schematic diagram of the general apparatus is illustrated in Figure 5. A tunable narrowband laser beam 10 is focused in the anterior chamber 11 of the eye 12 through an objective or ocular lens 13 through the lens 15 and the filter 16, the ray separator 17 and the filter 20. A non-fluorescent objective lens with a suitable numerical aperture is chosen (eg 0.2-0.5), so that the Raman dispersion of the aqueous humor is carried to the maximum while dispersing adjacent structures (for example, crystalline, cornea and iris) is reduced to the maximum. The objective lens must have an adequate working distance to allow the focus of the laser in the middle of the anterior chamber of the eye without touching the cornea. An integrated fixation target projected from the deployment screen 25 projects through the lens 26 through the same objective lens as the laser, but focuses on the retina of the eye. The focus of this target fixation on the retina controls the direction and focus of laser light in the anterior chamber at the same time. The light collected by the objective lens is directed through the holographic notch filters 20, 21 to remove the scattered light from Raleigh. The scattered light of Raman passes through these filters with minimal attenuation and is focused through a small confocal aperture 22 through the lenses 23, 24. The hole and focal point in the anterior chamber of the eye are confocal, so that light from adjacent structures in the eye seeps into this opening. Also, the hole serves as the entrance opening to the spectrometer. The illustrated spectrometer is an imaging spectrograph with a grid 30, a lens 31 and a set of CCD detectors 32. A computer 35 controls the laser 10, the fixation target and the read display 25, and receives the data from the CCD detector 32. The architecture of the spectrometer is just one example of one that is suitable for this application. Many types of spectrographs could be used including Fourier transform spectrographs, spectrographs that use tuneable liquid crystal filters or other tuneable elements. The information can be transmitted to a remote source, such as a ___ tß¿ ?? _ t_hJ¿¿ ^ ^ computer, database, remote medical or similar via modem or other connection through an appropriate line of communication 36 through the Internet, global information network (www), etc. . The CCD detector in the spectrograph is of a thick epitaxial design with a thin posterior wall facing red, so that its sensitivity is optimal in the spectral region of 700-1100 nanometers. Other types of very sensitive detectors may also be suitable. As already mentioned, a digital computer processes the output of the CCD detector and controls the frequency and energy of the laser source. The computer changes the frequency of the laser to allow fluorescence subtraction as described above. The computer also provides information for a digital display of images in the retina. The patient reads the results of the analysis on the display screen. The fibers 40, 41 can be used to supply the laser beam and collect the scattered Raman light as illustrated in Figure 6 (the same numbers are assigned as components in Figures 5 and 6). In this embodiment, the confocal opening is a circular opening 42 positioned at the end of a multimodal fiber. As with the small opening in Figure 5, the end of said fiber is positioned so that it is confocal with the focal point of the objective lens in the anterior chamber of the eye. A single-mode fiber is used to supply the laser beam in the anterior chamber.

