HK1103004B - Determination of a measure of a glycation end-product or disease state using tissue fluorescence - Google Patents

Abstract

A method of determining a measure of a tissue state (e.g., glycation end-product or disease state) in an individual. A portion of the tissue of the individual is illuminated with excitation light, then light emitted by the tissue due to fluorescence of a chemical with the tissue responsive to the excitation light is detected. The detected light can be combined with a model relating fluorescence with a measure of tissue state to determine a tissue state. The invention can comprise single wavelength excitation light, scanning of excitation light (illuminating the tissue at a plurality of wavelengths), detection at a single wavelength, scanning of detection wavelengths (detecting emitted light at a plurality of wavelengths), and combinations thereof. The invention also can comprise correction techniques that reduce determination errors due to detection of light other than that from fluorescence of a chemical in the tissue. For example, the reflectance of the tissue can lead to errors if appropriate correction is not employed. The invention can also comprise a variety of models relating fluorescence to a measure of tissue state, including a variety of methods for generating such models. Other biologic information can be used in combination with the fluorescence properties to aid in the determination of a measure of tissue state. The invention also comprises apparatuses suitable for carrying out the method, including appropriate light sources, detectors, and models (for example, implemented on computers) used to relate detected fluorescence and a measure of tissue state.

Description

Determination of a glycation end-product or disease state using tissue fluorescence

CROSS-REFERENCE TO RELATED APPLICATIONS

This application claims U.S. patent application No. 10/116,272 entitled "Apparatus And Method For spectral Analysis Of Tissue To Detect reagents In Ind, filed prior To 2002, 4/4 (incorporated herein by reference) And U.S. provisional application No. 60/515,343 entitled" Determination Of a measurement Of a catalysis End-product Using Tissue Fluorescence, "filed 10/28/2003 (incorporated herein by reference), In accordance with 35 U.S. C120.

Technical Field

The present invention relates generally to the determination of tissue state from tissue fluorescence. More particularly, the invention relates to a method and a device for determining a model relating tissue fluorescence to tissue state and a method and a device for determining the fluorescence properties of tissue, as well as a method and a device for determining a tissue state from the fluorescence properties and a suitable model.

Background

Diabetes is a major health problem that is prevalent in the united states and in developed countries of the world, as well as in developing countries. In 2002, the American Diabetes Association (ADA) estimated that 1820 million americans (with feet accounting for 6.4% of citizens) had some form of diabetes. Of these, 90 to 95% suffer from type II diabetes, while 35% (or about 600 million individuals) are undiagnosed. See ADA report, Diabetes Care, 2003. The World Health Organization (WHO) estimates that 17,500 million people worldwide have diabetes; type II diabetes also accounts for 90% of all diagnosed patients worldwide. Unfortunately, predictions show that this severe situation worsens over the next 20 years. WHO predicts that the total number of diabetic patients will double before 2025. Similarly, ADA is expected to infect this disease by 8.0% of the us population (approximately 2,500 million individuals) in 2020. Assuming that the detection rate is not before, this indicates that there are 3 "silent" diabetics in every 100 americans in less than 20 years. It is not surprising that many people have described worldwide outbreaks of diabetes as epidemics. Diabetes causes a significant impact on personal health and national economy. The U.S. health care costs associated with diabetes in 2002 exceed $ 1,320 million. Since chronic hyperglycemia can cause numerous complications, the overhead is distributed across a variety of public health care services. For example, 5% to 10% of all expenditures in the united states in the areas of cardiovascular disease, kidney disease, endocrine and metabolic complications, and ocular disorders are attributed to diabetes. See ADA report, Diabetes Care, 2003. These economic and health burdens mask the fact that most complications associated with diabetes are preventable. Clinical studies on Diabetes Control and Complications Trial (DCCT) with milestone significance confirm that a strict regimen involving glucose monitoring, exercise, proper diet and insulin treatment can significantly reduce exacerbation of and risk of developing diabetic Complications. See DCCT research group, NEng J Med, 1993. Furthermore, the Diabetes Prevention Program (DPP) being implemented has demonstrated that individuals at risk for Diabetes can significantly reduce their chances of suffering from this disease by changing lifestyle, such as losing weight and increasing physical activity. See DPP research group, NEng J Med, 2002. ADA has suggested that a healthcare provider could screen individuals with one or more disease risk factors, following: "if DPP demonstrates that one or more [ tested ] interventions can reduce the incidence of type II diabetes, it is reasonable to screen … … more extensively". See ADA Statement of Position status, Diabetes Care, 2003.

The Fasting Plasma Glucose (FPG) test is one of two accepted clinical criteria for diagnosing or screening for diabetes. See ADA Association Report (Committee Report), Diabetes Care, 2003. The FPG test is a carbohydrate metabolism test that measures plasma glucose levels after 12 to 14 hours of fasting. Fasting stimulates the release of the hormone glucagon, which in turn raises plasma glucose levels. In non-diabetic individuals, the body produces and processes insulin to counteract the rise in glucose levels. In diabetic individuals, plasma glucose levels remain high at all times. ADA recommends that the FPG test be performed in the morning because afternoon testing tends to produce lower readings. In most healthy individuals, FPG levels will drop to between 70 and 100mg/d 1. Drugs, exercise and recent illness can affect the outcome of this test, so the medical history should be properly understood before the test is conducted. FPG levels of 126mg/d1 or higher indicate the need for subsequent retesting. If the same level is reached upon retesting, a diagnosis of diabetes is typically made. Measurements only slightly above the normal range require additional tests, including the Oral Glucose Tolerance Test (OGTT) or the postprandial plasma glucose test, to confirm the diagnosis of diabetes. Other conditions that may lead to increased outcomes include pancreatitis, Cushing's syndrome, liver or kidney disease, convulsions, and other acute diseases such as sepsis or myocardial infarction.

Because the FPG test is easier and more convenient for the patient to conduct, this test is strongly recommended by ADA and its range of application is broader than the other acceptable diagnostic standard OGTT. Even with various shortcomings, OGTT remains the clinical gold standard for diabetes diagnosis. Following the onset of hunger, the patient is administered an oral dose of glucose solution (75 to 100 grams of dextrose) which typically causes the blood glucose level to rise within the first hour and return to baseline within three hours, since the body produces insulin to normalize the glucose level. Blood glucose levels can be measured 4 to 5 times during a 3 hour administration of OGTT. On average, the levels typically peak to 160 to 180mg/dl 30 minutes to 1 hour after administration of oral glucose and then return to fasting levels of 140mg/dl or less in 2 to 3 hours. Factors such as age, weight, and race can affect outcomes, such as recent diseases and certain medications. For example, an elderly individual over the age of 50 years has an upper limit of 1mg/dl of increased annual glucose tolerance. Current ADA guidelines indicate that diabetes can be diagnosed if the post-2 hour loading blood glucose values in two OGTT's conducted separately on different days are greater than 200 mg/dl.

In addition to these diagnostic criteria, ADA also recognizes two "pre-diabetic" conditions that reflect off-normoglycemia, which, although abnormal, are insufficient as criteria for the diagnosis of diabetes. An individual is considered to have "impaired fasting glucose" (IFG) when the individual's FPG test falls between 100 and 126 mg/dl. Similarly, "impaired glucose tolerance" (IGT) is typically diagnosed when the 2 hour post-loading glucose value of the OGTT is between 140 and 200 mg/dl. Both of these disorders are considered risk factors for Diabetes and IFG/IGT is used as inclusion criteria in the Diabetes Prevention Program (Diabetes Prevention Program). IFG/IGT is also associated with increased risk of cardiovascular disease.

Collectively, such tests are inconvenient and expensive to administer to patients when performing OGTT and FPG tests due to the need for fasting, invasive blood draw and repeated tests for multiple days prior to testing. In addition, the diagnostic accuracy of such tests has had much room for improvement. See, e.g., m.p. stern et al, Ann Intern Med, 2002 and j.s.yudkin et al, BMJ, 1990. Various attempts have been made in the past to avoid the disadvantages of FPG and OGTT in diabetes screening. For example, risk assessment based on patient history and pen and paper testing has been attempted, but such techniques often result in poor diagnostic accuracy. In addition, diabetic screening using glycated hemoglobin (HbAlc) has been proposed. However, because HbAlc is an average glycemic indicator over a period of weeks, its inherent variability combined with experimental uncertainties associated with currently available HbAlc analyses makes HbAlc rather poor as an indicator of diabetes. See ADA association report, Diabetes Care, 2003. HbAlc levels in diabetic patients may overlap with HbAlc levels in non-diabetic patients, making HbAlc problematic as a screening test. There is a need for a reliable, convenient and cost-effective method for screening for diabetes. Also, a reliable, convenient and cost-effective method for measuring the effects of diabetes would be helpful in treating the disease and avoiding complications thereof.

U.S. patent No. 5582168 (Samuels) discloses an apparatus and method for measuring properties of biological tissues and similar materials. The devices and methods described relate to human eye measurements. Furthermore, the calibration methods described by these inventors only involve the measurement of elastically scattered excitation light. Samuels describes a simple linear correction technique. Samuels does not disclose an algorithm or method by which tissue disease status can be discriminated through non-invasive measurements.

