Method for determining a substance concentration and detector device

Abstract

The invention concerns a method for determining a substance concentration in a sample comprising liquid containing particles, in particular glucose in blood, wherein a refractive index of the liquid is dependent on a concentration of the substance dissolved therein and a density of particles in the liquid is substantially constant. The method comprises obtaining a PPG first signal during a first time period and estimating at last a portion of a noise spectrum from the first obtained signal in particular the frequency domain. At least a portion of the noise spectrum is used to derive the substance concentration or a change in the substance concentration.

Description

[0001] METHOD FOR DETERMINING A SUBSTANCE CONCENTRATION AND DETECTOR DEVICE

[0002] The present application claims priority from German application DE 10 2023 130 812 . 2 dated November 7 , 2023 , the disclosure of which is incorporated in its entirety herein by reference . The present invention relates to a method for determining a substance concentration in a sample containing particles in a liquid, in particular glucose in blood, wherein a refractive index of the liquid is dependent on a concentration of the substance dissolved therein . The invention also relates to a detector device .

[0003] BACKGROUND

[0004] The current standard for blood glucose measurement often uses an invasive technique in which a small amount of blood is drawn, and subsequent electrochemical analysis is performed using a handheld device . This method is not suitable for continuous monitoring because for each measurement , the finger must be pricked to obtain a fresh blood sample .

[0005] A more recently developed technology uses a button that sits on the skin and misses interstitial fluid in parts of the subcutaneous adipose tissue with a small , needle-like sensor . However , the needle penetrates permanently into the skin . Such approach bears the ris k of infection . Moreover , it may have to be removed under certain circumstances , e . g . during sport activities , swimming and the like .

[0006] Besides these invasive methods , there are also non-invasive methods based on optical IR measurements or Raman spectroscopy . While in the first case , a suitable choice of emitter and detector leads to difficulties , an approach based on Raman spectroscopy is challenging due to the very poor signal-to-noise ratio . Recently, optical measurements using portions of the visible spectrum have been proposed, in which the back scattered light is analyzed . However , depending on skin type , environment and other parameters , the signal / noise ration may vary and degrade the overall signal quality significantly . In this respect , there is a need for a method that can detect a substance in a liquid in a simpler way and allows for a continuous measurement .

[0007] SUMMARY OF THE INVENTION

[0008] This and other obj ects are addressed by the subj ect matter of the independent claims . Features and further aspects of the proposed principles are outlined in the dependent claims .

[0009] Complementing some other ideas that are based on various parameters depending on the scattering of light within the tissue , the present idea relies on the evaluation of the noise level of the modulation depth, the latter also referred to as perfusion index . The inventor surprisingly found that not only the perfusion index itself depends on the blood glucose concentration, but also the noise itself . This seems to be due to the fact that the scattering introduces systematic noise in the overall noise , which relies on the glucose concentration .

[0010] The perfusion index or modulation depth is a result during so-called Photoplethysmography or PPG, i . e . , the optical measurement of volume changes of human blood, e . g . , due to the beating of the heart , at a certain point of the human body, for instance the fingertip , or some other suitable location . The volume changes are observable by signal variations during the heartbeats resulting from less or more scattering based on the different blood volume . The PPG measurement itself is performed during a short period of time , e . g . 10s in intervals of 2 or 5 minutes for example .

[0011] Due to scattering and absorption of light , which is initially directed towards the skin and may subsequently propagate through portions of shallow- and / or deeper lying tissue , which in turn hosts a network of blood vessels , the amount of light that re-emerges at the s kin-to- ambient interface some distance from the entry point will vary according to the optical path length ( distance ) travelled in blood . The path length is referred to as "blood optical path length or ( BOPL ) , " , which in turn is directly affected by the beating of the heart . Consequently, the BOPL varies with time , while the properties of surrounding portions of tissue can be considered constant , at least on the timescale of the heart rate and the short measurement .

[0012] When the BOPL is at a maximum or minimum, the emerging signal after interaction is conversely at a minimum or maximum . The difference between minimum and maximum signal is called modulation, and the ratio of modulation to average signal is defined as modulation depth, ac-to- dc ratio , or perfusion index .

[0013] According to Beer-Lambert ' s law, the absorption of light in blood is largely determined by the presence of Hemoglobin within the erythrocytes , as well as the cumulative BOPL . Under the assumption that the density of erythrocytes does not change during the measurement period, one can say that light propagating nominally through the more or less transparent blood plasma, undergoes scattering at the red blood cells , such that the cumulative BOPL from entry point to detection point , both of which are fixed by the device configuration, may depend on the scattering characteristics in the sense of a random walk, i . e . , it will become a stochastic process .