»I i t * * &**, This single-mode fiber ensures a limited diffraction spot size at the focal point of the lens. Said fiber supply and collection system illustrated in Figure 6 can be connected to an alternative detection system illustrated in Figure 7a-b. This alternative detection system will allow the subtraction of the fluorescence spectrum, as well as the selected sampling of the frequencies, more important for the calculation of the glucose concentration. Said alternative detection system will decrease the size and cost of the instrument. The system illustrated in Figure 7 uses one or more fixed frequency or tuneable lasers to illuminate the aqueous humor of the eye using the optical delivery system illustrated in Figure 6. Figure 7a illustrates that each spectral characteristic 50a, 50b, 50c of scattered Raman light is related to the excitation wavelength by a fixed phase shift, usually expressed in wave numbers. The change in the excitation wavelength causes the Raman spectra to change a wavelength, as illustrated by the different feature captured on the D? 2 51 scale of the detector based on the Raman spectra 52 for the frequency of Excitation L ?? in comparison with the Raman 53 spectra for the L? 2 excitation frequency. An apparatus that takes advantage of the above is illustrated schematically in Figure 7b. As components to figures 5 and 6, • j t t S -frtfa. assign similar numbers. The laser controllers and / or tunable electronic components 60 are operatively related to a tuneable laser or a plurality of fixed wavelength lasers 61, 62, 63. A series of one or more bandpass detector / filter elements 65. , 66, 67 are operatively related to the amplifiers, and the analog to digital converter 68 is used to make a sample of the spectrum of the collected light. The bandwidth and wavelength of the center of each filter is chosen to correspond to a Raman spectral peak different from the aqueous humor important for glucose quantification. If this laser is tunable on a nanometer or the like, it would also be possible to subtract the fluorescent components from the acquired spectrum, as already mentioned. The use of a laser with a wider tunable scale would allow the spectra changed by Raman to be scanned or slid through a smaller number of bandpass detectors / filters. Since semiconductor lasers with an extremely large control scale are not readily available, a set of narrow-tunable-scale lasers could be used instead, each with a different center wavelength as the laser means along with a Detector series / bandpass filter for this purpose. For convenience, the systems described herein can be configured, so that the optical components do not need or make contact with the cornea of the eye during use, which many patients consider - * '' "-" • --- - objectionable (for example, providing a mono-ocular eyewash suitable for making contact with the orbit around the eye). The apparatus of the invention can be established as a spectrometer base unit linked by an optical fiber cable to an ocular probe, or as a single integrated unit including optical system, spectrometer, detector, computer and screen. As already mentioned, a visual fixation target, such as a mirror, LED or the like, can be constructed in the optical apparatus to facilitate focusing of the excitation light in the anterior chamber and preserve the stability of the eye. In the embodiment of Figures 5-7, a display screen is used, such as a liquid crystal display. As illustrated in Figure 8, an intermittent fixing target in the form of an hourglass can be displayed 70, focused on the retina, during the acquisition of the spectra by the device. This will help optimize the Raman signal from the aqueous humor and reduce the light exposure to the other structures of the eye. The fixation target is active and unfolds during the time that the laser beam is active and illuminates the accuse humor of the subject. As illustrated in Figure 9, a visible indication of the test results in the form of alphanumeric indicia 71 (or other suitable form, such as a graphical display) can be displayed on the same screen just after the acquisition step. The indications in Figure 9 provide a reading for all active Raman compounds in the aqueous humor, but only one needs to be deployed.

What is it? What is it? What is it? What can I say? mimi? -pn MY ifUÉnt iH ÜMfflfÉirif The instrument of the invention can be operatively related to an insulin pump for patients (for glucose D) or with a dialysis machine (for urea) by electrical or fiber optic lines, transmitters and receivers of radiofrequency or the like to provide information on the appropriate analyte, which can then be used to control that apparatus and increase or decrease the output of the insulin pump in response to blood glucose levels, or regulate the dialysis machine. of telemedicine, the instrument of the invention can be operably related for convenience with a remote reading terminal through a computer, modem, internet connection or other communication line with any of the appropriate means (such as electrical or fiber lines). optical, radio frequency transmitters and receivers, etc.) to provide information about the analyte of blood to a doctor or provider of remote medical services (for example, through the Internet or global information network). The present invention is explained in more detail in the following non-limiting examples, wherein "μL" means microliters; "dL" means deciliters, "mW" means milliwatts, "nm" means nanometers, "kg" means kilograms, "J" means joules, "cm2" means square centimeters, and temperatures are given in degrees centigrade. _tj ^^^^^^ EXAMPLE 1 Measurement of glucose in aqueous humor in vitro with Raman spectroscopy Other researchers obtained the aqueous humor of sixteen New Zealand white rabbits in one minute of sacrifice, these animals had suffered infarction to experimental myocardium 48 before euthanasia. They were sacrificed by rapid exsanguination under ketamine and xylazine anesthesia. The aqueous humor samples were kept frozen until glucose levels could be measured and Raman spectroscopy performed. The glucose concentration was measured with a commercial glucometer (Elite Glucometer, Bayer, Elkhart, Indiana, E.U.A.) and was confirmed against concentration standards. Each measurement was repeated, and the average measurement was considered the actual glucose concentration. The samples were placed in conical quartz cuvettes designed to retain a volume of 80 μL and allow direct optical access to the solution by means of the spectrometer without crossing the coverslips or glass walls. The Raman spectroscopy was performed with a holographic imaging spectrograph f / 1.8 (Kaiser Optical Systems, Ann Arbor, Michigan, USA) attached to an Olympus BX60 microscope with a 10X objective. The data was collected using a Princeton Instrumets camera (Trenton, New Jersey, E.U.A.) with a CCD set of 1024 x 256 (EEV, United Kingdom) cooled to -80 ° C with liquid nitrogen. The illumination of the sample through the microscope objective was achieved with a Ti: sapphire laser (Spectra Physics 3900S, Mountain View, California, USA) pumped by an argon laser (Spectra Physics 201 OE). Spectrographic data were integrated while the sample was illuminated at a wavelength of 785.2 nm (30 mW) for 10 seconds. This was then repeated at a wavelength of 787.2 nm. The spectra integrated into the two slightly different wavelengths were then subtracted from each other. This eliminates with efficiency the broadband fluorescence that does not change in relation to the excitation wavelength. This allows the spectra changed by Raman to appear as a bipolar pattern, (figure 1) The multivariate analysis of the spectra was performed using software packages Holograms (Princeton Instrumets, Trenton, NJ) and Grams (Galactic Industries, Salem, New Hampshire E.U.A.). Thirty-two samples of aqueous humor (of sixteen rabbits) were evaluated using a "circular" approach to group them in an iterative manner except for one of the samples in the concatenated series. Consequently, the system concatenates all, except one of the samples, estimates the glucose level in that shows, then rotates the test sample in the concatenated series. This cycle is repeated until all the samples have served as an unknown test sample.

