WO2009004541A1

WO2009004541A1 - Spectroscopy measurements of the concentration of a substance in a scattering tissue

Info

Publication number: WO2009004541A1
Application number: PCT/IB2008/052570
Authority: WO
Inventors: Gerhardus W. Lucassen; Golo Von Basum; Markus Laubscher; Natalia Uzunbajakava; Miguel A. Palacios; Vouter H. J. Rensen; Peter De Peinder
Original assignee: Koninklijke Philips Electronics N.V.
Priority date: 2007-07-03
Filing date: 2008-06-26
Publication date: 2009-01-08
Also published as: CN101688832A

Abstract

A method and device (1) for calculating the concentration of glucose in skin (12) are described, wherein the skin (12) is illuminated by polarized NIR light (5) provided by a NIR source (2) and a polarizer (3) and a absorbance spectrum of the skin is detected by passing backscattered light (6) through an analyzer (7) to a spectrometer (8). The device may further comprises patches (9,10) for stretching the skin in a stretching direction (11). By measuring a first absorbance spectrum with the skin being in a first, unstreched state and a second absorbance spectrum with the skin being in a second, stretched state the contribution of collagen to the absorbance spectrum may be determined and a corresponding correction may be applied in the calculation of the glucose concentration.

Description

Spectroscopy measurements of the concentration of a substance in a scattering tissue

Field of the invention

The invention relates to spectroscopy measurements of the concentration of a substance in a scattering tissue.

Background of the invention

Spectroscopy may be used to measure the concentration of a substance in a scattering tissue. A light beam is sent on the tissue and the light that has interacted with the medium (either backscattered or transmitted) is detected, so as to deduce therefrom an absorbance spectrum. The concentration of the target substance can be deduced from the absorbance spectrum by means of a mathematical model, making use of known spectral characteristics of the substances contained within the tissue.

The scattering tissue may be the skin of a person. Spectroscopy on skin permits to estimate the person's blood concentration of a substance, in vivo and non- invasively. A type of spectroscopy that may be used is Near Infra-Red (NIR) spectroscopy, in which near infrared (NIR) light or infrared light is irradiated on skin. NIR light has a wavelength approximately comprised between lOOOnm and 2500nm; it is used because it penetrates more easily in the skin and is not immediately absorbed nor heavily scattered. NIR spectroscopy is for instance used for determining the glucose concentration in blood. US 6,990,364 describes a method for non-invasive determination of blood analytes, such as glucose, through NIR spectroscopy.

In order to extract the information from the acquired optical spectrum, it is necessary to separate the signals that are associated with glucose from those of the many other substances present in skin. In the NIR spectrum of skin, in the 1000- 2500nm wavelength region, several substances have a spectral contribution. Fig.l shows an exemplary absorbance spectrum of skin; it is known that the principle substances contributing to the spectrum are: - water, with bands at around 1450nm and 1920nm,

- collagen, with a band at around 2200nm,

- fat, with bands at around 1200nm, 1700nm and 1800nm,

- glucose, with bands at around 1550nm, 1700nm and 2100nm. As can be seen, the glucose peaks cannot be detected easily on the spectrum, since they are all masked by peaks corresponding to the absorption bands of other substances; collagen is one of the masking substances. Collagen is a fibrous protein which gives to the skin its tone and elasticity. It is one of the important scattering substances in skin and its main band is close to a glucose band.

Summary of the invention

It is therefore an object of the present invention to provide a method for measuring, by spectroscopy, within a scattering tissue comprising at least a first substance and a second substance masking the spectral contribution of the first substance, the concentration of the first substance, which provides an improved sensitivity to the first substance with a removal or at least a reduction of the second substance's contribution in the measurements.

In accordance with the present invention, there is provided a method for calculating the concentration of a first substance in a scattering tissue. The method includes an initial step of determining a first and second absorbance spectrum of the scattering tissue in a first and second stretching state. The first and second stretched spectra permit to obtain the contribution to the spectrum of the scattering tissue of the second substance. Next, a mathematical model is applied to another the concentration of the first substance by applying a mathematical model to another absorbance spectrum of the scattering tissue obtained by spectroscopy after correction of the contribution of the second substance.