[0014] Consequently, the perfusion index or modulation depth is primarily a result of direct interaction and particularly the scattering between light and erythrocytes . This principle is currently used for measurements of the oxygen saturation in blood, also referred as SPO2 measurement . SPO2 measurements are based on the ratio of PPG signals recorded at two different wavelengths , i . e . in the red and IR spectrum . More specifically, the SPO2 measurement is performed by generating the ratio of perfusion indices at the two different wavelengths . This is accomplished with a setup, where a red and an IR emitter for example are used to illuminate tissue , e . g . , by alternately transmitting short pulses of light at a sufficiently high repetition rate , e . g . , 20 Hz , 50 Hz , 200 Hz , etc . , and sampling the corresponding photo current generated in the photodiode . The raw photodiode current consists of an average signal level , referred to as DC, and a modulation of the signal due to the beating of the heart , referred to as AC . The AC signal can be estimated, for instance , by the difference between the peaks and the valleys of the raw signal . The ratio of AC to DC is called the perfusion index ( PI ) or relative modulation or modulation-percent . The reference "In Vivo Reflectance of Blood and Tissue as a Function of light wavelength" , Wei ia Cui et al . IEEE Transactions on biomedical engineering, Vol 17 , no 6 . June 1990 . The ratio of the relative modulations is computed to estimate the oxygen saturation . This process is also called the ratio of ratios . By proper calibration of such values with a reference device , one can identify early 100% of oxygen saturation and lower levels thereof .

[0015] However, the ratio of ratios for a given oxygen saturation and thus an SPO2 measurement depends on optical design choices , in particular the relative separation between the light source and detector for each wavelength, the emission characteristics of each light source , the detection characteristics of the photo diode , and the performance of the optical barrier in between sources and detectors , as well as other system aspects that affect how light may propagate from a source to a detector . Still , the above-mentioned approach to optical SPO2 measurements is widely discussed in the literature and scientifically accepted .

[0016] It has now been observed that for a human test subj ect under steadystate conditions and constant blood glucose levels , the perfusion index remains substantially constant . When the blood glucose level increases , the properties of the blood plasma and more particularly the refractive index of the plasma changes . As a result , the difference between the refractive indices of the erythrocytes and the blood plasma decreases , which will cause a decrease of scattering light . Likewise , a decrease in the glucose increases the difference between the refractive indices of the erythrocytes and the blood plasma causes an increase in scattering . This is explained in a model , in which scattering is mainly caused by diffraction differences , thus in a transparent medium, in which the refractive indices of the liquid and the scattering particles were the same , no scattering would be observable , since light would simply propagate through the medium without interaction . However, once there is an index mismatch, some light is being scattered ( e . g . , backwards ) by the particles and thus can be distinguished from the surrounding ambient , i . e . , plasma .

[0017] Consequently, one can use the perfusion index for so-called Oral Glucose Tolerance Test ( OGTT ) , in which a test subj ect starts with a baseline blood glucose level established by not eating for several hours followed by ingesting a liquid or substance of caloric value . Usually, changes in the perfusion index become visible after around 5 to 15 minutes after the intake .

[0018] It has now been observed that the overall noise level is changing with the blood glucose level as well . More particularly, using sigma or the standard deviation, which is a measure of the noise in the remainder of the spectrum, i . e . , that portion not modulated by the heartbeat , correlates with the blood Glucose concentration . One can therefore conclude that noise in the PPG signal might be a carrier of additional information about a substance and its concentration in the liquid .

[0019] Specifically, the proposed principle utilizes an algorithm for signal decomposition into constituents that are ( 1 ) attributed to modulation by the heartbeat and ( 2 ) a remainder, which is commonly referred to as noise and not considered a carrier of relevant information . These constituents are monitored during a so-called oral Glucose tolerance test , and the noise is correlated to Glucose reference measurements performed during the same period with an Abott Freestyle Libre 3 system . Signal descriptors , such as signal-to-noise ratio ( SNR) , standard deviation (□ ) , perfusion index ( PI ) , etc . , are calculated from the decomposed signals .

[0020] Apparently, the tissue become more transparent as the Glucose level or more general the level of the substance to be measured increases . This is evidenced by the decreasing DC level during the same period when using a configuration in reflection mode . The observed behavior may indicate that the detected light has interacted with deeper lying tissue layers and that various physiological mechanisms , such as blood flow, muscle activity, etc . , could add relatively more noise to the system . This interaction with deeper layers is in contrast with superficial layers , where there is little , or no blood flow and the skin is composed of rather passive material layers .

