This invention relates to measurements of material properties by determination of the response of a sample to incident radiation, and more specifically to the measurement of analytes such as glucose or alcohol in human tissue.
Noninvasive glucose monitoring has been a long-standing objective for many development groups. Several of these groups have sought to use near infrared spectroscopy as the measurement modality. To date, none of these groups has demonstrated a system that generates noninvasive glucose measurements adequate to satisfy both the U.S. Food and Drug Administration (“FDA”) and the physician community. Spectroscopic noise introduced by the tissue media is a principal reason for these failures. Tissue noise can include any source of spectroscopic variation that interferes with or hampers accuracy of the analyte measurement. Changes in the optical properties of tissue can contribute to tissue noise. The measurement system itself can also introduce tissue noise, for example changes in the system can make the properties of the tissue appear different. Tissue noise has been well recognized in the published literature, and is variously described as physiological variation, changes in scattering, changes in refractive index, changes in pathlength, changes in water displacement, temperature changes, collagen changes, and changes in the layer nature of tissue. See, e.g., Khalil, Omar: Noninvasive glucose measurement technologies: an update from 1999 to the dawn of the new millennium. Diabetes Technology & Therapeutics, Volume 6, number 5, 2004. Variations in the optical properties of tissue can limit the applicability of conventional spectroscopy to noninvasive measurement. Conventional absorption spectroscopy relies on the Beer-Lambert-Bouger relation between absorption, concentration, pathlength, and molar absorptivity. For the single wavelength, single component case:
I
λ
=I
λ,O10−ε
αλ=ελlc
Where Iλ,O and Iλ are the incident and excident flux, ελ is the molar absorptivity, c is the concentration of the species, and l is the pathlength through the medium. αλ is the absorption at wavelength λ (−log10(Iλ/Iλ,O)). These equations assume that photons either pass through the medium with pathlength l, or are absorbed by the molecular occupants.
Unfortunately, optical measurement of tissue does not match the assumptions required by Beer's law. Variations in tissue between individuals, variations in tissue between different locations or different times with the same individual, surface contaminants, interaction of the measurement system with the tissue, and many other real-world effects can prevent accurate optical measurements. There is a need for improvements in optical measurement methods and apparatuses that allow accurate measurements in real-world settings.
The present invention provides methods and apparatuses for accurate noninvasive determination of tissue properties. Some embodiments of the present invention comprise an optical sampler having an illumination subsystem, adapted to communicate light having a first polarization to a tissue surface; a collection subsystem, adapted to collect light having a second polarization communicated from the tissue after interaction with the tissue; wherein the first polarization is different from the second polarization. The difference in the polarizations can discourage collection of light specularly reflected from the tissue surface, and can encourage preferential collection of light that has interacted with a desired depth of penetration or path length distribution in the tissue. The different polarizations can, as examples, be linear polarizations with an angle between, or elliptical polarizations of different handedness.
A smoothing agent can be applied to the tissue surface to discourage polarization changes in specularly reflected light, enhancing the rejection of specularly reflected light by the polarization difference. The spectroscopic features of the smoothing agent can be determined in resulting spectroscopic information, and the presence, thickness, and proper application of the smoothing agent verified. The illumination system, collection system, or both, can exploit a plurality of polarization states, allowing multiple depths or path length distributions to be sampled, and allowing selection of specific depths or path length distributions for sampling. The rejection of specularly reflected light by polarization allows the sampler to be spaced from the tissue, reducing the problems attendant to contact samplers (e.g., tissue measurement trends due to pressure or heating). Separation of the sampler from the tissue enables a large area, e.g., 20 mm2, to be sampled. The illumination system and collection system can be disposed so as to communicate with different portions of the tissue surface, e.g., with portions that are separated by a fixed or variable distance.
The illumination system and collection system can be configured to optimize the sampling of the tissue, for example by changing the optical focus or the distance from the tissue surface in response to in interface quality detector (e.g., an autofocus system, or a spectroscopic quality feedback system). The portion of the tissue sampled can be identified with a tissue location system such as an imaging system that images a component of the vascular system, allowing measurements to be made at repeatable locations without mechanical constraints on the tissue.
Advantages and novel features will become apparent to those skilled in the art upon examination of the following description or may be learned by practice of the invention. The advantages of the invention may be realized and attained by means of the instrumentalities and combinations particularly pointed out in the appended claims.
The pathlength assumptions used for Beer's law are not well satisfied in the realities of making measurements in human tissue. In a medium such as tissue, photons are scattered and do not travel a single path but instead travel a distribution of paths. The distribution of paths results in a distribution of pathlengths (the length of a path traveled by a photon; a set of pathlengths having a particular distribution of lengths a “pathlength distribution” or “PLD”). In simple terms, this distribution will have a number of rays that traveled the typical path length, as well as rays that traveled shorter and longer paths through the sample via the random nature of scattering interactions. The properties of this path length distribution can be further characterized with statistical properties, such as the distribution's mean and standard deviation. These properties are not necessarily fixed for a measurement system as they can depend, in complex ways, on sample properties including the number of scattering particles, size and shape of the scatter particles, and wavelength. Additionally, the PLD of a specific volume of tissue is sensitive to the inherent properties of the tissue as well as the way in which the tissue is sampled. Any change in the PLD between noninvasive measurements or during a noninvasive measurement will cause a change in path such that the assumptions of Beer's law are not satisfied. The net result is an error in the noninvasive measurement. Changes in the optical properties cause changes in the observed PLD. Changes in the PLD can result in analyte measurement errors.