With the invention, the concentration of the first substance can be calculated more accurately, since the contribution of the second substance is corrected for the calculation of the concentration of the first substance. The invention is based on the following observation: the anisotropy of scattering of the second substance is affected by stretch; a comparison between the un-stretched and stretched spectra is therefore linked to the second substance's contribution. When reference is made to "the" scattering tissue in the method of the invention described above, it should be understood that it concerns the same type of tissue, but not necessarily the same tissue sample. Indeed, the measurements on stretched and un-stretched tissue samples, so as to deduce the contribution of the second substance in the tissue, could be done on tissue samples different from the one in which the first substance concentration is calculated, if the approximation can be made that the contributions can be considered similar when performing measurements on another skin sample. They could also be done on a plurality of samples and averaged. A tissue may represent any suitable part of the body of a person or of an animal, for instance skin, hair, a tendon, etc.

According to an embodiment, the first stretching state is an un-stretched state and the second stretching state is a stretched state.

According to an embodiment, steps a) and b) are performed on a scattering tissue sample that is the same as the one on which spectroscopy measurements are performed in order to determine the first substance's concentration.

According to an embodiment, steps a) and b) are performed on a plurality of samples and the measurements are averaged.

According to an embodiment, measuring an absorbance spectrum of the scattering tissue, in a first and a second stretching states, is performed by: - irradiating the scattering tissue with light linearly polarized in a polarization direction and

- detecting light that has interacted with the scattering tissue in a detection direction that is perpendicular to the polarization direction.

According to an embodiment, the scattering tissue being skin and the second substance being collagen, the irradiating light is polarized substantially in the direction of the collagen fibers and/or stretching is performed substantially in the direction of the collagen fibers.

According to an embodiment, the direction of the collagen fibers is determined according to one of the following: - anisotropy factors of skin are determined;

- measurements are performed with irradiating light polarized in two directions perpendicular with respect to each other and the measurements are summed, averaged or the most important measurement is taken into account, in order to obtain the un-stretched and stretched spectra.

According to an embodiment, the stretched and un-stretched spectra are compared to simulated spectra in order to determine the contribution of the second substance.

According to an embodiment, the scattering tissue is skin, the first substance is glucose and the second substance is collagen.

According to an embodiment, spectroscopy is near infrared spectroscopy.

According to the invention there is also provided a device for calculating, for a scattering tissue (12) comprising at least a first substance and a second substance that presents scattering anisotropy under stretch, the concentration of the first substance. The device comprises a spectrometer and a processing arrangement. The spectrometer measures a first absorbance spectrum of the scattering tissue (12), in a first stretching state and a second absorbance spectrum of the scattering tissue (12) in a second stretching state. The processing arrangement then determines from the spectra the contribution of the second substance in an absorbance spectrum of the scattering tissue obtained by the spectrometer. The processing arrangement further computes the concentration of the first substance by applying a mathematical model to the absorbance spectrum after correction of the contribution of the second substance.

According to an embodiment, the device comprises a light source, a polarizer for linearly polarizing, in a polarization direction, light from the light source, a beam splitter, an analyzer for detecting light that has interacted with the scattering tissue in a detection direction that is perpendicular to the polarization direction. According to an embodiment, the device comprises patches to stretch the scattering tissue.

These and other aspects of the invention will be more apparent from the following description, with reference to the attached drawings.

Brief description of the drawings

- Fig.1 is a diagram showing an exemplary absorbance spectrum of skin, on which the substances contributing to the spectrum are quoted; - Fig.2 is a diagram representing the anisotropy factor of skin as a function of a stretching strain and

- Fig.3 is a schematic block diagram representing a device for implementing a first embodiment of the invention.

Detailed description of the embodiments

The invention will be described with relation to particular embodiments, where the concentration of glucose in blood is measured in vivo, within the skin of a person, the substance with scattering anisotropy being collagen. The method of the invention could also be applied to an animal.

A few definitions will firstly be given.