[0021] One could hypothesize that increasing the glucose concentration opens a sort of optical window for optical measurements towards the deeper lying layers of tissue . Such a hypothesis is corroborated by the fact that the PPG modulation amplitude (AC, PI ) increases substantially (here 3 . 8x for example ) when sugar levels increase : this is only possible when the portion of the signal modulated by the pulsating blood is gaining in weight .

[0022] As a result of this observation, the inventor proposes a novel method for determining a substance concentration in a sample comprising liquid containing particles , in particular glucose in blood, wherein a refractive index of the liquid is dependent on a concentration of the substance dissolved therein and a density of particles in the liquid is substantially constant . The proposed method is based on the extraction of a noise spectrum in the PPG data, and particularly the separation of the signal portion from the noise portion in the frequency spectrum of a PPG measurement .

[0023] While the method was developed mainly for Glucose measurement , it is not limited thereon, but can be generalized into a method for determining a substance dissolved in a liquid containing particles as long as the liquid is excited by a pressure or volume modulation .

[0024] By obtaining the standard deviation from the noise spectrum and correlating it to the blood Glucose or generally to the substance concentration in the liquid, one can obtain the change of the substance level and if a base reference is known also the overall absolute value .

[0025] Several further results can be obtained from such an approach . For example , one can quantify the deviation from a shot-noise limited performance by the ratio sigma2 / DC and correlate the deviation from shot-noise performance to the substance concentration in blood . In some further aspect , one could also utilize the pure modulation signal , calculate the heart rate from it and correlate it to the blood glucose concentration . Finally, it is possible to calculate the so-called modulation-to-noise ratio , which quantifies the excess modulation of the signal due to the heartbeat over the system noise , including the noise from the specimen .

[0026] In some aspects , a method is proposed for determining a substance concentration in a sample comprising liquid containing particles , in particular glucose in blood, wherein a refractive index of the liquid is dependent on a concentration of the substance dissolved therein and a density of particles in the liquid is substantially constant . In the proposed method, the liquid is modulated in its volume or in its density . If the liquid is blood, with a substance concentration in blood to be measured, the heartbeat provides the required modulation .

[0027] For determining the substance concentration, a first signal is obtained during a first time period . The acquisition rate of the first obtained signal is set to be at least two times larger than the periodicity or frequency of the liquid ' s modulation to fulfil the Nyquist criteria . For example , the acquisition rate may be 10 times , 20 times or even 100 times larger . Usually, the acquisition rate is at least 40 times larger than the modulation frequency . In the case of blood with a heartbeat of approximately 1 Hz to 2Hz , the acquisition rate may be at least 10 Hz , and more particularly between 20 Hz and 200 Hz .

[0028] Likewise , the first time period is adj usted that a plurality of periodicity of the liquid volume or pressure modulation fit into the first time period . For example , the first time period may be between 3 times and 30 times larger than the time period of the liquid' s modulation . In case of heartbeat with a frequency of 1 Hz , the acquisition time may be between 5 seconds and 20 seconds .

[0029] This approach ensures that noise with a sufficient resolution and enough signal peaks are captured during the first time period and the acquisition of the first signal .

[0030] In a second step , at least a portion of the noise spectrum is extracted from the obtained signal , in particularly in the frequency domain . The estimation provides at least a frequency dependent amplitude of the noise portion in the first obtained signal .

[0031] With regard to the term "portion" , it is understood that the overall signal spectrum itself is band limited due to the real-world signal processing . It contains a noise portion, the actual signal referred to as AC portion and a DC portion . For the proposed method, a portion of the noise spectrum is obtained, estimated or extracted ( these terms are used synonymously in the present application ) from the overall signal spectrum. Various approaches are possible for such extraction, some of which are presented herein . This noise portion extracted can be further band limited and only comprise the noise up to a certain frequency . The reason for this approach lies in the fact that systematic noise (which is the interesting portion for the proposed method ) decays with 1 / f and higher frequency noise does not correlate with the substance concentration .

[0032] In a final step, the substance concentration is estimated and determined based on the extracted portion of the noise spectrum and a reference value . Alternatively, a change in the substance concentration can be determined based on the extracted portion of the noise spectrum and a reference value . In this regard, the reference value may be zero or any other base value . For example , the reference value may correspond to a ground or normal level of the substance concentration in the liquid .

[0033] The proposed method utilizes the fact that under certain circumstances the substance concentration can be determined by a deviation from a shot noise limited system, because the substance or a change thereof introduce an additional term of systematic noise . The approach may be beneficial in highly noisy system or when the signal-to-noise ratio significantly deteriorates .