U.S. patent No. 6505059 (Kollias) discloses an apparatus and method for non-invasive tissue glucose level monitoring. Kollias does not describe any method by which the tissue absorption and scattering effects of the measured fluorescence can be corrected. While Kollias indicates that tissue scatter can be directly measured by performing tissue reflectance measurements, it does not indicate how one can use this information to obtain information about tissue fluorescence spectra. Moreover, Kollias does not disclose an algorithm or method by which the status of tissue disease can be determined from non-invasive measurements.

U.S. patent No. 6571118 (Utzlnger) discloses a method and apparatus for performing fluorescence and spatially resolved reflectance spectroscopy on a sample. Although utsinger describes a technique in which biological tissue properties are determined using a combination of fluorescence and reflectance measurements, the application is not related to skin spectroscopy. Furthermore, the reflectance measurements described by utsinger are spatially resolved in nature, i.e., reflectance spectroscopy can be performed at one or more specific light source-receiver spacings. Finally, no algorithm or method is described in which the measured fluorescence can be corrected using tissue reflectance measurements to obtain or approximate the intrinsic fluorescence spectrum of the tissue.

U.S. patent application No. 20030013973 (Gaorgakoudi) discloses a system and method for fluorescence, reflectance, and light scattering spectroscopy for measuring tissue properties. Georgakoudi discusses the use of reflectance properties to assess intrinsic fluorescence (e.g., for the detection of esophageal cancer and Barrett's esophageal). Georgakoudi does not teach any specific technique for such assessment.

U.S. patent No. 6088606 (Ignotz) discloses a system and method for determining the duration of a medical condition. Ignotz mentions fluorescence, but does not use reflectance spectra to obtain or evaluate intrinsic fluorescence spectra. Furthermore, Ignitz describes methods for determining the duration of a disease rather than diagnosing or screening for the presence of a disease or quantifying the concentration of a particular chemical analyte. Finally, Ignotz does not suggest that skin can be used as a measurement site.

U.S. patent No. 5601079 (Wong) describes an apparatus for non-invasively quantifying glucose control, aging, and advanced Maillard reaction products via stimulated fluorescence. Wong specifically quantifies advanced glycation end products in the blood rather than in the skin and/or its structural proteins. Furthermore, fluorescence correction methods only involve measurement of elastically scattered excitation light. Wong only describes a simple linear correction technique. Finally, Wong does not disclose an algorithm or method by which tissue disease status can be discriminated via non-invasive measurements.

International patent application No. WO 01/22869 (Smits) describes a device for measuring skin autofluorescence in a non-invasive manner. The device consists of a broadband uv light source (blacklight) illuminating the skin through an interchangeable optical band pass filter. The resulting skin fluorescence was coupled to a pocket spectrometer via an optical fiber. The application proposes that AGE concentration in skin can be inferred from quantitative assessment of skin autofluorescence, but it does not describe any method by which AGE content can be quantified using the device and measurement techniques. The device is intended to be used to assess skin fluorescence in healthy individuals and the use of the device in disease determination is not proposed. The application states that individual skin color and texture may be a measure of interference, but it does not mention techniques or methods that can compensate for these variable characteristics.

Disclosure of Invention

The present invention provides a method for determining the status of a tissue in a body. A portion of the individual's tissue is illuminated with excitation light, and light emitted by the tissue due to fluorescence emitted by a chemical within the tissue in response to the excitation light is detected. The detected light can be combined with a model correlating fluorescence to disease status to determine the disease status of the individual. The present invention may include excitation light of a single wavelength, excitation light scanning (illuminating tissue at a plurality of wavelengths), detection at a single wavelength, detection wavelength scanning (detecting emitted light at a plurality of wavelengths), and combinations thereof. The present invention may also include calibration techniques that reduce errors in the measurement due to light detection rather than from fluorescence of chemicals within the tissue. For example, tissue reflections can cause errors if proper correction is not employed.

The invention may also include a variety of models that enable fluorescence to be correlated with disease states, including a variety of methods for forming such models. Other biological information may be used in combination with the fluorescence properties to help determine the tissue state, including, for example, the age of the individual, the height of the individual, the weight of the individual, the family history of the individual, ethnicity, skin melanin levels, or a combination thereof. Raman or near infrared spectroscopy may also be used To provide additional information, for example, as set forth In U.S. patent application No. 10/116,272 (entitled "Apparatus And Method for spectroscopic Analysis Of Tissue To Detect Diabetes In And Ind manual", filed on 4.4.2002). The invention also includes apparatus suitable for carrying out the method, including suitable light sources, tissue sampling devices, detectors, and models (e.g., implemented on a computer) for correlating the detected fluorescence with disease states.

As used herein, "determining a disease state" includes determining the presence or likelihood of onset of diabetes; the degree of diabetes progression; changes in the presence, likelihood of onset, or progression of diabetes; the likelihood of having, not having, forming, or not forming diabetes; the presence, absence or likelihood of onset of diabetic complications. "Diabetes" includes a variety of glucose regulating conditions including type I, type II and gestational Diabetes, other types of Diabetes recognized by the american Diabetes association (see ADA committee report, Diabetes Care, 2003), hyperglycemia, impaired fasting glucose, impaired glucose tolerance, and pre-Diabetes. "tissue reflectance properties" include any tissue reflectance property that can be used to correct for detected light, including, for example, tissue reflectance at the fluorescence excitation wavelength, tissue reflectance at the fluorescence emission wavelength, and tissue reflectance at other wavelengths that can be used to assess the tissue's intrinsic fluorescence spectrum. "chemical change due to glycemic control" means any change in histochemical properties due to glycemic control, examples including measurement of concentration, presence of glycation end products in the tissue, concentration, or change in concentration; a measure of the rate of accumulation or change in the rate of accumulation of the end product; (ii) a measure of the thickness of the tissue membrane or the change, rate of change or direction of change in the thickness; a property of the tissue, such as tensile strength, strain, or compressibility, or a change, rate of change, or direction of change in the property. "amount of glycation end-product" means any measure of the presence, time, extent, or status of tissue associated with hyperglycemia, e.g., including a measure of the presence, concentration, or change in concentration of glycation end-product in the tissue; a measure of the rate of accumulation and rate change of the end product; a measure of the presence, intensity, or change in intensity of fluorescence at a wavelength known to be associated with tissue glycation endproducts; and a measure of the rate of accumulation or rate change of said fluorescence. "determination of a tissue state" includes determination of a disease state, determination of a chemical change caused by glycemic control, determination of a measure of glycation end products in a tissue, or a combination thereof. It should be understood that when it is said that light has a "single wavelength," the light may actually comprise light at a plurality of wavelengths, but it should also be understood that most of the energy in the light is transmitted at a single wavelength or over a range of wavelengths close to a single wavelength.

Drawings

The drawings, which are not necessarily to scale, depict exemplary embodiments and are not intended to limit the scope of the invention.

FIG. 1 is a graph of an excitation spectrum in which the excitation wavelength is scanned in the range of 315 to 385nm and the emitted fluorescence is measured at a fixed wavelength of 400 nm.

FIG. 2 is a graph of emission scan data in which the excitation is fixed at 325nm and fluorescence is monitored by scanning the detection subsystem in the range of 340 to 500 nm.

Fig. 3 is a graphical representation of the interpolated variance of the measured spectrum (solid line, "uncorrected") and the intrinsically-corrected spectrum (dashed line, k 0.5, n 0.7) of the spectrum of fig. 1 and 2.

FIG. 4 is a schematic diagram of the model construction steps typically employed when the ultimate goal is to use a model to assess tissue disease state.

FIG. 5 is a graphical representation of the manner in which the discrimination function can achieve optimal separation between the two groups.

FIG. 6 is a graphical representation of a data set and corresponding wavelength ranges.

FIG. 7 is a box and whisker plot of the cross-validation posterior probability of diabetes class membership for all study participants.

Figure 8 is a graphical representation of a subject's working profile associated with the present invention and a subject's working profile associated with a fasting plasma glucose test.

Figure 9 is a graphical representation of a cross-validation result in which all data from a single study participant is used in round-robin fashion in each iteration.

Figure 10 is a graphical representation of a subject's working profile associated with the present invention and a subject's working profile associated with a fasting plasma glucose test.

FIG. 11 is a diagram of a device component or subsystem of the present invention.

FIG. 12 is a diagram of an exemplary skin fluorometer.

FIG. 13 is a schematic view of a portion of an apparatus of the present invention.

FIG. 14 is a schematic view of a portion of an apparatus of the present invention.

FIG. 15 is a schematic view of a tissue interface suitable for use with the present invention.

FIG. 16 is a schematic representation of a multi-channel fiber optic tissue probe geometry.

FIG. 17 is a schematic illustration of a circular arrangement of multi-channel fiber optic tissue probes.

FIG. 18 is a schematic illustration of a linear arrangement of a multi-channel fiber optic tissue probe.

FIG. 19 is a diagrammatic cross-sectional view of a portion of a multi-channel fiber optic tissue probe in a vertical arrangement.