Simplified Physical Model. A simplified model can be useful in understanding the principles of operation of the present invention. With recognition that tissue is a very complex layered media, a simplifying physical model provides a useful construct for explanation and dissection of the problem into simpler parts. Consider the case of making a spectroscopic measurement in a layered set of sponges. The sponges resemble tissue in that sponges have a solid structure with surrounding fluid. This physical model is similar to tissue in that tissue has a solid matrix composed of cells and collagen surrounded by interstitial fluid. This physical model of a sponge and its relationship to tissue will be systematically described with increasing complexity.
Consider a sponge as a heterogeneous structure. Depending on the size of the sampling area relative to the variation in the sponge, different observations of the sponge at different locations can look quite different. Tissue is a heterogeneous medium and thus location to location differences can exist.
Consider the simplified case where two sponges have the same composition but different densities. Density defined here as the ratio of solid sponge material to either air (if dry) or water (if wet) per unit volume. These density differences will cause changes in the light propagation characteristics due to changes in scatter. These differences will then translate into differences in the PLD between sponges. The collagen to water relationship differs in tissue and causes differences in the observed PLD.
Water is able to move into and out of the sponge based upon compression. Compression changes the density of the sponge in a transient manner and thus changes the observed PLD. Tissue is a compressible medium as evidenced by the indents one can make in tissue. Thus, compression of tissue can change the water to collagen ratio and alters the observed PLD.
Skin is composed of different skin layers, similar to a stack of sponges. Each layer in a layered stack of sponges can be of different thickness, and can have different properties (e.g., different densities). The differences in the thickness and other properties of the sponge layers can modify the optical properties of the stack and can cause a change in the observed PLD. The skin thickness of people can vary, e.g., between men and women, and as a result of aging. Thus, differences in skin thickness can cause changes in the optical properties of the media and the observed PLD. See
Returning to Beer's law:
αλ=ελlc
where Iλ,O and Iλ are the incident and excident flux, ελ is the molar absorptivity, c is the concentration of the species, and l is the pathlength through the medium, αλ is the absorption at wavelength λ. The same recorded absorbance can be obtained if the product of pathlength and concentration are maintained, see
Sources and Causes of Tissue Noise. The following discussion of sources of tissue noise and their resulting influence on pathlength distribution can help understand the operation and benefits of various aspects of the present invention.
Inherent Differences Between People. Human tissue is a complex structure composed of multiple layers of varying composition and varying thickness. Structural differences between people influence how light interacts with the tissue. Specifically, these tissue differences can cause changes in the scattering and absorption characteristics of the tissue. These changes in turn cause changes in the PLD. In experiments with more than a hundred different people, the PLD has been found to differ significantly between people.
Tissue Heterogeneity Differences. Human tissue is a complex structure composed of multiple layers of composition and varying thickness. Additionally, tissue can be highly heterogeneous with site-to-site differences. For example, skin on a person's palm is quite difference from skin on the same person's forearm or face. These structural differences between varying locations can influence how light interacts with the tissue. Experimental data indicates that the PLD differs depending upon the exact location sampled. Sampling the same tissue volume, or at least tissue volumes that largely overlap, with each repeat sampling of the tissue can reduce the PLD differences. For a given amount of overlap, a very small sampling area will have very tight requirements on repositioning error while a larger sampler will have less stringent requirements. In human testing with a fiber optic sampler we have observed that repositioning errors of only a few millimeters can create significant spectral differences. These spectral differences due to site-to-site differences cause changes in the PLD and result in prediction errors. Thus, a sampling system that samples a large area with a significant amount of overlap between adjacent samples has distinct advantages.
Tissue samplers (sometimes known as optical probes) that sample using multiple path lengths can also be susceptible to PLD differences. In multi-path samplers that use a different physical separation between the illumination and collection sites to generate different paths, slightly different locations of the tissue are sampled, introducing additional tissue noise.
Tissue Compression Issues. In addition to the inherent PLD differences described above, tissue is not a static structure and the PLD can change appreciably during the measurement period. As an example, consider the imprint left in tissue when skin is placed in pressure contact with any hard object. When sampling the arm with a solid lens or surface, the tissue can become slightly compressed during the sampling period. The compression of the tissue occurs due to movement of water and the compression of the underlying collagen matrix. The water and collagen changes result in both absorption (composition) changes and changes in scatter. The influence of contact sampling on absorption and scattering coefficients is described in U.S. Pat. No. 6,534,012. The patent describes a moderately complex system for controlling the pressure exerted on the arm. Changes in the absorbance or scattering coefficients due to the sampling process results in a variable PLD during the sampling period, and a corresponding detrimental effect on measurement accuracy.