A polarizer is a device that permits to polarize light. In the method of the invention, light may be polarized linearly, that is to say, along a direction perpendicular to its trajectory direction. A linearly polarized light is a planar wave. An analyzer is a device that permits to detect backscattered or transmitted light in a particular polarization direction. In other words, an analyzer filters the backscattered or transmitted light in order to only get one polarization component of this light.

In the method of the invention, polarized incident light may be irradiated on skin, where it is backscattered or transmitted and collected into an analyzer so as to be detected. Two directions of detection of the backscattered or transmitted light by the analyzer may be interesting in the described embodiments of the method of the invention: the direction perpendicular to the polarization direction of the incident light and the direction parallel to the polarization direction of the incident light. Those two directions will be referred to as, respectively, the perpendicular detection direction and the parallel detection direction. Those directions are respectively perpendicular and parallel to the polarization direction of the incident light, whatever the polarization direction of the incident light is.

Some directions of polarization of the incident light will sometimes be referred to: the direction parallel to the collagen fibers and the direction perpendicular to another direction, notably. This has no link with the perpendicular or parallel nature of the direction of detection of the backscattered or transmitted light, which is always perpendicular or parallel to the polarization direction of the incident light, whatever the direction is.

The skin, on which spectroscopy measurements are performed in order to calculate the glucose concentration, will be referred to as the "probed skin" (and, notably, the probed skin volume or the probed skin sample); the corresponding reflectance (or transmission) and absorbance spectrum will be referred to as the "probed reflectance" and the "probed absorbance spectrum" or "probed spectrum". Similarly, for the skin in a normal state (that is to say, in an un-stretched state), on which measurements are performed in order to deduce the collagen contribution, the following expressions will be used: "un-stretched skin", "un-stretched reflectance" and "un-stretched absorbance spectrum" or "un-stretched spectrum". Similarly again, for the skin in a stretched state, on which measurements are performed in order to deduce the collagen contribution, the following expressions will be used: "stretched skin", "stretched reflectance" and "stretched absorbance spectrum" or "stretched spectrum". The probed, un-stretched and stretched skin samples may be different skin samples. They should be of the same type of skin, but not necessarily the same sample, that is to say, a sample from the same part of the body or a sample from the same person. According to an embodiment, the stretched and un-stretched skin samples are the same skin sample being un-stretched or stretched, so as to be sure that the scattering anisotropy is only linked to collagen and not to sample changes; however, approximations could be done in such a way that different samples are used for the un- stretched and stretched states. According to an embodiment, the probed skin sample is the same as the un-stretched and stretched skin samples; in such a case, the collagen contribution is calculated for a particular skin sample and not in general, providing personalized results; such a method is more precise but more complex to implement, compared to a method where preliminary measurements are performed beforehand in order to determinate the collagen contribution, optionally as an average over different samples, the collagen contribution being used in the subsequent measurements on the probed skin volume. According to an embodiment, measurements are done on a set of persons in un-stretched and stretched states and the results are averaged, in order to calculate an average collagen contribution. According to an embodiment, a database of un-stretched results and a database of stretched results are developed. The aforementioned concerning the probed, un-stretched and stretched skin samples applies to all the embodiments of the invention; therefore, it will not be referred to again, and it should be understood that, for each embodiment, those skin samples may be the same ones or not, averaged or not. In the described embodiments of the method of the invention, the measurements may be realized in vivo, that is to say, on the skin of a living person. The invention also applies to ex-vivo skin samples excised from a body.

NIR spectroscopy will be alluded to in the description; it is performed in a conventional manner. The skin sample is irradiated with an incident light beam and spectroscopy is performed with the light that has interacted with the skin sample; this light is scattered light that may either be backscattered light or transmitted light, depending on which side of the sample the scattered light is collected. In the embodiments described hereinafter, measurements are made on backscattered light and transmission will not be referred to in the following. However, it should be understood that measurements could also be performed on transmitted light; the person skilled in the art shall transpose easily. The backscattered light is detected within a spectrometer, where its diffuse reflectance is measured; the logarithm of this diffuse reflectance is proportional to the absorbance of the skin sample; an absorbance spectrum of the skin sample can therefore be obtained with the measured diffuse reflectance. The features of the absorbance spectrum are due to the scattering and absorption of light by all the components in the skin sample, in particular glucose and collagen fibers.