Subsequently, a backpropagation neural network (D. Rumelhart et al., Nature 323, 533 (1986)) was used to determine whether a non-linear regression method would better predict the glucose concentration of the aqueous humor Raman spectra. This model can compensate 5 interactions still unknown between analytes in aqueous humor. A two-layer back-propagation neural network (Neutral Ware, Inc., Pittsburgh, Pennsylvania, USA) was employed using a sigmoid function as the non-linear element. The factors derived from the algorithm of partial least squares served as the entrance to the neural network. It could have been resorted to pure spectral data, but would have required a prohibitively large concatenated series. As in the linear regression performed previously, tests and circular concatenation were used. During the concatenation, the weights of the neural network were adjusted to reduce to the maximum the total square error between the concentration of actual glucose and glucose concentrations predicted. Each sample was tested using a concatenated neural network in the remaining 31 samples.

EXAMPLE 2 Correlation of glucose in aqueous humor with elevation of glucose in blood Nine New Zealand white rabbits were used for this part of the study. They were anesthetized with ketamine (50 mg / kg) and xylazine (7.5 mg / kg) given as a single intramuscular injection. Xylazine blocks the release of insulin from the pancreas and causes an elevation of blood glucose (K. Chalabi et al., Ophthalmic Res. 19, 289 (1987); J. Ambjerg et al., Ophthalmic Res. 22, 265 (1990)). Blood samples were taken from the arteries of the central ear at various times after the injection of anesthesia, once the animals were adequately anesthetized. All the blood was measured immediately regarding the glucose concentration with a commercial glucometer (Elite Glucometer, Bayer). In most cases, two measurements were made. The average measurements were reported. If the measurements differed by more than 20%, a third measurement was made. If a measurement differed by more than 20% from the mean, it was discarded. In cases where the concentration of glucose in the blood appeared stable, occasionally only one measurement was made. At various times after the induction of anesthesia, samples of aqueous humor were taken. This was performed by paracentesis with a 25 gauge needle through the cornea near the limbus after the administration of proparacaine eye drops. The glucose concentration in aqueous humor was measured with the glucometer in a manner similar to blood. In rabbits, there is a severe failure of the aqueous blood-humor barrier after a single paracentesis of the anterior chamber of the eye (W. Unger et al., Exp. Eye Res. 20, 255 (1975)). The aqueous humor becomes too viscous to repeat the paracentesis for at least thirty minutes, consequently, the results are reported for the first paracentesis of an eye. However, in some cases, a second paracentesis was performed one hour after assessing how the failure of the blood-aqueous barrier could affect the correlation of glucose in aqueous humor with plasma glucose. The average blood glucose concentrations for each animal were plotted against time after the anesthetic injection. Since it was impossible to obtain blood samples at the same time For a time with samples of aqueous humor, a polynomial of second order of optimum adjustment for each animal was calculated. This was used to calculate the concentration of glucose in the blood at the time of taking the aqueous humor sample. The concentration of glucose in aqueous humor was then recorded on a graph against the concentration of glucose in the blood calculated at the same time for all animals. The DeltaGraph software (Delta point, Inc., Monterrey, California, E.U.A.) was used for all statistical calculations.