Skin Surface Issues. In addition to internal changes, the interface between the tissue and the optical interface can also change over time. Skin is a rough surface with many wrinkles and cracks. Changes in the skin surface can occur between days, during a single day, and even during a single measurement period. Between day changes can occur, for example, due to sun exposure. Within day changes can occur, for example, due to activities such as taking a shower. Measurement period changes can occur, for example, due to changes in the air spaces or tissue cracks. As cracks or spaces decrease in size, the amount of contact between the lens and the skin improves. This improved contact can change the efficiency of light transfer into and out of the tissue and also can change the effective numerical aperture of the light entering the tissue. The numerical aperture is defined as the cone angle of the light entering and exiting the tissue. A change in the numerical aperture can cause a change in the PLD, resulting in analyte measurement errors. Sampling the tissue with a contact-based sampler can also cause the skin to perspire over the sampling period. Perspiration can change the optical coupling into the tissue and influence the measurement result.
Tissue Location Relative to Sampling System Issues. Many tissue sampling systems are based upon an assumption that the tissue is in contact with an optically clear element or that the tissue is in a spatially repeatable location. The use of an optically clear element in contact with the skin was discussed above. The fact that tissue is not a rigid structure causes significant difficulty in satisfying the criteria associated with a spatially repeatable location. Most optical systems have a focal point (e.g. like a camera) and location of the tissue in a different position effectively blurs or degrades the spectral data. The location of the tissue, specifically the front surface plane of the tissue, is influenced by differences in the elasticity of tissue, skin tension, activation of muscles, and the influence of gravity. Differences in location can be a source of tissue noise that degrades measurement performance.
Tissue Surface Contamination Issues. To make a useful noninvasive analyte (e.g., glucose or alcohol) measurement, radiation must interact with a material (e.g., a bodily fluid) that appropriately represents the blood or systemic value of the analyte of interest. Radiation that simply reflects off the front surface of the tissue generally contains little or no useful information, since it has little interaction with the bodily fluid. Radiation that reflects from the front surface or from very shallow depths of penetration will be referred to as specular light. Even radiation that penetrates deeply into the tissue and contains analyte information can be influenced by contaminating substances on the surface because the light passes through the layer of contamination twice. For example syrup on the arm of a patient undergoing glucose testing can result in a measurement error.
Accuracy of spectroscopic measurements in tissue can be improved by reducing the sources of tissue noise, and/or by increasing the information content of the spectral data. Generally, any sampler system that enables the procurement of spectra with a constant or more constant PLD will positively influence measurement accuracy. Any sampler system that provides more unique spectroscopic measurement scenarios (e.g., binocular vs. monocular, or controllable path length sampling) can increase the information content of the spectral data.
The present invention comprises tissue sampling systems that reduce tissue noise, and that can increase the information content of the spectral data acquired. Various embodiments of the present invention include various combinations of the following characteristics:
No contact between the sampler and the tissue. The lack of contact can reduce the influence of tissue compression as well as physiological changes at the tissue surface.
Illumination and collection optics that cover a relatively large area of tissue allowing the signal to be averaged over a large area, and thereby reducing site-to-site variations.
A means of varying the distribution of path lengths or depth of penetration through the tissue in order to exploit these differences in the data processing to arrive at a more accurate estimation of the analyte concentrations.
Easy assembly and overall low cost of implementation.
Ability to sample the same tissue location or have a significant amount of overlap between different samplings of the tissue. A high amount of overlap between sampling can reduce the spectral variation due to site-to-site differences.
System that compensates for differences in the location of the tissue surface and/or provides feedback to the user such that the tissue sampling site is located in a repeatable manner. Rejection of specular light from the measured spectrum. Since specular or short path length spectral data contain little or no useful analyte information, the rejection of specular light removes or decreases another source of noise.
As illustrated in
Light diffusely reflected from the tissue after interaction with the tissue can be collected by condenser optics 213 and communicated to a detector 212. The optical path from the tissue surface 208 to the detector 213 can also include a second polarizer 211 (sometimes referred to herein as an “analyzer”), a second quarter wave plate 210, or both. The illumination optics 221 and collection optics 222 can be disposed relative to each other and to the tissue surface 208 to discourage collection of specularly reflected light 209. As an example, the tissue can be placed at the intersection of the optical axis of the illumination optics 221 and the collection optics 222, with the tissue surface forming different angles with the two axes. In one implementation of the present invention, the optics were selected to illuminate an area of tissue approximately 10 mm in diameter, and a positioning apparatus (not shown) used to maintain the tissue surface at the desired location and orientation. Note that the spectrometer can be in either the illumination or the collection side.
The sampling system of
A matrix representation of the way a medium changes the polarization properties can be used in measuring and analyzing polarized light, e.g., the Mueller matrix, a square matrix containing 16 elements. The Stokes vector can be used to describe the state of polarization of the illuminating and collected light. See, e.g., C. Bohren and D. Huffman, Absorption and Scattering of Light by Small Particles (John Wiley & Sons, New York, 1983), pp 41-56, incorporated herein by reference. It can be derived from four independent polarization states, such as vertical linear polarization, horizontal linear polarization, +45 degree linear polarization, and left circular polarization. By illuminating the medium with each of these states and then, at each illumination state, observing the response using an analyzer set to each of these states, a set of 16 independent states can be observed (4 collection states for each of 4 illumination states), making up the elements of the Mueller matrix. Multiplying the input Stokes vector by the Mueller matrix produces the output Stokes vector. Although determining a complete Mueller matrix for individual tissue samples might be useful for characterizing differences between people, it is not necessary to do so to obtain useful information. Measurements using only a few polarizer positions can provide insight into the way one tissue sample scatters light differently than another tissue sample, allowing an improved calibration model to be constructed that takes advantage of this knowledge.