FIG. 20 is a diagrammatic cross-sectional view of a portion of a multi-channel fiber optic tissue probe in an angled arrangement.

FIG. 21 is a diagrammatic cross-sectional view of a portion of a multi-channel fiber optic tissue probe in an angled arrangement.

FIG. 22 is an isometric perspective view of a fiber optic tissue probe.

FIG. 23 is a schematic representation of a multi-channel fiber optic tissue probe seeking tissue volumes at various excitation and receiver spacings.

Detailed Description

Exposure of proteins to glucose typically results in non-enzymatic saccharification and sugar oxidation, a process known as the maillard reaction. The stable end products of the maillard reaction are collectively referred to as advanced glycation end products (AGEs). In the absence of significant clearance, the AGEs aggregate at a rate proportional to the mean blood glucose level. The maillard reaction can be regarded as a regular occurrence in a healthy state and an accelerated appearance of the aging process in diabetic patients due to the presence of chronic hyperglycemia. In skin, collagen is the most abundant protein and is easily subject to glycation. Skin collagen AGEs are typically in the form of fluorescent cross-links and adducts; pentosidine (cross-linker) and carboxymethyl lysine (CML, adduct) are two well studied examples of skin-collagen AGEs. Other AGE examples include fluorobis (fluorolink), pyrraline, crosline, N' - (2-carboxyethyl) lysine (CEL), glyoxal-lysine dimer (GOLD), methylglyoxal-lysine dimer (MOLD), 3DG-ARG imidazolone, vesperysine a, B, C, and threosidine. One commonly used measure of aggregated AGE production and associated collagen crosslinking is the level of collagen-related fluorescence (CLF). CLF is typically measured by monitoring chemically separated collagen in the range of 400 to 500nm in vitro with a fluorescence emitter after excitation at or near 370 nm. See Monnier, NEJM, 1986.

Relatively long half-life (t) of skin collagen_1/215 years) the fluorescent properties of AGEs associated with many of them make these substances potential indicators of tissue blood glucose accumulation. CLF intensity and levels of specific skin AGEs are associated with the occurrence and severity of end-organ diabetic complications such as joint stiffness, retinopathy, nephropathy and arterial stiffness. See Buckingham, Diabetes Care, 1984; buckingham, J Clin Invest, 1990: monnier, NEJM 1986; monnier, J Clin Invest 1986; sell, Diabetes, 1992. To date, in a number of such studies, the DCCT Skin Collagen-aided research Group (DCCT Skin Collagen research Group) has evaluated a variety of Skin Collagen variables by needle biopsies donated by most Study participants. These researchers found that skin AGE is closely related to the occurrence and clinical grade of diabetic neuropathy, nephropathy and retinopathy. See Monnier et al, Diabetes, 1999.

The present invention can use one or more non-invasive fluorescence measurements to determine the diabetic status of a subject. The present invention may illuminate a portion of an individual's tissue (e.g., a portion of skin) with excitation light and detect fluorescence emitted by the tissue. The fluorescence measurements may include at least one set of excitation and emission wavelengths corresponding to the CLF window described above. The fluorescent properties may convey information about the disease state of the tissue in the sought state. The present invention can apply additional processing algorithms to the measured fluorescence, followed by providing simple numerical thresholds or more detailed mathematical models to correlate the optical information with the disease state. In other embodiments, the output of the thresholding method or mathematical model may be a quantitative measure of diabetes-induced chemical changes in the tissue of the individual that are not taken into account when measuring the chemical changes. In other embodiments, the present invention may use quantitative measures of diabetes-induced chemical changes to further infer or classify the diabetic condition in the individual undergoing the measurement.

Determining fluorescence properties of tissue

Tissue fluorescence is initiated when tissue is illuminated with light that causes electrons in various molecular species to reach an excited level. Some of the excited molecules decay in emission, emitting light with electrons returning to lower energy states. The released fluorescence always has a longer wavelength (lower photon energy) than the excitation light. The absorption and fluorescence spectra of biomolecules are typically broad and overlap. Most tissues absorb a wide range of wavelengths. The emitted fluorescence spectrum is typically correspondingly broad for a given excitation wavelength. Several factors can affect the usable range of excitation and emission wavelengths. The fluorescing substance (e.g., pentostatin) is generally most absorbing in the UVA (315 to 400nm) range and is released in the UVA to the entire short wavelength visible light range (340 to 500 nm). The long wavelength limit of the excitation and emission range is typically influenced by the electronic structure of the fluorescing component. The shortest practical excitation wavelength will be limited to the UVA or longer wavelength range for optical safety reasons. The threshold limit for optical exposure is significantly reduced for wavelengths below 315 nm. Therefore, the safe exposure time for wavelengths in the UVB (280 to 315nm) range may be too short for efficient spectral data acquisition.

If the spectral selectivity of the excitation or emission portion of the fluorometer is relatively coarse, then only gross tissue information regarding biochemistry and morphology will be obtained. A more useful approach is to consider emission-excitation/emission pairs at a particular wavelength (or narrow range of wavelengths) as a result of excitation in response to light having a single or narrow range of wavelengths. In practice, the fluorescence signal at a particular wavelength pair may be monitored, or a signal corresponding to a set of excitation/emission pairs may be obtained. An emission spectrum (or emission scan) is formed with a fixed light source wavelength and a fluorescence signal is acquired over a range of emission wavelengths. Similarly, the excitation spectrum may be acquired by fixing the wavelength of the detected emitted fluorescence despite the light source wavelength variation. Excitation-emission spectra can be used to represent fluorescence signals as topographs covering a range of excitation and emission wavelengths. The emission and excitation spectra correspond to the vertical profile of this spectrum. The point of value falling on the diagonal of the excitation-emission spectrum (i.e., where the excitation and emission wavelengths are equal) represents the intensity of the elastically scattered photons reflected back to the detection system by the tissue. These "reflectance" measurements can be obtained by simultaneous scanning of both the excitation and emission monochromators in the fluorometer or by separate dedicated devices. Both fluorescence and reflectance measurements can be used to determine the true or "intrinsic" fluorescence properties of an optically turbid medium, such as biological tissue.

When the excitation light is emitted onto the tissue, it undergoes scattering and absorption processes, which vary with the optical properties of the sought site, the excitation wavelength and the optical probe geometry. Since the emitted fluorescence propagates through tissue before exiting and collecting, it also undergoes wavelength-and position-dependent absorption and scattering. Typically, the tissue property of interest is its "intrinsic" fluorescence, which is defined as the fluorescence emitted by a sample that is homogeneous, scatter-free, and optically thin. To accurately distinguish the intrinsic fluorescence spectrum of the tissue of interest, the spectral altering effects of scattering and absorption imposed on the excitation and emission light can be removed. Changes caused by subject-to-subject differences and site-to-site differences outweigh subtle spectral changes indicative of tissue state. The intrinsic fluorescence spectrum of the molecule of interest can be found by spectral correction based on each subject's tissue optics (at the same site as the fluorescence measurement, or at a different site with a predictable relationship to the site). This intrinsic correction can mitigate variations across and within the subject, finding spectral features that are relevant to the presence of disease and disease state.

The data described in this example were collected using a SkinSkan fluorometer (sold by Jobin-Yvon, Edison, NJ, USA). The excitation side and emission side of the skinnscan system have a dual scan 1/8-m grating monochromator, achieving a system bandpass of-5 nn. The excitation light was supplied by a 100W Xe-arc lamp and was matched to the f/number of a double fiber probe containing 31 source fibers and 31 detection fibers. The fibers had a core diameter of 200 microns and could be randomly arranged in a ferrule in a 6-mm diameter annular bundle with the distal end serving as the skin interface. The output ends of the detection fibers are stacked within an input ferrule and the fiber width forms an entrance slit to a first input monochromator. Optical detection can be achieved with a photomultiplier tube whose gain can be controlled by software. Whenever non-invasive spectroscopy is performed, background measurements of a uniform reflective material (2% Spectralon, Uabsphere, North Sutton, NH, USA) are required to facilitate shifting of instrument lineshapes. In addition, the skinnscan system provides a silicon photodetector that can independently monitor the excitation lamp for correcting for fluctuations in lamp brightness. Thus, the "measured" skin fluorescence value (F) is reported as follows_meas)：

Equation 1

Wherein λ_xFor excitation wavelength, λ_mTo emission wavelength, F_tissFor "untreated" fluorescence on the detector, I_DCIs PMT dark current, L is excitation lamp brightness, t represents time, back represents Spectralon background, and R_backIs a reflection of the Spectralon background. Similarly, the measured skin reflectance value (R) may be reported as follows_meas)：

Equation 2

Wherein R is_tissIs the "unprocessed" tissue reflection signal on the detector. When the skinnscan system is used for both fluorescence and reflectance measurements, a different PMT bias voltage needs to be used for each measurement format to avoid detector saturation.