The spectrum of the probed skin volume is the basis for the calculation of the glucose concentration. However, as explained above, collagen has an influence on that spectrum and its contribution should be corrected in the calculation of the glucose concentration, in order to better distinguish the contribution of glucose and get a more acute measure of the glucose concentration.

The principle of the invention is the following: an absorbance spectrum is obtained for un-stretched skin and stretched skin, for instance by NIR spectroscopy. The contribution from the collagen fibrils of skin show anisotropy changes when mechanically loaded. Based on this anisotropy, the contribution of collagen is deduced from the stretched and un-stretched spectra. The collagen contribution can therefore be corrected in the calculation of the glucose concentration, which is made on the basis of the probed reflectance (of the probed skin sample).

The influence of stretching on the scattering anisotropy can be checked on Fig. 2, where the anisotropy factor (AF) is shown, as a function of the strain of stretch (which is expressed as a percentage of the original length of the stretched sample of skin). The anisotropy factor AF is herein defined as the following value: AF = I-L / 11 , where:

- I-L is the intensity of backscattered light in the perpendicular detection direction and - 11 is the intensity of backscattered light in the parallel detection direction.

We can note that, since the anisotropy factor is a ratio of intensities of backscattered light, it can also be expressed as the ratio of the corresponding reflectances in the perpendicular and parallel detection directions. As can be seen on Fig. 2, when the strain is 0%, the anisotropy factor is approximately equal to 1 , which means that light is backscattered the same way in all directions. When the strain is 20%, the anisotropy factor is approximately equal to 3, which means that the reflectance in the perpendicular detection direction is three times more important than the reflectance in the parallel detection direction. Scattering anisotropy of skin has therefore been affected by stretching. This is due to the fact that, since collagen is in the form of fibers, it is much more influenced by stretching than water, cells and the other skin components. Indeed, when skin is stretched out, the fibers are stretched along the stretching direction, whereas interstitial components are not really affected; this is in particular the case if the stretching is performed in the direction of the fibers, which therefore align more easily in the stretching direction. Hereby, the anisotropy of scattering from the collagen layers is affected, whereas the scattering of the interstitial components does not substantially change.

It is therefore assumed, in the presently described embodiments of the method of the invention, that only collagen is influenced by stretching, the influence of the other components being neglected. This is not totally exact and, if known, the influence of the other components under stretch could be taken into account to get more acute results. In the described embodiments of the invention, collagen is considered as the only skin component influenced by the stretching of skin. Such an approximation permits to conclude that the difference between un-stretched and stretched spectra is only due to collagen and is therefore representative of the collagen contribution in the skin spectrum. This collagen contribution can therefore be corrected in the calculation of the glucose concentration, in order to calculate this concentration more accurately.

An embodiment of the invention for measuring an un-stretched spectrum and a stretched spectrum will now be described, with reference to Fig.3.

A device 1 for implementing the method may comprise a NIR light source 2, a polarizer 3, a beam splitter 4, which permits to the incident NIR light 5 to pass through and reflects the backscattered light 6 into an analyzer 7, after which the light is received into a spectrometer 8. Two patches 9, 10 are used to stretch the skin 12 in a stretching direction 11. The patches 9, 10 may be formed of plastic or metal, for instance, and may be glued to the skin in order to drive it so as to stretch it when they are displaced. The patches 9, 10 may for example be displaced by 1 to 10mm between the un-stretched and stretched positions.

The method may be performed as described in the following. Firstly, an un-stretched absorbance spectrum is calculated. The skin 12 is left normal, that is to say, un-stretched. NIR light is emitted by the NIR light source 2 and linearly polarized by the polarizer 3. The polarized incident light 5 passes through the beam splitter 4 and is irradiated on the un-stretched skin 12. The backscattered light 6 is reflected by the beam splitter 4 and passes through the analyzer 7, which is tuned on the perpendicular detection direction. This filtered backscattered light enters the spectrometer 8 where its diffuse reflectance is measured and an un-stretched spectrum is calculated, by a mathematical model, in a conventional manner.