The example embodiment represents a major advancement in tissue sampling: a sampler that samples a relatively large area, without requiring contact with the tissue, with strong specular rejection capabilities, and the ability to generate multi-path data by changing the state of polarization between the illumination and collection optics.
A sampling system such as described in the example embodiment above can be modified for specific performance objectives by one or more of the additional embodiments and improvements described below.
Auto Focus. A motorized servo system along with a focus sensor, such as that used in autofocus cameras, can be used to maintain a precise distance between the tissue and the spectral measurement optical system during the measurement period. The tissue, the optical system, or both can be moved responsive to information from an autofocus sensor to cause a predetermined distance between the tissue and the optical system. Such an autofocus system can be especially applicable if the sampling site is the back of the hand or the area between the thumb and first finger. For example if a hand is placed on a flat surface, the auto focus mechanism could compensate for differences in hand thickness.
Tissue Scanning. The tissue can be scanned during a measurement to create an extremely large sampling area. The scanning process can involve scanning a tissue site by moving the tissue site relative to the sampler, or by moving the sampler relative to the tissue site, or by optically steering the light, or a combination thereof.
Location Feedback on Tissue Surface. The measurement system can inform the user if the tissue site is inserted into the correct focal plane or location. Many optical location or measurement systems exist, such as those commonly used for the determination of interior wall dimensions. Such a system can provide information of the general location of the tissue plane as well as the tilt of the tissue plane.
Use of Different Input Polarization States. Because of anisotropy in the structure of the tissue, e.g., anisotropy due to collagen strands, uniquely different path length distributions can be obtained by collecting data at different illumination polarizer angles. These changes in input polarization angle coupled with concurrent changes in collection polarization angle can provide a diversity of pathlength observations.
Use of Different Types of Polarization. Circular and linearly polarized light can behave differently. The use of different types of polarization can be used to enhance pathlength differences. Circularly polarized light can maintain a larger portion of its original polarization state with each forward scattering event. Thus, the use of different types of polarization can be used for the generation of different pathlength data.
Use of Different Collection and Illumination Angles. The angles of the illumination optics and collection optics relative to each other and relative to the tissue surface can influence the path length distribution. As described above, the illumination and collection optics are arranged to avoid the collection of direct specular reflection from the tissue surface. Depending upon the relationship between the illumination and collection optics, the system can be configured such that the collected light must undergo the required polarization changes and required changes in direction. Generally, greater required change of direction means longer pathlength in the tissue.
Separation of Illumination Area and Collection Area. The amount of specular light can be further reduced by separating the illumination and collection areas. With separated illumination and collection areas, any light collected by the system must have entered the tissue and propagated through the issue to the collection location.
Reduction of Skin Surface Artifacts. Tissue surface roughness can cause polarization changes that are unrelated to changes in polarization state due to propagation through tissue. The potential problem can be mitigated by coating the tissue surface with a fluid having no or few interfering absorbance features in the spectral region of interest. The use of such a skin smoothing fluid reduces polarization changes due to surface roughness. An oil with few absorbance features is Fluorolube, a fluorinated hydrocarbon oil. A light coating with such a smoothing agent can reduce the signal produced by surface scatter with minimal disturbance of the observed tissue spectra. The proper application of the smoothing agent (e.g., presence, thickness, material) can be determined from spectral features distinguishable as properties of the agent. For example, additives with known absorbance properties can be added to Fluorolube, and the spectroscopic system can determine the characteristics of the Fluorolube agent from observation of those properties. Additionally, the removal or minimization of hair can reduce artifacts due to tissue roughness.
The use of fluids to facilitate interfaces between tissue and contact-based samplers has been described previously, for example in U.S. Pat. Nos. 6,622,032; 5,655,530; 6,152,876; and 5,823,951; each of which is incorporated herein by reference. In connection with a noncontact sampler like those described herein, such a fluid can form an interface with the tissue that reduces scatter at the tissue surface and also allows rays of higher NA to make it into and out of the tissue, thereby improving the throughput and reducing the effects of variability in the tissue surface topology.