Typical measured skin fluorescence spectra are shown in the left hand graphs of fig. 1 and 2. These figures show spectra obtained under different collection modalities at two different wavelength ranges. FIG. 1 shows excitation spectra in which the excitation wavelength is scanned in the range 315 to 385nm, while the emitted fluorescence is measured at a fixed wavelength of 400 nm. FIG. 2 presents emission scan data with excitation fixed at 325nm and passing at 340 to 500The detection subsystem was scanned in the nm range to monitor fluorescence. All spectra were obtained from the arms of 17 diabetic and 17 non-diabetic subjects between the ages of 40 and 60. The middle illustration of these figures depicts the measured reflectance spectrum. Each reflectance spectrum corresponds to a specific fluorescence spectrum and is acquired at the same site of the same subject. Fluorescence and reflectance spectra demonstrate typical changes due to poor repositioning of the probe, environmental changes, and subject-to-subject physiological differences. These changes can exceed the spectral changes caused by the disease state and prevent the measured spectrum from being used for diagnosis. To accurately discriminate or quantify disease status, additional tissue is applied. An approximation (F) for evaluating the intrinsic fluorescence spectrum_corr) Including dividing the measured fluorescence spectrum by the root mean square of the measured reflectance at the excitation and/or emission wavelengths (see, e.g., Finlay et al, Photochem Photobiol, 2001 and Wu et al, Appl Opt, 1993):

equation 3

The optimum values of n and k are determined based on the arrangement of the source and detector fibers and may be determined empirically. Intrinsic fluorescence spectra obtained from the spectra of fig. 1 to 2 using the calibrated function of equation 3 and with values of k =0.5 and n =0.7 are shown in the right hand graphs of these figures. Note that the intrinsic correction has excluded a large proportion of the patient's internal variations and allows visual identification of a rough set of spectra corresponding to the disease state.

The values of n and k used in the intrinsic corrections set forth in fig. 1 and 2 were selected to minimize spectral variations associated with repeated insertion of the measurement instrument into the forearm of the study participant. If multiple spectra are collected by each participant in a patient exam, the spectra of the ith spectrum of the subject may be interpolated (S)_insert) Expressed as the absolute deviation from the median spectrum of the subjects:

S_{insert i,j}(λ，n，k)=abs[F_{corr i,j}(λ，n，k)-median(F_corr.,j(λ，n，k))]/median(F_corr.,j(λ, n, k)). equation 4

The integrated insertion error is S_insertVariance of (a):

V_insert(λ，n，k)=var(S_insert(λ, n, k)). Equation 5

Fig. 3 depicts the (dashed line, k =0.5, n =0.7) interpolated variance of the measured (solid line, "uncorrected") spectrum and the intrinsically-corrected spectrum in the spectra of fig. 1 and 2. It follows that the inherent correction process can reduce the insertion variance by approximately a factor of 4 over the entire wavelength range. This procedure mitigates the destructive effects of changes in the optical properties of a portion of the tissue, assuming that the intrinsic tissue fluorescence is not altered by insertion.

A variety of other procedures can achieve intrinsic fluorescence correction. For example, various methods have been described by which measured fluorescence can be corrected using knowledge of measured reflectance, tissue optical properties, and probe-dependent parameters. See, for example, Gardner et al, ApplOpt, 1996, Zhang et al, Opt Lett, 2000; muller et al, Appl Opt, 2001. In addition, intrinsic fluorescence correction may be performed using a procedure in which correction parameters for a given fluorescent probe may be developed by measuring one or more tissue-mimicking materials for which fluorescence, absorption and scattering properties have been sufficiently distinguished. The procedure may also be accomplished by Monte-Carlo simulations or other computer simulations of the response of optical probes to media with known optical properties. Any of these methods can be used to correct for the effects of tissue optical properties in non-invasive skin fluorescence measurements. The multi-channel optical probe described herein may enable measurement of tissue optical properties. The optical properties can be determined by solving an analytical formula given by the multi-channel fluorescence and/or reflectance measurements. Alternatively, the optical property may be estimated from the spectral measurements by comparison with a look-up table that correlates the measured values with predetermined values of the optical property. The look-up table may be formed from a numerical model that simulates multi-channel intensity measurements over a range of simulated optical properties. The look-up table may also be constructed from experimental measurements of tissue-like modeling materials across a range of optical properties. The measured or estimated optical properties can then be applied to correct for their spectral distortions induced to incident light and fluorescence. Calibration can be achieved by comparison with numerically or empirically derived probe calibration tables. The inverse of fluorescence spectroscopy can also be used to determine the intrinsic skin fluorescence after the measured or estimated tissue optical properties have been determined. Alternative methods for tissue fluorescence multi-channel optical correction include soft model techniques, such as described above (equation 3). The effects of epidermal pigmentation and surface blood content can be mitigated using multi-channel measurements. For example, by calculating the ratio of the reflection measures in adjacent channels (equation 6), the epidermal filtering effect can be substantially excluded, resulting in a ratio of the transfer functions of the two channels and thus of the tissue layers it seeks.

Equation 6

Applying the technique according to equation 6 to the fluorescence signals of the individual channels can produce a fluorescence transfer function that can provide useful fluorescence information in which epidermal and dermal upper layer shadowing has been largely eliminated. The spectral data from the various channels may be combined and/or combined in order to provide additional spectral information for multivariate techniques that can yield more accurate and/or powerful quantitative and classification models.

Although the examples described herein generally do not take polarization into account when referring to steady state fluorescence measurements, it is possible to apply these methods to other fluorescence measurement modalities. For example, frequency domain fluorescence spectroscopy may be applied, where the excitation light is amplitude modulated at an RF frequency and the phase and modulation of the emitted light is monitored. Another suitable method includes a time-resolved technique in which a short pulse of excitation light is applied to the tissue, after which the time evolution of the resulting fluorescent emission is spot-examined. Both frequency domain and time resolved measurements increase monitor performance, e.g., fluorescence lifetime (a parameter that can provide additional discrimination). Furthermore, it is possible to measure the fluorescence anisotropy using polarized excitation light and polarization sensitive detection, by r = (I)_‖-I_⊥)/(I_‖+2I_⊥) Definition of wherein I_‖And I_⊥Respectively, the intensity of the fluorescence light with a polarization parallel and perpendicular to the polarization of the linearly polarized excitation beam. Fluorescence anisotropy measurements can separate signals from fluorophores with overlapping spectra but with different rotational correlation times or molecular orientations. Furthermore, any of these techniques may be used in conjunction with imaging methods (such as microscopic microscopy or macroscopic scanning of an excitation beam) to acquire the relevant fluorophore nullsAnd (4) information of the distribution. Any of the above methods can be used in conjunction with a depth-discernable measurement technique, such as a confocal detection system or optical coherence tomography, to augment the information about the depth of distribution of the fluorophore beneath the tissue surface.

Determining a model for correlating fluorescence properties with disease state or chemical change

The relationship between tissue fluorescence properties and diabetic disease status at one or more wavelengths is not evident when the spectra are visually inspected. In view of this, it is necessary to construct multivariate mathematical relationships or "models" using intrinsic fluorescence spectroscopy to classify or quantify chemical changes in tissue disease states. The construction of such a model is generally performed in two phases: (I) collecting "calibration" or "training" data, and (ii) establishing a mathematical relationship between the training data and reference concentrations present in the disease state or training data.

During training data collection, it may be desirable to collect fluorescence data from multiple individuals to represent all disease states or reference values that are desired to be characterized by the model to be constructed. For example, if it is desired to construct a model that can separate diabetic from non-diabetic patients, it may be necessary to collect numerous representative spectra from both types of individuals. It is important to collect these data in a manner that minimizes the correlation between the disease state and other parameters that can lead to changes in fluorescence. For example, the natural formation of collagen AGEs under healthy conditions can lead to a correlation between skin AGE content and chronological AGE. It is therefore important to obtain spectra from both diabetic and non-diabetic patients across the age for which an applicable classification model is desired. Alternatively, if it is desired to construct a model that can quantify specific skin collagen AGE levels, it may be advisable to collect spectral data daily across a wide range of AGE reference values rather than measuring all individuals with the smallest AGE concentration early in the study and all individuals with larger AGE concentrations later in the study. In the latter case, a false correlation between AGE concentration and time may result, and if there is an instrument trend during the study, the resulting model may be corrected for device status rather than for analyte concentration.

While training data is being collected, additional reference information can be collected for subsequent construction of a suitable classification model. For example, if the classification model is used to predict a diabetes state, the diabetes state of some or all of the individuals represented in the training set may be collected and correlated with corresponding spectral training data. Alternatively, the classification model may predict the level of certain chemical substances in the skin, such as glycated collagen, glycated elastin, specific AGEs (such as pentosan or CML), or other proteins modified by the hyperglycemic conditions associated with diabetes. In these cases, skin biopsy samples may be collected from multiple individuals during training data collection. Furthermore, if other ancillary information (such as age, body mass index, blood pressure, HbAlc, etc.) is used in making the later disease state assessment, such information may be collected for some or all of the spectra in the training set.

After the training data is collected, a multivariate model can be constructed to correlate the disease state associated with the training data with the corresponding spectral information. An accurate model may be selected based on the final goal of the training session. There are at least two types of multivariate models that one can construct. In the first category, the goal of the training process is to form a model that can correctly classify the measured tissue disease state. In this case, the output of the model is an assignment of one or more discrete classes or groups. These categories or groups can represent different grades or manifestations of a particular disease. It can also represent various degrees of risk of contracting a particular disease or other subpopulation of the population associated with the disease state. For the second type of model, the goal is to provide a quantitative estimate of certain diabetes-induced chemical changes in the system. The output of such a model varies continuously over the relevant range of variation and does not necessarily indicate a disease state.