Secondly, a stretched spectrum is calculated. In that purpose, the patches 9, 10 are moved in the stretching direction 11 in order to stretch the skin. As well as before, an absorbance spectrum of this stretched skin is calculated, which is the stretched spectrum. The analyzer has a polarization detection direction perpendicular to the polarization direction of the incident light. Direct (specular) reflection of light does not change polarization and is therefore not detected in the perpendicular analyzer detection direction. Diffuse reflectance is depolarized with multiple scatter events and is thus detected. With perpendicular polarization detection with respect to incident polarization, the multiply scattered reflectance is therefore measured once with the unstretched skin and once with stretched skin. The difference spectrum is due to collagen stretch.

The scattering anisotropy is best detectable when the incident light is polarized parallel to the direction of the collagen fibers. As well, the stretching should preferably be performed parallel to the natural predominant direction of the collagen fibers. Now, as demonstrated by Karl Langer, the fibers in the skin have determined orientations, called the Langer's lines; in order to take this into account in the measure of the change in scattering of collagen due to the stretching, two embodiments may be contemplated.

According to a first embodiment, anisotropy factors (as shown in Fig.2) of the skin sample are calculated, so as to deduce therefrom the direction of the Langer's lines; indeed, since the anisotropy factor is linearly depending on the stretch strain, the values of the anisotropy factor close to the straight line of Fig.2 correspond to measurements which have been done with incident light polarized parallel to the Langer's lines. Once the direction of the Langer's lines has been determined, the measurements are performed on un-stretched skin and stretched skin with an incident light with a polarization parallel to the determined direction of the Langer's lines, light being detected in a perpendicular detection direction, in order to obtain the unstretched and stretched spectra.

According to a second embodiment, measurements (on un-stretched and stretched skin) are performed in two perpendicular directions, that is to say, measurements on un-stretched and stretched skin are performed with incident light with a first polarization direction (detection being performed in the perpendicular detection direction) and then the same measurements are performed with incident light with a second polarization direction perpendicular to the first polarization direction (detection being performed in the perpendicular detection direction, which is perpendicular with respect to the second incident polarization direction). The results obtained in the two directions are different. Those results may be summed, averaged or the most important result may be taken into account, in order to obtain the un-stretched and stretched spectra.

According to the method of the invention, the un-stretched and stretched spectra are used to deduce information on the collagen contribution, which is used to improve the quality of the calculation of the glucose concentration. In that goal, the contribution of collagen is corrected in the calculation of the concentration of glucose; the concentration of collagen in skin may be calculated and used to correct the contribution of collagen. The calculation of the glucose concentration is based on applying a mathematical model to an absorbance spectrum of skin obtained by NIR spectroscopy. According to a first calculation embodiment, this calculation is corrected by directly correcting the absorbance spectrum of skin with a spectrum of collagen, corresponding to its contribution, and then applying a mathematical model to this corrected spectrum. According to a second calculation embodiment, the calculation is corrected by calculating the concentration of collagen and using it as an additional input to an adapted mathematical model, so that the mathematical model corrects the contribution of collagen in the calculation of the concentration of glucose, the mathematical model having, as inputs, the absorbance spectrum of skin and the concentration of collagen.

Before describing in details those two calculation embodiments of the invention, an example of an application of a mathematical model to a spectrum, in order to obtain the value of the glucose concentration, according to the prior art, will now be described. Such methods are known by the person skilled in the art and this is why only the principles of such a method will be described, in order to help for the subsequent explanations of the embodiments of the invention. The person skilled in the art will have no difficulty for materially implementing the method of the invention if he knows the main principles of it, and how the prior art principles are modified and applied to the invention. This prior art description is therefore a way to give a few definitions in that goal.

Two steps are generally performed: 1) developing the mathematical model;

2) applying this mathematical model to the measured absorbance spectrum. A chemometric mathematical model may be used. The International Chemometrics Society (ICS) defines chemometrics as the science of relating measurements made on a chemical system or process to the state of the system via application of mathematical or statistical methods. For the purpose of the invention, a chemometric mathematical model is used to relate an absorbance spectrum to the value of the glucose concentration.