In connection with non-contact samplers such a fluid can also perform another important function that goes beyond its function in contact type samplers. In one example embodiment of a non-contact sampler the collection of unwanted light scattered from the surface can be reduced in two ways, both applicable even without a smoothing fluid (also called an index matching fluid). One way is by placing a collection aperture in such a position that the direct specular reflection (as by a mirror) is blocked, leaving the aperture to collect only non-specularly reflected, i.e. scattered, light. This works to reduce the amount of surface reflected light because most scattering surfaces, including human tissue, will generally reflect more light into the specular direction than into other directions. The rejection afforded by this method is, at best, only partially effective. The second way is by using crossed polarization between the illuminator and collector. The reason this works is that light undergoing a single reflection largely retains its original polarization, especially if it is reflected into the same plane as the plane of illumination. One problem with reliance on this approach when sampling skin is that when encountering a rough surface some of the light is reflected into planes not parallel with the original plane of illumination. This light will have its polarization rotated, and when encountering a polarizer crossed with respect to the original plane some of this unwanted light will be transmitted to the detector.
The use of smoothing fluid can facilitate both of these surface light rejection methods. The first is by reducing the reflectance of the fluid/tissue interface. Light reflected from the tissue surface may still have the polarization rotation issue described above but the intensity will be greatly reduced. A second way the fluid helps is that if it is applied in a thick enough layer and is viscous enough not to run off, it can form a smooth mirror-like layer above the tissue. Reflections from this surface will be of similar intensity to that of the tissue without the fluid but can be more effectively eliminated by the aperture mentioned above and by the crossed polarizers.
In some non-contact instrument embodiments the same lens is used for illumination and collection, making it impractical to reject specular light based on aperture placement. An advantage of such a concentric arrangement is that a much larger NA can be used in both illumination and collection, since the apertures do not interfere mechanically with each other. The use of a fluid with a smooth top surface allows crossed polarizers to effectively eliminate this specular component.
Sampling of the Same Tissue Volume. Due to the heterogeneous nature of tissue, it is desirous to sample the same tissue location or tissue volume. Several patent applications or patents have sought to address this problem by using an adhesive to temporarily attach various mechanical devices to the arm, such as a metal plate or EKG probes. See, e.g., U.S. Pat. No. 6,415,167, incorporated herein by reference. The arm is then placed on the sampler using these devices to position the arm into a mating receptacle. These devices are, at best, a very temporary means of helping to repeatedly relocate the arm during a short set of measurements. They cannot be used as a permanent fiducial to reduce measurement error over a long period of time.
Two or more ink spots on the arm outside the measurement region have been demonstrated in our laboratory to be useful in guiding positioning of the tissue. A TV camera looking at the arm from the sampler side can be used to visually guide placement of the arm onto the sampler, allowing the person being measured or an assistant to move the arm around until the ink spots are aligned with spots placed on the screen of the TV monitor. This scheme can be used over a long term by permanently tattooing the marks into the skin. Users have generally deemed this unacceptable. It also precludes easily changing measurement locations should a given sampling area become desirable.
Vein or capillary imaging can be used instead of ink spots or tattoos to provide lasting reference marks for positioning of the tissue. Vein or capillary imaging can use an optical illumination and image capture method to make veins or capillaries near the tissue surface visible, for example, on a TV monitor. In practice for analyte measurements, a measurement site can originally be located according to criteria dictated by an end application, such as non-invasive blood glucose measurement. A vein or capillary image can then be recorded either coincident with the measurement site or from surrounding regions. This recorded image can then be used as a template to guide relative placement of the tissue and sampling system in future measurements. It can be used as a visual aid to manually place the tissue in the correct location or it can be used in a servomechanism using image correlation to automatically place and maintain the instrument or tissue in the correct location. An automated system might be especially useful in maintaining position when there is no direct physical contact between the measurement apparatus and the tissue at the measurement location.
Methods of vein imaging have been described in the literature for other applications including biometric identification and assistance devices for blood withdrawl. Vein imaging techniques generally seek to obtain maximum contrast between veins and surrounding tissue. In one described technique, polarized light at 548 nm was used to illuminate the tissue in a small region. See, e.g., http://oemagazine.com/fromTheMagazine/nov03/vein.html, visited Jan. 15, 2006; U.S. Pat. No. 5,974,338, “Non-invasive blood analyzer,” issued Oct. 26, 1999, each of which is incorporated herein by reference. As the light penetrates the tissue it is scattered, illuminating a larger volume of the tissue. Light back scattered from shallow regions maintains some of its original polarization and thus can be attenuated by a crossed polarizer on the video camera. Light penetrating deeper loses its polarization and is detected by the camera, effectively back illuminating veins in the path. At a selected wavelength, blood has an absorption peak allowing a vein to be seen as a dark object against the brighter background of light scattered from underlying tissue. In other references polarized light from LEDs at 880 nm or at 740 nm are used to flood illuminate the tissue and again a crossed polarizer on a CCD camera helps to reject surface reflections and shallow depth scattered light. See, e.g., http://www.news-medical.net/?id=5395; http://www.luminetx.com/home.html; http://www.nae.edu/NAE/pubundcom.nsf/weblinks/CGOZ-65RKKV/$file/EMBS2004e.pdf, all visited Jan. 15, 2006. At these longer wavelengths the tissue scattering is less than at the shorter wavelength of 548 nm so the light can penetrate a larger distance, allowing deeper veins to be observed. Absorbance of blood at 880 nm is much less than at 548 nm so computer processed contrast enhancement may be needed to clarify the vein images. Other techniques involve injecting a contrast enhancing dye into the blood stream, which might not be acceptable for many analyte measurement applications.