Classification of tissue disease states

The model construction steps that are typically performed when the ultimate goal is to assess the disease state of a tissue using a model are graphically illustrated in fig. 4. The first step is spectral pre-processing, involving pre-processing of the spectral data (if necessary), including, for example, the background correction and intrinsic fluorescence correction steps described above. In the second step, the dimensionality of the data set can be reduced by employing a factor analysis method. Factor analysis allows an individual spectrum to be described by its score over a set of factors rather than by its spectral intensity at each collected wavelength. A variety of techniques may be used in this step: principal Component Analysis (PCA) is a suitable method. For example, factors generated by Partial Least Squares (PLS) regression on reference variables related to disease state may also be used. Having generated various factors, those factors that are most useful for classification can be selected. The valued factors typically exhibit a larger separation between classes with smaller intra-class variance. Factors may be selected based on the separability index; one method that can be used to calculate the separability index for factor f is: :

equation 6

Wherein x_1，fIs the average score, x, of Category 1_2，fIs the average score for Category 2, and s²Representing the variance within a class.

Finally, a technique for classifying data into various categories may be selected. Various algorithms are suitable, and the optimal algorithm can be selected according to the structure of the training data. In linear classification analysis (LDA), a single linear function is constructed that can optimally separate the multi-dimensional spectral data into reference classes observed during training. In the second order discrimination analysis, a second order discrimination function is constructed. Fig. 5 illustrates the way in which the discrimination function can find the best separation between the two groups-which depends on the data structure. In some cases (fig. 5(a)), a linear discrimination function is sufficient to discriminate between the categories. However, as each class of multi-dimensional structures becomes more complex, more complex classification functions, such as quadratic functions, are required (fig. 5 (b)). In some cases (fig. 5(c)), the data structure makes it difficult to use even a second order discriminative analysis while other classification methods are more appropriate. There are a number of suitable classification algorithms. For example, k-nearest neighbor, logistic regression, hierarchical clustering algorithms such as classification and regression trees (CART), and machine learning techniques such as neural networks are all suitable and useful techniques. A detailed discussion of the technology can be found in the following documents: huberty, Applied Discriminiant Anaylsis, Wiley & Sons, 1994 and Duda, Hart and Stork, Pattern Classification, Wiley & Sons, 2001.

Quantification of diabetes-induced chemical changes

If the ultimate goal is to quantify the concentration of an analyte or class of analytes contained in a tissue, a different approach is employed in the model construction process. In this case, a set of (typically continuous) reference values for the one or more analytes may be obtained for some or all of the spectra in the training set. For example, where the model is for quantifying the level of pentostatin in skin collagen, the reference concentrations associated with each spectrum in the training set may be obtained from a pentostatin analysis performed on a skin biopsy sample (obtained during calibration). In cases where the biopsy is too invasive for the study participants, some alternative to AGE-related chemical changes may also be used. For example, in the case where FPG values are assumed to increase with increasing degree of diabetes progression, a modest compromise is to collect FPG data as a surrogate for skin AGE concentration. HbA1c and OGTT information can be used similarly.

A calibration model for predicting quantitative values associated with a test set may be constructed by forming a mathematical relationship between reference values and associated spectral data. There are a variety of algorithms available. For example, in Principal Component Regression (PCR), calibration data is first decomposed into sets of orthogonal scores and loading values, and then reference values are regressed onto the scores of the original N PCA factors. Another suitable method is Partial Least Squares (PLS) regression, in which a set of factors is constructed to maximize the squared covariance between the reference value and the score on each successive PLS load vector. These programs and others are outlined by Martens and Naes in Multivariate Calibration, Wiley & Sons (1989).

Of course the quantitative calibration model is not limited to the regression technique described above. Those skilled in the art will recognize that a variety of other methods may be used, including other regression techniques, neural networks, and other non-linear techniques.

Determination of disease state or chemical change from fluorescence properties

After the model is constructed, fluorescence measurements can be made on new samples with unknown disease states or diabetes-related chemical changes. The method by which a new sample disease state or chemistry can be determined will depend on the type of model constructed during the training period.

Classification of tissue disease states

As described above, various models can be used to discriminate various diabetes states from the measured fluorescence properties. For example, when using quadratic discriminant analysis, the new fluorescence spectra are projected onto the factors formed during the classification model construction using the training data to form a new score vector, x, for the test spectra_i. Calculating for each class j the average x of the training set scores covering the previously selected factor_jSum covariance matrix S_j. For example, for two categories (i.e., diabetic versus non-diabetic) of problems j ═ 1, 2. The Mahalanobis distance (Mahalanobis distance), D, from sample i to class j is then calculated for each score vector by_i，j。

D_i，j＝(x_i-x_j)^TS_j ^-1(x_i-x_j) Equation 7

The posterior probability p that a test sample i is a member of a class j (i ∈ j) can be calculated using equation 8. This number, like all probabilities, ranges from 0 to 1; a probability close to 1 indicates that the observation is close to the diabetic category, while a probability close to 0 indicates that the observation is close to the non-diabetic category. The probability that sample i is a member of a class j is given by:

equation 8

Wherein pi_ijA priori probabilities based on other knowledge (risk factors, etc.) for testing that sample i is a member of a class j. The prior probabilities are parameters that can be adjusted during a prediction period that depends in part on the diagnostic application of the classification algorithm.

Finally, thresholds for assigning new fluorescence measurements to specific tissue disease states can be used. For example, it can be determined that all fluorescence measurements that yield a posterior probability of diabetes greater than 0.75 are assigned to the diabetes class. As with the prior probabilities, the exact threshold used in validation will depend on a number of factors, including the application, the prevalence, and the socio-economic results of the positive and negative test results.

Quantification of diabetes-induced chemical changes

The output of the quantitative calibration model may be a regression vector that converts the corrected fluorescence spectrum to a quantitative analyte prediction by inner product:

equation 9

WhereinIs the analyte prediction and b is the regression vector.

The method used to generate the quantitative output may vary with the model constructed over the training period. Final analyte quantification using, for example, neural networks can be performed by different methods but will also produce similar outputs.

After various types of multivariate models (i.e., quantitative models for chemical changes or classification models for tissue disease states) are constructed, the accuracy of the models can be tested by predicting the disease state associated with a well-characterized "confirmed" spectrum. There are also a number of techniques for accomplishing this task. In leave-one-out cross-validation, individual spectra or groups of spectra from the training set are removed from the model construction process, and the resulting model is then used to predict disease states associated with the removed spectra from the model. By repeating this process a sufficient number of times, a mathematical assessment of the performance of the model can be made under new conditions. A more rigorous test of a newly constructed model is to apply the model to an entirely new data set or "test" set. In this case, the disease state associated with each spectrum is known, but the "test" spectrum is collected at a different time than the training data collection (e.g., after model construction). By comparing predictions of "test" data with reference values associated with these data, the diagnostic accuracy of the model can be assessed independent of training data.

Example embodiments

Fig. 6-10 show the results of a large calibration study conducted over a 3 month period. In these experiments, noninvasive fluorescence and reflectance spectra were obtained from the skin of the study participants' arms using a commercially available fluorometer (SkinSkan, Jobin-Yvon, Edison, NJ, USA). During the training period, 57 type II diabetic subjects and 148 non-diabetic subjects were measured by fluorescence spectroscopy. Study participants were selected based on their age and their self-reported diabetic status. In addition to the subject's own reported disease status, FPG and OGTT reference information was also collected for all diabetic patients and a subset of non-diabetic patients in the study. For these individuals, FPG and 2 hour OGTT values were collected on two different days, respectively. Spectral measurements were collected on day 3 and were not prepared prior to imposing specific fasting requirements and other tests on the study participants.

In this study, several fluorescence data sets were obtained. 3 different sets of emission scans were collected at 2.5-nm data intervals: (1) 325nm, 340 to 500nm, (2) λ_x＝370nm，λ_m385 to 500nm, and (3) λ_x＝460nm，λ_m475 to 550 nm. In addition, 3 different sets of excitation scans (data intervals of 2.5 nm) were also collected: (1) lambda [ alpha ]_m＝460nm，λ_x325 to 445nm, (2) λ_m＝520nm，λ_x325 to 500nm, and (3) λ_m＝345nm，λ_xLow-resolution (data interval of 10-nm) excitation-Emission Spectra (EEMs) were also collected, as well as skin reflectance data across the excitation and emission wavelength ranges used in fluorescence data acquisition. These data sets and their corresponding wavelength ranges are graphically depicted in fig. 6, where the black open circles represent excitation scans, the gray filled circles represent emission scans, the gray x-symbols represent EEMs, and the black x-symbols represent reflection scans. Two replicates of each of these data sets were obtained for each study participant. Each repetition of the set of spectral data is obtained from a different physical region of the arm.