The mathematical model may be developed with a "partial least squares" (PLS) regression method, well known by the person skilled in the art. Other examples of chemometric models are principal component regression, principal components analysis, genetic algorithms, artificial neural networks, support vector models, etc., which are all well known in the art. The example of PLS will be described. In order to develop the model, NIR measurements are performed on skin samples of which the concentration of glucose is known, for instance, by a reference method, such as fingerstick blood glucose meters, or by blood analysis with standard clinical laboratory methods. The objective of the PLS-regression method is to calculate the vector, called the regression vector, which represents the translation (or correlation) between an absorbance spectrum and the corresponding concentration of glucose. Each substance (here glucose) is related to a particular regression vector.

An absorbance spectrum is represented by a vector (absorbance values versus wavelengths). To develop the model, several absorption spectra are measured on skin samples, which have calibrated glucose concentrations, for instance known by reference methods as explained above. On the one hand, all those spectra are gathered into a matrix, which we will call the calibration matrix. On the other hand, the corresponding calibrated glucose concentrations are gathered into a vector, where each element corresponds to a calibrated glucose concentration; we will call this vector the calibration vector. The PLS-regression method is then applied in order to calculate the regression vector, which represents the translation from the calibration matrix to the calibration vector.

Once the regression vector is obtained, it is possible to perform measurements on unknown skin samples and predict therefrom the glucose concentration of those skin samples. Indeed, spectroscopy is performed, providing an absorbance spectrum, which can be represented in the form of a vector, which is taken as the input of the mathematical model. With the model, the vector of the absorbance spectrum is multiplied by the regression vector, this multiplication resulting in the value of the unknown glucose concentration in the measured skin sample.

The two calculation embodiments announced above will now be described. In both cases, we have seen that the absorbance spectrum of the probed skin volume has been obtained by NIR spectroscopy, while the contribution or the concentration of collagen has been determined by a comparison between un-stretched and stretched spectra of skin samples. A detailed description of how to obtain the collagen contribution or concentration will be described later on and, in the present description of the calculation embodiments, this collagen contribution or concentration is considered as already determined.

According to the first calculation embodiment, a certain spectrum - corresponding to the contribution of collagen and which will be called the collagen contribution spectrum - is subtracted from the absorbance spectrum of the probed skin volume obtained by NIR spectroscopy. A new spectrum, which will be called the corrected absorbance spectrum, is obtained, which contains the peaks of all the substances in the skin volume, except the collagen features, which have been subtracted. The concentration of glucose can be calculated therefrom, not being masked by the peaks of collagen. According to an embodiment, in order to build the collagen contribution spectrum, the collagen concentration in skin may be calculated on the basis of its contribution to the un-stretched and stretched spectra. The concentration of collagen may then be used with the known absorbance spectrum of pure collagen in order to get the collagen contribution spectrum. According to another embodiment, the collagen contribution spectrum is directly derived from the comparison between the un- stretched and stretched spectra.

More precisely, the corrected absorbance spectrum is used as the entry of a chemometric mathematical model, so as to obtain the glucose concentration. As described above, this mathematical model has to be developed on beforehand, with calibration skin samples. The calculated regression vector will be different from the regression vector described above for the prior art mathematical model because, according to this embodiment, the regression vector has to be applied on an absorbance spectrum from which the peaks of collagen have been removed. The development of the regression vector is done with calibration skin samples for which the glucose as well as the collagen concentrations are known. The collagen concentration of the calibration skin samples may be determined by way of Monte Carlo simulations, as will be explained later on. NIR spectroscopy is performed on each calibration sample, in order to get a calibration absorbance spectrum. The collagen contribution spectrum - which is known since the collagen concentration is known - is subtracted from this calibration absorbance spectrum in order to obtain a calibration corrected absorbance spectrum. The calibration corrected absorption spectra of the different samples are filled into a calibration matrix. A calibration vector is provided, which comprises the glucose concentrations of the calibration skin samples. As well as before, the regression vector is calculated as the vector representing the translation from the calibration matrix to the calibration vector.