Additional Capabilities
Removal of surface contaminants. Light scattering by tissue gradually randomizes the original polarization state of the illuminating light. Unscattered or weekly scattered light maintains its polarization state, whereas multiple-scattered light is randomly polarized and contributes equally to both copolarization and cross polarization states. Simple subtraction of the two states enables the weakly scattering component to be reduced See, e.g., Morgan, Stephen et al, Surface-reflection elimination in polarization imaging of superficial tissue, Optics Letters Vol 28, No 2, Jan. 15, 2003, incorporated herein by reference. Thus, surface contamination issues such as powered sugar for glucose measurements or liquor on the surface of the arm for noninvasive alcohol measurements can be largely eliminated by effectively processing data from different polarization states.
Processing of the Spectra for Minimization of PLD Differences. Information from multiple path lengths can be used to explicitly define or resolve the PLD. Another, simpler approach uses the different pathlength data to minimize the differences in the PLD and to create a PLD with the narrowest possible distribution. Suppose that the scattering resulted in photons taking one of two possible pathlengths, l1=1 and l2=3 (each with 50% likelihood), then the resulting measured transmission or absorbance is
This result is unfortunately not linear with respect to concentration. Suppose, however that the optical sampling mechanism can measure the l2=3 pathlength in isolation. Its reflectance is simply
In this trivial case, subtracting eq. 4 from eq. 1 gives a differential reflectance
And RΔ actually has a discrete pathlength of l1. This simple example can be extended to situations where two or more distinct path lengths are generated, as shown in
Use of Different Spectral Resolutions. Spectral data from the front surface of the tissue often contains little useful analyte information. As shown in
Adaptive Sampling. Experimental studies as well as simulation studies have shown that the parameters of the optical sampler can influence the PLD obtained. Specifically, the PLD obtained can be influenced by the configuration of the sampler. Important parameters include the numerical aperture of the input and output optics, the launch and collection angles, the separation between the input and output optics, and the polarization (linear or circular) of the input and output optics. The optical system can be adjusted real-time to generate the desired PLD. The adjustment of these parameters alone or in combination allows the system to procure a single spectrum with the most desirous PLD.
Direction of Change Measurements. In the management of diabetes, the individual with diabetes typically receives a point measurement associated with the current glucose level. This information is very useful but the value of the information can be dramatically enhanced by the concurrent display of the direction of change. It has been desired that the measurement device report the glucose concentration, the rate of change, and the direction of change. Such additional information can lead to improved glucose control and greater avoidance of both hypoglycemic and hyperglycemic conditions. Such a measurement has not been possible with current contact samplers because the tissue becomes compressed during the measurement process. Thus, the path length distribution changes and the highly precise measurement need for direction of change can not be obtained. With a non-contact sampler like that described herein, the tissue is not compressed and the sampling surface does not change due to contact with the sampler, allowing determination of the direction of change of the analyte concentration. See, e.g., U.S. patent application Ser. No. 10/753,50, “Non-Invasive Determination of Direction And Rate Of Change of an Analyte,” incorporated herein by reference.
Various additional example embodiments are described to help illustrate advantages possible with the present invention. The example embodiments are illustrative only; those skilled in the art will appreciate other arrangements and combinations of features.
Example Embodiment. The sampler discussed above changes the amount of cross polarization between the illumination and collection optics to measure light that has traveled at two or more different path length distributions. The spatial spread of the light can also be used to generate path length differences in the collected spectra. If the tissue is illuminated by a point source and the diffusely reflected light is received by a collection point, the path length distribution can change as the collection point is moved to different distances from the illumination point. The rate of falloff of the light intensity with distance from the origin will be dependent on the scattering and absorption properties of the tissue. The samplers described in the following text take advantage of this phenomenon.
In an example embodiment incorporating this feature, a variable path sampler uses light from a small source focused onto the tissue by a lens or mirror. A second lens or mirror collects light from a point on the tissue and focuses it onto a detector. Although, in principle, the same lens or mirror can be used for both illumination and collection, it can be advantageous to use separate optical components. This allows for the placement of baffles to help in eliminating collection of light scattered directly from the source-illuminated optics (i.e., without interacting with a sufficient depth of tissue). A spectrometer can be placed either in the path from the source to the tissue or in the path from the tissue to the detector. The physical separation between the illumination and collection spots on the tissue determines the shortest possible path length of light traveling through the tissue. To obtain different path length distributions, data can be collected with different physical separations between the input and output optics.
In practice the input and output need not be limited to single points.
This example sampler has numerous advantages: no mandatory contact with tissue in measurement region; surface scattered light can be rejected through baffling and the imaging properties of the optical system; and path length distribution, especially the minimum path, can be easily changed by changing the physical separation between input and output spots or lines. In some applications, it can be important to position the tissue accurately to maintain the lines in sharp focus. The area of tissue interrogated is not as large as with the sampler previously described, providing less averaging of tissue signal.
Example Embodiment.