Two different multivariate models were constructed using these training data. The first model classifies the new measurements according to their apparent diabetes status. The second model uses FPG reference values as a surrogate for skin-collagen AGE content to quantify diabetes-induced chemical changes.

Classification of tissue disease states

After completion of training data collection, all non-invasive measurements are collected together with reference information (self-reported diabetes status, FPG and OGTT reference values). Post-processing was first performed on all fluorescence data, including correction of intrinsic fluorescence using the method described in equation 3 (k 0.5 and n 0.7). The results presented here were obtained by combining the 3 sets of excitation scans described above into a single large fluorescence spectrum. The data set was dimensionality reduced using PCA factor analysis and classification functions were constructed using QDA using scores for 5 of the top 25 principal components using the separability indices indicated in equation 6 to identify those PCA factors that were best suited for class discrimination. The diagnostic accuracy of the QDA classification function was assessed using the leave-one-out cross-validation method. In this example, all spectral data for a single patient is provided by training data, constructing a separate QDA model, and calculating the posterior probability of membership of each spectrum in the diabetes class. FIG. 7 is a box and whisker plot of the cross-validation posterior probability of diabetes class membership among all study participants. From which it is known that diabetic individuals generally exhibit a higher probability of diabetes than non-diabetic patients. For diagnostic tests it is often the case that no single test threshold can completely separate all diabetic patients from all non-diabetic patients using the example data. One method of summarizing the diagnostic accuracy of the QDA classification function is to plot a true positive rate (i.e., sensitivity) -false positive rate (i.e., 1-specificity) curve for a test threshold range. The area under the Receiver Operating Characteristic (ROC) curve is close to 1 for a perfect classification test and close to 0.5 for an almost random chance test. The ROC curves from the QDA cross-validation program described above are depicted as solid lines in fig. 8. The area under the ROC curve is 0.82 and at the curve inflection point, a sensitivity of about 70% is obtained when the false positive rate is about 20%. The relevant equal error rate (the point at which the sensitivity and the false positive rate are equal) is about 25%. All of these ROC parameters are advantageously compared to comparable values from the FPG ROC curve (shown as a dashed line for comparison). The ROC curve for the FPG test was calculated from a database covering 16,000 individuals participating in the Third National Health and Nutrition Survey (Third National Health and Nutrition evaluation Survey) conducted in 1988 to 1994. The curves were generated using the diabetic status of the study participants themselves as faithfully disclosed by applying various test thresholds to the FPG test values.

Quantification of diabetes-induced chemical changes

Rather than directly assigning the diabetic condition to an unknown sample using fluorescence measurements, it is more valuable to produce a quantitative degree of chemical change associated with the occurrence or progression of diabetes. For example, the concentration of pentosan, CML, or another skin collagen AGE may be analyzed by skin biopsy. The reference values may be used to construct a multivariate model as described above. However, in this example, the reference data is not available and the FPG values collected during the training period are used as a substitute for this chemical information.

A quantitative PLS calibration model was constructed from the same corrected fluorescence data described above. The results presented here were obtained by combining the 3 excitation scans described above into a single large fluorescence spectrum. A total of 3 latent variables or PLS factors were constructed from the non-invasive fluorescence data and used to model changes in the FPG reference values. Since most of the fluorescence wavelength is concentrated around the CLF window, it is speculated that the spectral changes are due (at least in part) to collagen cross-linking and associated diabetes progression. Therefore, the FPG test values are not expected to be an optimal alternative to disease progression.

The results of the cross-validation are presented in fig. 9, where all data from a single study participant was applied in round-robin fashion at each iteration. PLS estimates for the 3 model factors are plotted on the y-axis; since the fluorescence change is presumed to originate from AGE chemistry, this axis is labeled "Chemical Progression" and the dimension can be arbitrary. The corresponding FPG values are indicated on the abscissa. Values from diabetic subjects are shown as gray filled circles, while non-diabetic subjects are represented by open circles. From this figure, it can be seen that in general a larger reference value corresponds to a larger estimate of PLS for chemical evolution, but as expected does not exhibit an ideal linear relationship. Furthermore, from this figure, on average, diabetic individuals exhibited a greater estimate of chemical progression than non-diabetic individuals. Reference values (such as one or more skin-collagen AGEs) that are more closely related to true disease progression can yield a model that more closely follows a linear relationship.

Although quantitative models of diabetes-related chemical changes can only report test values (i.e., fail to represent a classification related to tissue disease state), it is also possible to use the output of such models for classification purposes. An example of this procedure is illustrated in FIG. 10, which is a ROC curve formed from the PLS chemical progression estimates described in FIG. 9 using the diabetic status faithfully reported by the study participants themselves. The FPG ROC curve from fig. 8 is reproduced in fig. 10 for comparison. The area under the ROC curve is 0.81 and a sensitivity of 65% is reached at a false positive rate of 20% at the inflection point of the curve. The relevant equal error rate (the point at which the sensitivity and the false positive rate are equal) is about 25%. All these ROC parameters are again advantageously compared to comparable values from the FPG ROC curve.

Example devices

The components or subsystems of the device for characterizing and/or quantifying disease states from tissue fluorescence are illustrated in fig. 11. The illumination subsystem includes a light source a adapted to illuminate the tissue to electronically excite endogenous chromophores within the tissue. The illumination subsystem includes an optical system B that couples light generated by the light source a to the tissue and collects resulting fluorescence from the tissue sample and couples the collected fluorescence to the detection subsystem C. In the detection subsystem, fluorescence is typically converted to an electronic signal. The signal corresponding to the tissue fluorescence is measured and characterized by an analysis or data processing and control system D. The processing/control system may also control or modify the actions of other subsystems.

Examples of such systems I include high intensity arc lamps, optical shutters, monochromators, and collimators as the core elements of the light source. The optical coupling subsystem consists of a two-pronged fiber bundle that couples excitation light to the tissue and collects fluorescence emitted by the tissue. The second lead of the dual-branched fiber bundle couples the collected fluorescence to the detection subsystem. The detection system comprises a monochromator (different from the monochromator of module a) and a detector, such as a photomultiplier tube. The electrical signal corresponding to the tissue fluorescence is digitized, processed and stored by a computer (component D). The computer may also control the functions of other subsystems, such as the tuning of the monochromator and the switching of the shutters.

In example II, the dual-leg bundle of example I was replaced with a lens and mirror system to deliver excitation light from the light source to the tissue and subsequently collect and transfer to the detection subsystem the fluorescence emitted by the tissue.

In example III, the broadband light source of example I, consisting of a high intensity arc lamp and a monochromator, is replaced by one or more discrete light sources, such as LEDs or laser diodes. LEDs may require suitable optical bandpass filters to produce excitation light of sufficiently narrow wavelengths. The LED or laser diode may be operated in a continuous wave, modulated or pulsed manner. The output of these light sources may be coupled to the tissue by an optical subsystem, such as a fiber optic bundle of example I or a collection of mirrors and/or lenses as described in example II.

In example IV, the detection system of example 1, consisting of a monochromator and a single detector, was replaced by a spectrograph and detector array or a CCD array.

An example of a skin fluorometer is presented in FIG. 12. The illumination subsystem consists of a xenon arc lamp coupled with a double monochromator. The narrow spectral output from the monochromator is coupled into a two-pronged fiber bundle. The fibers in the ferrule that contact the tissue may be randomly arranged (as shown in fig. 13) or may be designed to be constructed with a specific source-detector fiber spacing (as shown in fig. 14). An example of a clamp fitted with a fiber bundle that can be brought into contact with the skin of the subject (in this case, a forearm support) is shown in fig. 15. The support provides a means for the subject to comfortably position their arm while the underside skin of the forearm is in contact with the fiber bundle delivery/collection end. The support also facilitates positional reconstruction of the arm portion relative to the fiber optic bundle. The fluorescence collected by the detector fibers within the dual-branched fiber bundle forms an entrance slit to the fluorometer's second monochromator, as shown in FIG. 12. The monochromator may filter the incoming fluorescence and allow a narrow band of frequencies to fall on the detector, photomultiplier tube (PMT), or channel photomultiplier tube. The PMT may be replaced by a sufficiently sensitive silicon avalanche photodiode or a conventional silicon photodiode. Tunable grating pairs in both the light source and detector monochromator may allow the wavelength of each portion to be tuned separately. The signal from the PMT is digitized and recorded by a computer, which may also tune the grating, adjust the detector, and control the monochromator shutter.

Which can be used to preferentially collect information from the dermis. FIG. 14 is an illustration of a tissue interface suitable for use with the present invention. The tissue interface includes a plurality of excitation fibers operable to be in optical communication with a light source and adapted to deliver excitation light to tissue. It further includes a plurality of receiving fibers operable for optical communication with the detector and adapted to receive light emitted by the tissue in response to the excitation light. The receiving fibers are spaced apart from each other and aligned relative to the excitation fibers so that fluorescence information from the dermis layer of the skin can be preferentially collected without physical exposure of the dermis.