In order to calculate the (unknown) glucose concentration of the probed skin volume, the corrected absorbance spectrum of the probed skin volume is entered as an input of the chemo metric mathematical model which has been developed as explained just above, where it is multiplied by the regression vector, resulting in the value of the glucose concentration in the probed skin volume.

According to the second calculation embodiment, the concentration of collagen in skin is calculated - thanks to the un-stretched and stretched spectra - and is used as an additional input for the chemometric mathematical model that serves to calculate the glucose concentration with the probed spectrum. The mathematical model therefore corrects the contribution of collagen in its calculation of the glucose concentration. More precisely, a regression vector has to be calculated for the mathematical model in order to permit, by inputting a probed absorbance spectrum and a collagen concentration, to obtain the glucose concentration of the corresponding probed skin volume. The regression vector has to take into account the collagen concentration in its calculation of the glucose concentration. Such a model is developed as follows and, again, by means of a PLS regression method.

The development of the regression vector is done with calibration skin samples for which the glucose as well as the collagen concentrations are known by another method; samples with various glucose concentrations as well as various collagen concentrations are tested. NIR spectroscopy measurements are performed on these calibration samples, in order to fill a calibration matrix, which comprises the absorption spectra data as well as the corresponding collagen concentrations of those calibration samples. Compared to the calibration matrixes that have been presented above, this calibration matrix has an additional dimension corresponding to the collagen concentration data. The calibration vector comprises the glucose concentrations of the calibration skin samples. As well as before, the regression vector is calculated as the vector representing the translation from the calibration matrix to the calibration vector. It also has an additional dimension, since it takes into account the concentration of collagen as an input.

In order to calculate the glucose concentration of the probed skin volume, the probed absorbance spectrum and the calculated collagen concentration are entered as inputs into the chemometric mathematical model, as a vector, which is multiplied by the regression vector, resulting in the value of the glucose concentration.

As an alternative to the second calculation embodiment, the whole collagen spectrum could be used for the development step of the regression vector, thus using a calibration matrix having double size. Such an embodiment will not be further developed herein. As explained above, the principle of the invention is based on obtaining an absorbance spectrum for un-stretched skin and stretched skin in order to deduce therefrom the contribution or concentration of collagen is the skin sample. The calculation step of the contribution or concentration of collagen may be carried out according different embodiments, three of which will be described here. Whatever the embodiment, having a priori precise information on the spectral positions of the collagen, glucose and other components is beneficial, in order to prevent over or under estimating the glucose concentration in the corrected spectra.

According to a first embodiment, the collagen contribution is determined with the help of Monte Carlo calculations; this type of calculation is known by the person skilled in the art and will not be described in a very detailed manner. In Monte Carlo calculations, the amount of collagen and its bands in the NIR spectrum of skin can be varied and simulated. This gives in the collagen band spectral region precise information on the band position, band width and band height for different concentrations of collagen in skin. A database (or Look-Up-Table) can hence be generated that contains all kinds of (simulated) possible spectra of the skin under various skin composition conditions, in stretched and un-stretched conditions. The measured spectra in the stretched and un-stretched states are compared with the database in order to determine to which concentration of collagen the spectra correspond. The concentration of collagen is hence determined, which permits to calculate, with the help of the spectrum of pure collagen, the collagen contribution spectrum, which can be subtracted from the probed spectrum. The forward Monte Carlo simulations have the advantage that the various experimental conditions

(stretched/unstretched skin) can be simulated easily, with a noise level that can be adjusted by the operator. In the Monte Carlo calculations, the stretched and un- stretched spectra are compared to simulated spectra in order to determine the contribution of collagen. According to a second embodiment, bandfϊtting of the spectra is performed, where each of the contributing components is simulated and varied until an acceptable level of fit is obtained. This is done for the un-stretched spectrum and the stretched spectrum, keeping in mind that only the collagen contribution should be changed to pass from the un-stretched spectrum to the stretched spectrum. From this comparison, the collagen contribution is the un-stretched spectrum can be determined. Again, the stretched and un-stretched spectra are compared to simulated spectra in order to determine the contribution of the second substance.