Another variation of this example embodiment illuminates an annular ring mask and focuses an image of the ring onto the tissue. Light is then collected from a small point in the center of the ring and focused onto the detector. By changing the annular ring mask a series of different separations between source and collector can be achieved. This embodiment can be extended with an optical system that focuses multiple images of the annular ring onto the tissue and collects light from multiple centered points onto a detector.
Any of the examples embodiments can be used with or without a sample positioning window or index matching fluid in contact with the tissue. They can also be used with the spectrometer either in the path before or after the tissue.
Example Embodiment.
Example Embodiment.
In operation, light from the light source 701 is optionally filtered by the source filter 732 and collimated by the collimating lens 702. The illumination aperture plate 731 allows control of the numerical aperture and the angle of incidence of light on the issue surface as described below. Light from the illumination system 733 interacts with the polarizing beam splitter 730. Light having a first polarization is substantially transmitted by the polarizing beam splitter 730; light having a second polarization, different from the first polarization, is substantially reflected by the polarizing beam splitter 730 and optionally absorbed by the light absorber 727. Light having the first polarization, transmitted by the polarizing beam splitter 730, interacts with the aperture plate 726 and is condensed by the condensing lens 725 before reaching the tissue surface 708. The light then interacts with the tissue, and some light is returned along a path that will reach the condensing lens 725. Such light can include light that was specularly reflected (from the surface of the tissue), and light that has scattered after interaction with constituents of the tissue. Such light is collimated by the condensing lens 725 and interacts with the aperture plate 726 before reaching the polarizing beam splitter 730. The polarizing beam splitter 730 substantially transmits light having the first polarization, and substantially reflects light having the second polarization along a path toward the detection system 736. Since light that does not interact with the tissue (e.g., light that was merely reflected from the surface of the tissue) will have the first polarization, and light that has interacted with the tissue will have a variety of polarizations including the second polarization, the polarizing beam splitter 730 consequently preferentially directs light that has interacted with tissue to the detection system 736. Light reaching the detection system 736 interacts with the collection aperture plate 728 and the condensing lens 713. An optional polarizer 729 can provide further rejection of light having the first polarization. The light then interacts with the spectrometer 703 and detector 722, allowing determination of the absorption of the light by the tissue at each of several wavelengths.
The example embodiment can operate with different arrangements of optical components, and some of the optical components mentioned can be deleted. For example, the spectrometer 703 can mount with the illumination system 733, for example in the location shown for the spectral filter 732. Placing the spectrometer 703 in the illumination system 733 can incur the spectrometer transmittance losses before light reaches the tissue, reducing the light intensity at the tissue which can be beneficial if the light intensity is such that tissue damage or burning can occur. As another example, the light absorber can be omitted if there is no significant path for light that would reach the absorber in the figure to reflect or otherwise be transmitted to the detector. In an enclosed housing, however, the light absorber can help ensure that such light does not reflect from the housing and cause noise by reaching the detection system 736.
The aperture plates can also be omitted, if their desirable effects are not needed. The diameter of apertures in the plates can be used to control the numerical aperture of the system. Shaping each apertures as a “D” can allow the angle of incidence range on the tissue surface to be different from the collection angle range. An off-center or annular aperture on the collection side can allow the collected angle range to be chosen independently of the illumination angle range. The angles of incidence and reflection can also be affected by tilting the apparatus with respect to the tissue surface.
The concentric arrangement of the example embodiment can allow a larger numerical aperture, since there is no need for maintaining spatial separation between input and output light beams. This can result in significantly higher efficiency in transmitting light to the tissue surface and can allow for the possibility of increased irradiance or decreased lamp power. The concentric arrangement is also more compact than some other arrangements, which can allow for equivalent signal to noise performance in a smaller mechanical housing than some other embodiments. It does not offer as much versatility in control of the relative polarizations as some other embodiments.
Experimental Results
A series of tests were conducted with the various tissue sampling embodiments previously discussed with a goal of demonstrating and measuring their improved performance. These experiments involved both a tissue phantom model composed of scattering beads and tests on human tissue.
The tissue phantoms were sampled in a back scattering mode or via diffuse reflectance similar to the way the samplers would be used to measure human tissue. The tissue phantoms consisted of water solutions in a container with a flat transparent window. Various concentrations of several analytes, such as glucose and urea were included at concentration ranges found in human tissue. A range of concentrations of suspended polystyrene beads was also included to vary the scattering level and thereby the path length distribution of light propagating through the solution. The set used for testing was composed of 9 different scattering concentrations from 4000 mg/dl to 8000 mg/dl. See, e.g., U.S. patent application Ser. No. 10/281,576, “Optically similar reference samples,” filed Oct. 28, 2002, incorporated herein by reference. This variance in scatter results in a path length variation of approximately +25%. Spectral response data were then collected using a sampler like that described in connection with
Human testing was also conducted with the same optical system. The arm was inserted by placing the elbow on an elbow cup and the subject's hand gripping or placed against a vertical post. The palm of the patient was perpendicular to the ground. No window or other locating device was used to control the subject's arm position.