EXAMPLE 3 Results for glucose measurement in aqueous humor in vitro with Raman spectroscopy Pure spectra of aqueous humor in rabbits show broad fluorescence peaks that obscure the underlying Raman identification. (Figure 1) When the spectra of the two slightly different wavelengths are subtracted from one another, the spectra changed by Raman come to appear as bipolar peaks. The glucose concentration in real aqueous humor, measured with the glucometer, varies from 37 to 323 mg / dL in the thirty-two samples. The multivariate analysis of the pure spectra of these samples with the partial least squares algorithm revealed a clear correlation (r2 = 0.76) between the predicted aqueous humor glucose concentration and the actual concentration. The multivariate analysis of the spectra subtracted from these samples with the partial least squares algorithm resulted in an improved correlation (r2 = 0.90) between the predicted aqueous humor glucose concentration and the actual concentration. When the backpropagation with an artificial neural network is applied in addition to the data, the correlation is excellent (1 ^ = 0.98). (Figure 2).

EXAMPLE 4 Results for the correlation of glucose in aqueous humor with elevation of blood glucose The rate of increase in blood glucose after injection of xylazine varies greatly from animal to animal. (Figure 3) In fact in a rabbit there was little change in the concentration of glucose in the blood over time. If only the samples taken in the first 10 fifteen minutes after the anesthetic injection are considered, the glucose concentration in an aqueous humor is higher than the blood glucose concentration (207 ± 28 mg / dL for aqueous humor; 27 mg / dL for blood). When the calculated glucose in the blood rises above 200 mg / dL, the glucose in simultaneous aqueous humor becomes parallel with the blood glucose almost linearly [glucose in aqueous humor = 1.18 (blood glucose) - 72.7; 1 ^ = 0.88]. (Figure 4). The aqueous humor sample number for paracentesis was insufficient to draw conclusions about its importance. However, they did not appear to correlate well with aqueous glucose levels of the initial paracentesis. (Figure 4). These data indicate that the baseline ratio of glucose in aqueous humor to blood glucose in rabbits is approximately 1.5. (Actually, it may be greater than this since a real baseline was not obtained, and it was likely that blood glucose levels would rise at the time of obtaining the first samples, even in the first 15 minutes after injection). This is much higher than in humans and much higher than in previous reports in rabbits. However, all previous reports in rabbits were prior to the recognition of the hyperglycemic effect of xylazine anesthesia, although many of them used xylazine anesthesia. In addition, most of the previous reports assumed that the rabbits suffered a constant state euglycemia without actually confirming that this was the case. These facts can count for some of the results that are too variable in previous reports. These data also show that glucose in aqueous humor in the rabbit responds almost immediately, once the blood glucose exceeds 200 mg / dL. The ratio of glucose in aqueous humor to blood glucose is almost linear, while blood glucose rises above 200 mg / dL. Below that level, glucose in aqueous humor seems stable. It has not yet been determined what happens with glucose in aqueous humor when blood glucose is set at a hyperglycemic level, or when the blood glucose concentration falls. However, the balance of glucose in aqueous humor with blood glucose probably occurs in a matter of minutes in rabbits. If the rapid equilibrium of glucose in aqueous humor also occurs in humans, it could serve as an excellent substrate for non-invasive glucose monitoring.