According to a third embodiment, multivariate analysis is used to decompose the spectra in their principal components. Measurements are made on un-stretched samples and measurements are made on stretched samples. For each spectrum, the main components contribution is first looked for, therefore the water contribution is firstly determined, then the fat and collagen contributions are determined. A chemo metric software tool or package then correlates the variations between the un- stretched and stretched spectra in order to deduce the collagen contribution. Then, loading vectors containing the spectral information on the collagen are subtracted from the probed spectra. Examples of methods that may be used for the determination of the collagen contribution are fitting, inverse Monte Carlo, genetic algorithms, look up tables, partial least-squares or principal component regression.

While the invention has been illustrated and described in detail in the drawings and foregoing description, such illustration and description are to be considered illustrative or exemplary and not restrictive; the invention is not limited to the disclosed embodiments.

Other variations to the disclosed embodiments can be understood and effected by those skilled in the art in practicing the claimed invention, from a study of the drawings, the disclosure, and the appended claims. In the claims, the word

"comprising" does not exclude other elements or steps, and the indefinite article "a" or "an" does not exclude a plurality. A single processor or other unit may fulfill the functions of several items recited in the claims. The mere fact that certain measures are recited in mutually different dependent claims does not indicate that a combination of these measured cannot be used to advantage. A computer program may be stored/distributed on a suitable medium, such as an optical storage medium or a solid- state medium supplied together with or as part of other hardware, but may also be distributed in other forms, such as via the Internet or other wired or wireless telecommunication systems. Any reference signs in the claims should not be construed as limiting the scope.

Claims

1- Method for calculating the concentration of a first substance in a scattering tissue (12), the method comprising: a) measuring a first absorbance spectrum of the scattering tissue (12) in a first stretching state, b) measuring a second absorbance spectrum of the scattering tissue (12) in a second stretching state, c) determining, from the first and second stretched spectra, the contribution to the spectrum of the scattering tissue (12) of a second substance present in the tissue, and d) determining the concentration of the first substance by applying a mathematical model to an absorbance spectrum of the scattering tissue obtained by spectroscopy after correction of the contribution to the spectrum of the second substance.

2- Method according to claim 1, wherein the first stretching state is an un- stretched state and the second stretching state is a stretched state.

3- Method according to claim 1, where the scattering tissue is skin (12), the first substance is glucose and the second substance is collagen.

4- Method according to claim 1, wherein steps a) and b), and spectroscopy measurement in step d) are performed on the same scattering tissue sample.

5- Method according to claim 1, wherein steps a) and b) are performed on a plurality of tissue samples and the measurements are averaged. 6- Method according to claim 1, wherein the scattering tissue is skin (12) and the second substance is collagen and the irradiating light is polarized substantially in the direction of the collagen fibers and/or stretching is performed substantially in the direction of the collagen fibers.

7- Method according to claim 1, wherein the contribution of the second substance is determined from a comparison of the first and second spectra with simulated spectra.

8- Method according to claim 1, wherein spectroscopy is near infrared spectroscopy.

9- Device for calculating, for a scattering tissue (12) comprising at least a first substance and a second substance that presents scattering anisotropy under stretch, the concentration of the first substance, the method comprising: a spectrometer for measuring a first absorbance spectrum of the scattering tissue (12), in a first stretching state and a second absorbance spectrum of the scattering tissue (12) in a second stretching state, a processing arrangement that determines from the first and second spectra, the contribution of the second substance in an absorbance spectrum of the scattering tissue obtained by the spectrometer and computes the concentration of the first substance by applying a mathematical model to the absorbance spectrum after correction of the contribution of the second substance.

10- Device according to claim 9, comprising: a light source (2) for illuminating the scattering tissue; a polarizer (3) for linearly polarizing, in a polarization direction, light from the light source (2); an analyzer (7) coupled to the spectrometer for detecting light that has interacted with the scattering tissue (12) in a detection direction that is perpendicular to the polarization direction. 11- Device according to claim 10, which comprises patches (9, 10) to stretch the scattering tissue.