As previously discussed, preferential collection of information from the dermis via multiple channels may also be employed for measurement of tissue optical properties. FIG. 16 is an illustration of a tissue interface suitable for use with the present invention. The tissue interface includes a plurality of excitation fibers (e.g., as shown by the solid circles) operable to be in optical communication with a light source and adapted to transmit excitation light to tissue. It further includes a plurality of receiving fibers (e.g., as shown by both open circles and horizontally hatched circles) operable to be in optical communication with the detector and adapted to receive light emitted by the tissue in response to the excitation light. In the illustration, the hollow circle includes a first channel to receive the fiber and the shaded circle includes a second channel to receive the fiber. In each channel, the receiving fibers are spaced apart from each other and aligned relative to the excitation fibers so that fluorescence information from the dermis layer of the skin can be preferentially collected without physical exposure of the dermis. The light collected from the skin by each receiving channel is detected individually by multiple detectors or by switching between channels and a single detector.

Fig. 17 and 18 show other arrangements of excitation and reception fibers that allow for multichannel collection of information. FIG. 17 shows a circular fiber arrangement in which the central (solid circle) fiber that transmits the excitation light is surrounded by the first channel of the receiving fiber (hollow circle), which is further surrounded by the second channel of the receiving fiber (shaded circle). FIG. 18 shows a linear fiber arrangement in which a plurality of excitation fibers (filled circles) are arranged in a row. The first channels (open circles) that receive the fibers are arranged in a row parallel to (and slightly spaced from) the excitation row. The second channels (shaded circles) that receive the fibers are also arranged in a row parallel to (and further away from) the active row.

Figures 19 to 22 show various views of possible arrangements of the multi-channel fibre optic tissue probe relative to the sampling surface. FIG. 19 is a diagrammatic cross-sectional view of a portion of a vertically aligned multi-channel fiber optic tissue probe in which solid fibers may represent excitation fibers, hollow fibers represent first receive channels, and the fibers with line shading represent second receive channels. In such an arrangement, the spacing between the excitation fibers and the first and second receiving channels may be selected to suit the proven desired information that may be used for tissue optical property determination. FIG. 20 is a diagrammatic cross-sectional view of a portion of a multi-channel fiber optic tissue probe in an angled arrangement. The angle of inclination from perpendicular to the excitation fiber may be 0 to 60 degrees. Also, the inclination of the first and second receiving channels (hollow and shaded fibers, respectively) may be inclined 0 to 80 degrees in opposite directions of the excitation fiber, and need not necessarily be inclined by equal or opposite amounts. FIG. 21 is a diagrammatic cross-sectional view of a portion of a multi-channel fiber optic tissue probe in an angled arrangement. Where the first and second receiving channels are placed on either side of the central excitation fiber, respectively. FIG. 22 is an isometric view showing how angled fibers can be arranged to increase light throughput.

FIG. 23 is a schematic diagram of a multi-channel fiber optic tissue probe seeking a volume of tissue at various excitation and receiver spacings. In each of the 4 illustrations, there is a single tilted excitation fiber indicated by a downward arrow pointing to the tissue volume (shown in black). Opposite the excitation fiber are 4 receive fiber channels, each separated by a distance from the excitation fiber. From left to right, the diagram shows the sought tissue region as a function of excitation fiber and receive channel spacing. These separate receiving channels allow for the preferential collection of information from the dermis that can be used for tissue optical property measurements.

Those skilled in the art will recognize that the present invention may be embodied in a wide variety of forms other than the specific embodiments described and contemplated herein. Accordingly, the present invention may be modified in form and detail without departing from the scope and spirit of the invention as set forth in the appended claims.

Claims (52)

1. An apparatus for determining a tissue state of an individual's tissue, the apparatus comprising:

a. an illumination system for illuminating a portion of the individual tissue with excitation light and detecting light emitted by the tissue from fluorescence emitted by chemicals within the tissue, the illumination system comprising: a plurality of excitation fibers and a plurality of receiving fibers spaced apart from each other and arranged relative to the excitation fibers such that fluorescence information from the dermis layer of the skin can be preferentially collected;

b. a detection system for converting light detected by the illumination system into an electronic signal;

c. an analysis system for determining the tissue state from the electronic signal and a model correlating fluorescence to tissue state.

2. The device of claim 1, wherein the excitation light has a wavelength in a range of 280nm to 500 nm.

3. The device of claim 2, wherein the excitation light has a wavelength in a range of 315nm to 500 nm.

4. The device of claim 1, wherein the excitation light has a first wavelength at a first time and a second wavelength different from the first wavelength at a second time.

5. The device of claim 1, wherein detecting light emitted from the tissue comprises detecting light at a wavelength greater than a wavelength of the excitation light.

6. The device of claim 1, wherein detecting light emitted from the tissue comprises detecting light at a wavelength between 250nm and 850 nm.

7. The device of claim 1, wherein detecting light emanating from the tissue comprises detecting light at each of a plurality of wavelengths.

8. The device of claim 7, wherein the excitation light has a single wavelength, and wherein detecting light emitted by the tissue comprises detecting light at a plurality of wavelengths.

9. The device of claim 1, wherein the excitation light wavelength varies over time, and wherein detecting light emitted by the tissue comprises detecting light at a first wavelength.

10. The apparatus of claim 1, wherein detecting light emitted from the tissue comprises:

a. determining tissue reflectance properties at an excitation wavelength;

b. detecting light returning from the tissue in response to illumination at the excitation wavelength such that light in contact with the dermis is preferentially detected compared to light not in contact with the dermis;

c. determining a corrected fluorescence measurement from the detected light and tissue reflectance characteristics;

d. and wherein detecting the tissue state comprises determining the tissue state from the corrected fluorescence measurements and a model that relates fluorescence to tissue state.

11. The apparatus of claim 1, wherein detecting light emitted from the tissue comprises:

a. determining tissue reflectance properties at a detection wavelength;

b. detecting light of the detection wavelength returned from the tissue in response to the illumination such that light in contact with the dermis is preferentially detected compared to light not in contact with the dermis;

c. determining a corrected fluorescence measurement from the detected light and the tissue reflectance characteristic;

d. and wherein detecting the tissue state comprises determining the tissue state from the corrected fluorescence measurements and a model that relates fluorescence to tissue state.

12. The apparatus of claim 1, wherein detecting light emitted from the tissue comprises:

a. determining a first tissue reflection characteristic at an excitation wavelength;

b. determining a second tissue reflection characteristic at the detection wavelength;

c. detecting light at the detection wavelength returned from the skin in response to illumination at the excitation wavelength such that light in contact with the dermis is preferentially detected compared to light not in contact with the dermis;

d. determining a corrected fluorescence measurement from the detected light and the first and second tissue reflectance characteristics;

e. and wherein detecting the tissue state comprises determining the tissue state from the corrected fluorescence measurements and a model that relates fluorescence to tissue state.

13. The apparatus of claim 10, wherein determining tissue reflectance characteristics comprises:

a. illuminating the tissue with reflected illumination light having an excitation wavelength;

b. detecting reflected light reflected from the tissue having the excitation wavelength using the same detector as used to detect light returning from the tissue; and

c. tissue reflection characteristics are established from a relationship between the reflected illumination light and the reflected light.

14. The apparatus of claim 10, wherein determining tissue reflectance characteristics comprises:

a. illuminating a portion of the tissue with reflected illumination light having an excitation wavelength;

b. detecting reflected light having the excitation wavelength that is reflected from the portion of the tissue; and

c. tissue reflection characteristics are established from a relationship between the reflected illumination light and the reflected light.

15. The apparatus of claim 11, wherein determining tissue reflectance characteristics comprises:

a. illuminating the tissue with reflected illumination light having the same wavelength as the detection wavelength;

b. detecting reflected light having the detection wavelength reflected from the tissue using the same detector used to detect light returning from the tissue; and

c. tissue reflection characteristics are established from a relationship between the reflected illumination light and the reflected light.

16. The apparatus of claim 11, wherein determining tissue reflectance characteristics comprises:

a. illuminating the portion of the tissue with reflected illumination light of the same wavelength as the detection wavelength;

b. detecting reflected light reflected from the tissue having the detection wavelength; and

c. tissue reflection characteristics are established from a relationship between the reflected illumination light and the reflected light.

17. The device of claim 1, wherein the detection of light comprises determining a relationship between excitation wavelength and fluorescence at a detection wavelength, and wherein determining the tissue state comprises comparing the relationship to a model defining the relationship between tissue state and fluorescence at the excitation wavelength and the detection wavelength.

18. The device of claim 17, wherein detection of light comprises determining a relationship between illumination at a plurality of excitation wavelengths and fluorescence at a detection wavelength, and wherein determining a tissue state comprises comparing the relationship to a model defining a relationship between a tissue state and fluorescence at the plurality of excitation wavelengths and the detection wavelength.

19. The device of claim 17, wherein detection of light comprises determining a relationship between illumination at an excitation wavelength and fluorescence at a plurality of detection wavelengths, and wherein determining a tissue state comprises comparing the relationship to a model defining a relationship between a tissue state and fluorescence at the excitation wavelength and the plurality of detection wavelengths.