Previous investigators have not discovered any failure in the blood-aqueous barrier of albino rabbits exposed to infrared radiation with energy densities up to 10G J / cm (D. Reddy, supra; G. Peyman et al., Exp. Eye Res. 42, 249 (1986), T. Kurnik et al., Inv. Ophthalmol, Vis. Sci. 30, 717 (1989)). However, low densities of infrared energy at 44 J / cm2 may be sufficient to give a failure in the blood-aqueous barrier in pigmented rabbits. This is even higher substantially than the energy density that would need to be applied with the Raman technique described herein.

EXAMPLE 5 Measurement of integrity of blood-aqueous and blood-brain barriers The blood-brain and blood-aqueous humor barriers block the passage of large molecules in the cerobroespinal fluid or aqueous humor. Many drugs and procedures against diseases cause alteration of the blood-aqueous barrier and the blood-brain barrier. In the case of said drugs or procedures against diseases, alteration of the blood-brain barrier can be inferred from the failure of the blood-aqueous barrier. Failure of the aqueous-blood humor barrier can be measured by measuring the protein content of the aqueous humor or by measuring the concentration of other substances (eg, drugs) within the aqueous humor. Said substances can be quantified by Raman spectroscopy and the protein can be calculated from the fluorescence spectrum, since the fluorescence spectrum is generated largely by protein. Accordingly, the device described herein for measuring glucose and other Raman scattering metabolites in the eye can be used to measure the integrity of the blood-aqueous and blood-brain barriers. Different size Raman dispersion molecules that do not normally cross the blood-aqueous or blood-brain barrier can be administered to a patient intravenously. The presence of these molecules is then quantified by Raman spectroscopy of the anterior chamber of the eye. This can be used to determine the size of molecules that pass through the blood-aqueous barrier. The amount of fluorescence that is subtracted to reveal the 15 Raman spectra of these substances, with the device described herein, would reflect the passage of natural proteins through the blood-aqueous and blood-brain barriers. Therapeutic drugs can also be measured to determine their effectiveness when crossing the blood-aqueous or blood-brain barrier. The foregoing illustrates the present invention and does not constitute a limit thereof. The invention is defined by the following claims, with the equivalents of the claims that will be included therein. ¿Í í * > ** - »*"

Claims (27)

NOVELTY OF THE INVENTION CLAIMS

1. - A non-invasive method for determining blood glucose levels in a subject comprising: generating an excitation laser beam at a wavelength of 700 to 900 nanometers; focusing said excitation laser beam on the anterior chamber of an eye of said subject so that the aqueous humor in said anterior chamber is illuminated; detecting a Raman spectrum of said illuminated aqueous humor; subtracting a fluorescence spectrum for said aqueous humor from said Raman spectrum to produce a spectrum of difference; and then determining the blood glucose level for said subject of said difference spectrum.

2. A method according to claim 1, further characterized in that said laser beam pulse lasts no more than thirty seconds.

3. A method according to claim 1, further characterized in that said laser beam pulse has an instantaneous power not greater than 30 milliwatts.

4. A method according to claim 1, further characterized in that said step of focusing is performed so that the energy in the retina of the eye of said excitation laser beam pulse is not greater than 3000 milliwatts per square centimeter.

5. - A method according to claim 1, further characterized in that it comprises the step of subtracting the spectrum of adjacent tissues from said Raman spectrum.

6. A method according to claim 1, further characterized in that said subject is a human subject.

7. A method according to claim 1, further characterized in that said determination step is performed with an empirical model of real test experience.

8. A method according to claim 7, further characterized in that said empirical model is generated by multivariate analysis.

9. A method according to claim 8, further characterized in that said multivariate analysis is a partial least squares analysis.

10. A method according to claim 7, further characterized in that said empirical model is produced with a concatenated series comprising at least 20 samples of aqueous humor, a Raman spectrum corresponding to each of said samples, and a level of blood glucose corresponding to each of said samples; with the glucose concentration in the aqueous humor samples or corresponding blood samples that vary at least 200 mg / dL from the sample with the lowest concentration to the sample with the highest concentration. ^ ¡G ^

11. - A method according to claim 10, further characterized in that the concentration of the main Raman active compounds in said samples varies substantially among said non-colinear samples.