Large Area Sampled. As shown in
Similar Information Content of Spectra. Spectral data were taken with both a conventional fiber optic sampler such as that shown in
Improved Stability during Tissue Measurement. In previous samplers, contact with the tissue compresses the tissue, and the interface between the tissue and the sampler changes over the sampling period. Data from the same subjects were obtained from a conventional sampler and from the previously described non-contact sampler of
Demonstration that Changing Polarization Changes Pathlength in Tissue Phantoms. The length of the path over which a photon becomes depolarized depends on its initial state of polarization (linear or circular), the number of scattering events it experiences, and the scattering anisotropy of the particles it interacts with. The degree of polarization of linearly polarized light is dependent on the azimuthal angle, but circular is independent of it. The experimental system was based upon linearly polarized light, and was used to demonstrate that path length could be influenced by changing the amount of cross polarization between the illumination and collection optics.
Demonstration that Changing Polarization Changes Pathlength in Tissue. The methodology used to demonstrate pathlength variation as a function of polarization angle was repeated in human subjects. Spectral data was acquired from 5 different subjects at 0°, 22.5°, 45°, and 90°. The data were averaged together by polarization angle and the change in pathlength quantified by calculating the area under the water absorbance peak at 6900 cm−1 following baseline correction. The resulting spectral data, presented in
Demonstration of the Ability to Quantify Path Length Differences in Scattering Solutions. With a conventional ‘monocular’ sampling system, the ability to determine the scattering characteristics of a given sample is very limited. Insertion error and changes in instrument performance can make this process even more difficult. A multi-path system such as that enabled by the present invention allows the determination of relative path length. A set of variable scattering tissue phantoms were created using 9 different scattering concentrations from 4000 mg/dl to 8000 mg/dl. This variance in scatter results in a path length variation of approximately ±25%. The 9 scattering levels were sampled at four polarizer settings: 0°, 50°, 63°, 90°. The data was processed in the following manner. (1) Determine the path for each sample at each polarization angle. (2) Using all of the acquired data determine the average path as a function of polarization angle across all scattering samples. (3) Plot the determined pathlength for each solution at each different polarization angle versus the average for the solution set, as shown in
Demonstration of Path Length Variance in People. The method described above was used to examine the pathlength variation between human subjects. The process entailed determination of the average path as a function of angle across multiple subjects, and plotting pathlength at different polarization angles per subject versus the average path for multiple subjects. The slope difference defines the percentage (%) difference between people. As can be seen in
Adaptive Sampling Demonstrated. For the procurement of tissue spectra that generates the most accurate glucose measurements, the optical system may change such that the desired spectral characteristic is obtained. For example, spectral data with the same or as similar as possible path length may be desirable in some applications. One method of minimizing path variation comprises defining a desired path length and then combining data from two or more different path lengths or polarizations. The method of combination is defined by the following equation:
NewSpectra=x %*spectra 63+(1−x %)*spectra90
x=Min(water peak(Averagespectra6900)−water peak(newspectra6900))
Samples from 20 different subjects at 630 and 900 cross polarizations were combined as defined by the above equation. The comparison metric was the variance under the 6900 cm−1 band. The results plotted in
Demonstration of Surface Smoothing. When using polarization as a method for specular rejection, it can be desirable to have any changes in polarization occur due to within-tissue scattering events. Scattering events on the surface that change the degree of polarization can degrade the quality of the spectral data by increasing the variance in the PLD. To demonstrate the value of skin smoothing, surface oil was applied to the tissue in a non-specific manner. The oil applied was Fluorolube, a fluorinated hydrocarbon oil. This particular oil was selected as it has almost no absorbance in the region of interest. Spectral data was taken on multiple days with and without the skin smoothing oil. Examination of variance in 6900 water band at each polarization angle shows dramatic improvements; see
Other Applications. An individual can be identified by their spectral differences. See, e.g., U.S. Pat. Nos. 6,816,605; 6,628,809; 6,560,352; each of which is incorporated by reference herein. Samplers according to the present invention can provide an improved biometric capability. Specifically the re-location capability and the additional information provided by multi-path sampling can improve the biometric results. Using the information available via PLD differences (either a system that changes source to detector separation or that changes polarization), one can create a biometrics identification system that can have superior performance to a system that contains information at only one PLD or depth of penetration. This information can be used like different tumblers on a combination lock: for access one must satisfy the biometrics determination at multiple layers.
The particular sizes and equipment discussed above are cited merely to illustrate particular embodiments of the invention. It is contemplated that the use of the invention may involve components having different sizes and characteristics. It is intended that the scope of the invention be defined by the claims appended hereto.
This application claims the benefit of U.S. provisional application “The Influence of Changing Pathlength Distributions in the Measurement of Analytes Noninvasively and Methods for Mitigation and Correction,” No. 60/651,679, filed Feb. 9, 2005, incorporated herein by reference. This application is a continuation in part of U.S. application Ser. No. 11/350,916, filed Feb. 9, 2006, “Methods and Apparatuses for Noninvasive Determinations of Analytes,” incorporated herein by reference, which application claimed priority to U.S. provisional application 60/651,679, filed Feb. 9, 2005.
Number | Date | Country | |
---|---|---|---|
60651679 | Feb 2005 | US |
Number | Date | Country | |
---|---|---|---|
Parent | 11350916 | Feb 2006 | US |
Child | 12239601 | US |