20. The device of claim 17, wherein detection of light comprises determining a relationship between illumination at a plurality of excitation wavelengths and fluorescence at a plurality of detection wavelengths, and wherein determining a tissue state comprises comparing the relationship to a model defining a relationship between a tissue state and fluorescence at the plurality of excitation wavelengths and the plurality of detection wavelengths.

21. The apparatus of claim 1, further comprising obtaining biological information related to the individual, wherein determining the tissue state comprises determining the tissue state from the biological information, the detected light, and a model that correlates biological information and fluorescence to tissue state.

22. The apparatus of claim 21, wherein the biological information comprises a height of the individual, a weight of the individual, a family history of the individual, an ethnicity of the individual, a skin melanin level of the individual, or a combination thereof.

23. The device of claim 1, wherein the tissue comprises the skin of the individual.

24. The apparatus of claim 1, wherein the model is determined according to the following description:

a. for each of a plurality of subjects:

i. determining a fluorescence property of a portion of the subject tissue;

determining a tissue state of the subject;

b. applying a multivariate approach to the determination of the plurality of fluorescent properties and the associated tissue state determination to form a model correlating the fluorescent properties to the tissue state.

25. The device of claim 1, wherein the tissue state comprises the presence of glycated end products, the concentration of glycated end products, a change in glycated end product concentration, the presence of glycated collagen, the concentration of glycated collagen, a change in glycated collagen concentration, a disease state of the individual, or a combination thereof.

26. An apparatus for determining a tissue state of an individual, the apparatus comprising:

a. an optical system for establishing an interface between the optical system and a portion of the individual's tissue, the optical system comprising: a plurality of excitation fibers and a plurality of receiving fibers spaced apart from each other and arranged relative to the excitation fibers such that fluorescence information from the dermis layer of the skin can be preferentially collected;

b. a control and analysis system for:

i. for each of the complex pairs of excitation and detection wavelengths, the following relationship is determined: a relationship between the illumination light at the excitation wavelength and the reaction at the detection wavelength that is predominantly from the dermis of the skin;

determining tissue reflectance properties of the skin at each of the illumination wavelengths and tissue reflectance properties of the skin at each of the detection wavelengths;

determining intrinsic fluorescence of the skin by a relationship between illuminating light and detected light and the tissue reflectance properties;

detecting a tissue state of the individual from the intrinsic fluorescence detection light by using a model correlating intrinsic fluorescence and tissue state.

27. The apparatus of claim 26, further comprising obtaining biological information related to the individual, wherein determining the tissue state comprises determining the tissue state from the biological information, the detected light, and a model that correlates biological information and fluorescence to tissue state.

28. The apparatus of claim 27, wherein the biological information comprises information from a raman spectroscopy study of the subject.

29. The device of claim 27, wherein the biological information comprises a height of the individual, a weight of the individual, a family history of the individual, ethnicity, a skin melanin level, a blood HDL cholesterol level of the subject, a blood LDL cholesterol level of the subject, a blood triglyceride level of the subject, laser-doppler information from tissue of the subject, or a combination thereof.

30. An apparatus for determining a tissue state of an individual, the apparatus comprising an optical system and an analysis system for:

a. for each of a plurality of subjects:

i. determining the fluorescent property of a portion of the subject tissue consisting essentially of the dermis of the subject using an optical system comprising: a plurality of excitation fibers and a plurality of receiving fibers spaced apart from each other and arranged relative to the excitation fibers such that fluorescence information from the dermis layer of the skin can be preferentially collected;

determining a tissue state of the subject;

b. applying a multivariate approach to the determination of the plurality of fluorescent properties and the associated tissue state determination to form a model correlating the fluorescent properties to the tissue state.

31. The device of claim 30, wherein determining fluorescent properties comprises determining intrinsic fluorescence of the portion of the tissue.

32. The device of claim 30, wherein determining fluorescent properties comprises determining intrinsic fluorescence of the portion of the tissue at each of a plurality of detection wavelengths in response to excitation light having an excitation wavelength.

33. The device of claim 30, wherein determining fluorescent properties comprises determining intrinsic fluorescence of the portion of the tissue at a detection wavelength in response to excitation light at a plurality of excitation wavelengths.

34. The device of claim 30, wherein determining fluorescent properties comprises determining intrinsic fluorescence of the portion of the tissue at a plurality of detection wavelengths and a plurality of excitation wavelengths in pairs.

35. The device of claim 30, wherein determining the fluorescent properties of the portion of the subject tissue comprises:

a. illuminating the portion of the individual tissue with excitation light;

b. detecting light emitted from the tissue due to fluorescence of a chemical within the tissue.

36. The device of claim 35, wherein measuring light emitted from the tissue comprises

a. Determining tissue reflectance properties at an excitation wavelength;

b. detecting light returning from the tissue in response to illumination at the excitation wavelength;

c. determining a corrected fluorescence measurement from the detected light and the tissue reflection characteristic.

37. The device of claim 35, wherein measuring light emitted from the tissue comprises

a. Determining tissue reflectance properties at a detection wavelength;

b. detecting light at the detection wavelength returned from the tissue in response to the illumination;

c. determining a corrected fluorescence measurement from the detected light and the tissue reflectance characteristic.

38. The apparatus of claim 35, wherein detecting light emitted from the tissue comprises:

a. determining a first tissue reflection characteristic at an excitation wavelength;

b. determining a second tissue reflection characteristic at the detection wavelength;

c. detecting light of the detection wavelength returned from the skin in response to illumination at an excitation wavelength;

d. determining a corrected fluorescence measurement from the detected light and the first and second tissue reflectance characteristics.

39. The apparatus of claim 30, wherein determining the tissue state comprises at least one of:

a. evaluating the subject according to OGTT;

b. evaluating the subject according to FPG;

c. evaluating the subject according to the HbA1c test;

d. assessing the subject based on the observed symptoms of the disease state;

e. determining the presence or extent of a related complication of the disease state;

f. determining a previous disease state;

g. determining the level of glycation end-products in the tissue of the subject.

40. The apparatus of claim 30, wherein applying a multivariate method comprises using a multivariate model constructed according to partial least squares, principal component regression, principal component analysis, classical least squares, multiple linear regression, ridge regression algorithm, linear discriminant algorithm, quadratic discriminant algorithm, logistic regression algorithm, or a combination thereof.

41. The apparatus of claim 30, wherein the portion of the tissue comprises skin of the subject.

42. An apparatus for determining a tissue state of an individual's tissue, the apparatus comprising:

a. an illumination subsystem comprising a plurality of excitation fibers and a plurality of receiving fibers spaced apart from each other and arranged relative to the excitation fibers such that fluorescence information from the dermis layer of the skin can be preferentially collected;

b. a detection subsystem, wherein the illumination and detection system is configured to cause the detection system to preferentially detect light emitted from the dermis of the subject's skin;

c. an analysis subsystem comprising a model correlating fluorescence properties of the skin of the subject to tissue states.

43. The apparatus of claim 42, wherein the model is determined according to:

a. for each of a plurality of subjects:

i. determining a fluorescence property of a portion of the subject tissue;

determining a tissue state of the subject;

b. applying a multivariate approach to the determination of the plurality of fluorescent properties and the associated tissue state determination to form a model correlating the fluorescent properties to the tissue state.

44. An apparatus for determining a tissue state in an individual, the apparatus comprising:

a. an optical system for determining a fluorescent property of a portion of the skin of a subject by collecting light that is primarily in contact with the dermis of the skin, the optical system comprising: a plurality of excitation fibers and a plurality of receiving fibers spaced apart from each other and arranged relative to the excitation fibers such that fluorescence information from the dermis layer of the skin can be preferentially collected;

b. an analysis system that applies a multivariate method to determine a tissue state of the subject from the fluorescence properties.

45. The apparatus of claim 44, wherein:

a. the fluorescent property comprises intrinsic fluorescence of the portion of the skin;

b. the tissue state comprises a concentration of glycation end products;

c. the use of multivariate methods involves the application of a multivariate model that relates intrinsic skin fluorescence to the concentration of the glycation end-products.

46. The device of claim 44, wherein determining fluorescent properties comprises determining a response of the tissue to amplitude modulated excitation light, short pulse excitation light, or polarized excitation light, or a combination thereof.

47. The apparatus of claim 44, wherein determining fluorescent properties comprises discriminating tissue depth from which the fluorescent properties are obtained using confocal detection or optical coherence tomography.

48. The device of claim 44, wherein determining fluorescent properties comprises using raster scanning or imaging optics to obtain information related to the spatial distribution of fluorescent properties.

49. The apparatus of claim 44, wherein determining fluorescent properties comprises discriminating tissue depth from which the fluorescent properties are obtained using an optimized optical probe.

50. The apparatus of claim 44, wherein determining fluorescent properties comprises discriminating tissue depth using a fiber optic probe having light source and receptor fibers arranged in a spatial pattern, the fluorescent properties being obtained from the tissue depth.

51. The apparatus of claim 44, wherein determining fluorescent properties comprises using an optical probe having a plurality of source bits and/or a plurality of collection bits, the optical probe providing multiple channels of optical information to preferentially collect light emitted from the dermis.

52. The apparatus of claim 44, wherein determining fluorescent properties comprises determining fluorescent properties at each of a plurality of optical channels through the tissue.