12. An apparatus for the non-invasive determination of a blood level of an analyte of interest in a subject, comprising: a laser for generating an excitation laser beam; an optical system operatively related to said laser for directing said excitation laser beam in the anterior chamber of an eye of said subject so that the aqueous humor in said anterior chamber has sufficient illumination to detect scattered Raman light; a detector operatively related to said optical system and configured to detect a Raman spectrum of said aqueous humor; means for subtracting a fluorescence spectrum for said aqueous humor from said Raman spectrum to produce a spectrum of difference; and a processor for determining the blood level of said analyte of interest to said subject of said difference spectrum; and a target of deployment projected through said optical system that visually displays said blood level of said analyte of interest.

13. An apparatus according to claim 12, further characterized in that it comprises a fixation target operatively related to and projected through said optical system to ensure proper direction and focus of the laser beam in the anterior chamber of the eye. itwitfMü AMa ^ _ _m ^ m ^ _m? MM ^ and Éa ^

14. - An apparatus according to claim 12, further characterized in that said optical system is a confocal optical system.

15. An apparatus according to claim 12, further characterized in that said laser beam has a wavelength of 700 to 900 nanometers.

16. An apparatus for the non-invasive determination of a blood level of an analyte of interest in a subject, comprising: a laser for generating an excitation laser beam; an optical system operatively related to said laser for directing said laser beam and excitation in the anterior chamber of an eye of said subject, such that the aqueous humor in said anterior chamber is illuminated sufficiently to detect scattered Raman light; a detector operatively related to said optical system and configured to detect a Raman spectrum of said aqueous humor; means for subtracting a fluorescence spectrum for said aqueous humor from said Raman spectrum to produce a spectrum of difference; a processor for determining the blood level of said analyte of interest for said subject of said difference spectrum, said processor including an empirical model of real test experience.

17. An apparatus according to claim 16, further characterized in that said empirical model is generated by multivariate analysis.

18. - An apparatus according to claim 17, further characterized in that said multivariate analysis is a partial least squares analysis.

19. An apparatus according to claim 16, further characterized in that said empirical model is produced with a concatenated series comprising at least 20 samples of aqueous humor, a spectrum of Raman corresponding to each of said samples, and a blood level of said analyte of interest corresponding to each of the samples; with the concentration of said analyte of interest in the aqueous humorous samples or corresponding blood samples which vary by at least a factor of 5 from the sample with the lowest concentration to the sample with the highest concentration.

20. An apparatus according to claim 16, further characterized in that it comprises a communication line 15 connected to said processor for transmitting said blood level of said analyte of interest to a remote location.

21. An apparatus according to claim 16, "Further characterized in that said optical system is an optical system * confocal

22. An apparatus according to claim 16, further characterized in that said laser beam has a wavelength of 700 to 900 nanometers. • taMhtíttlkíi

23. - An apparatus according to claim 16, further characterized in that said analyte of interest is glucose.

24. An apparatus for the non-invasive determination of a blood level of an analyte of interest in a subject, comprising: a laser means for generating a plurality of excitation laser beams at a plurality of different wavelengths; an optical system operatively related to said laser means and configured to direct said excitation laser beam in the anterior chamber of an eye of said subject, such that the aqueous humor in said anterior chamber has sufficient illumination to detect scattered light of Raman; a detector means operatively related to said optical system and configured to direct a Raman spectrum of said aqueous humor in a plurality of detector lengths; a control means operatively related to said laser means for sliding said Raman spectrum through said plurality of wavelengths of the detector; means for subtracting a fluorescence spectrum for said aqueous humor from said Raman spectrum to produce a spectrum of difference; and a processor for determining the blood level * of said analyte of interest for said subject of said spectrum of s difference.

25. An apparatus according to claim 24, further characterized in that said laser means comprises a tunable laser. ^^ tófafejAlg ^^^^^ j ^ j ^ j ^ j ^ j ^^ j ^^^^^^^^^^^^^ j ^^^^^^^^^^^^^^^^ ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ J ^^^ j ^

26. - An apparatus according to claim 24, further characterized in that said laser means comprises a plurality of fixed frequency lasers.

27. An apparatus according to claim 24, further characterized in that said optical system is a confocal optical system. go