[0034] In some aspects , the reference value is based on a processed second signal . Consequently, the proposed method includes the step of obtaining a second signal during a second time period with an acquisition rate that is at least two times larger than a periodicity of the volume or pressure modulation of the liquid . The second time period can be after or prior to the first time period, with a certain time gap between the measurements . Further measurements can be performed to obtain a plurality of such signals . It is suitable that the first and second time periods are the same to reduce possible residual effects , although the subsequent method step can compensate for different periodic measurement , e . g . by normalization .

[0035] A portion of a noise spectrum is extracted from the second obtained signal in particular the frequency domain . The reference value is based in some aspects on the extracted portion of the noise spectrum of the second obtained signal . This allows to obtain a change in substance concentration between two or more measurements . In some cases , in which the substance concentration changes over time , one can obtain the change rate and other parameters of the substance concentration by evaluating a plurality of obtained signal in accordance with the proposed method . Likewise , the reference value is given in some aspects given by a pre-determined estimated portion, or a value derived by such estimated portion .

[0036] In some aspects , the step of estimating at last a portion of a noise spectrum comprises the steps of identifying AC portions in the spectrum of the obtained signal , said AC portions corresponding to the fundamental frequencies and its harmonics of the liquid' s volume or pressure modulation . Usually, the harmonics appear at multiples of the fundamental frequency . The Identified AC portions are then removed or separated from the spectrum, leaving the noise portion and the DC portion .

[0037] It should be noted in this regard, that noise is also present at frequencies corresponding to the fundamental and the harmonics of the liquid' s volume or pressure modulation . One could say, the noise portion is buried beneath the actual signal peaks . Simply removing the peaks at those frequencies , will also remove or separate that noise portion, which is not desired as it contains valuable information . Consequently, the inventor proposes an approach to estimate the noise portion beneath the peaks at the fundamental frequency and the harmonics . For this purpose , the step of removing the identified AC portion from the spectrum comprises the step of estimating the noise portion within the spectrum at the fundamental frequencies and its harmonics of the liquid' s volume or pressure modulation based on the values of adj acent points in the spectrum. This step first uses the peak width of the fundamental frequency or a given mas k to define the width of the noise portion to be estimated . The estimation itself uses the noise directly adj acent to the respective peaks . It is based on the assumption that the noise beneath the peaks is constant . As an alternative approach, one can estimate the noise portion under the masked portions of the spectrum based on points in the spectrum at the mas ked portion' s borders .

[0038] Hence , gaps in the noise spectrum caused by removal or separation of the AC portion are filled with the average value of the adj acent points .

[0039] In some further aspects , signal processing to identifying the AC portion in the spectrum of the first and second obtained signal may contain several different tasks . These can be either used separately or combined with other steps to be able to distinguish the peak signal from a noise portion .

[0040] In some aspects , a smoothing function is applied onto the spectrum of the first and second obtained signal . Smoothing reduces the noise , hence it should not be too strong , but also helps to reduce outliers . The smoothing function can be applied prior to obtaining the spectrum to even out a potentially j agged structure of the harmonic .

[0041] In some aspects , at least the peak of the fundamental frequencies maybe enhanced, in particular by taking the absolute values of the spectrum to the power of M with M being an integer larger than 2 . This is especially useful when the data is noisier and the peaks in the frequency spectrum are only marginally protruding over the adj acent noise . By applying this function, the peaks are significantly enhanced over the noise , which also simplify the subsequent process steps . In some aspects , a mas k is determined with a width based on a predetermined amplitude level of at least the peak of the fundamental frequencies . For this purpose , a base level may be set to define the mask borders , for example at 10% of the peak level , or at 5 % of the peak level . Other levels are possible like3% , 2% or even 1% and below . The base level set to define the noise can be dependent on the overall signal level , i . e . a portion thereof , and also depend on the enhancement level .

[0042] In some aspects , the dynamic peak width is determined given for example by fmax-- fmin, whereas the frequencies are the given by the respective base level at the increasing edge and decreasing edge of the respective peak . In some aspects , a static offset may be added to the dynamic width, but the advantage is that the offset needs to be determined only once for all measurements .

[0043] In some aspects , a center of gravity of at least the peak of the fundamental frequencies is determined, optionally by using the width based on a pre-determined amplitude level of at least the peak of the fundamental frequencies . The center may be determined by summing the product of frequency and FFT amplitude over the width of the fundamental frequency that has been determined previously . The result is then divided by the sum of the amplitudes . It should be noted that the center of gravity of the fundamental peak may not necessarily correspond with the maximum this would only be the case for symmetric peaks , which are rarely the case . The center of gravity of the fundamental frequency also provides the information used to identify the correct frequencies for the harmonics .

[0044] The mask with the determined width is applied over a predetermined number of possible peaks and more particularly over the respective center of gravity of the fundamental and then repeated at the frequency of the harmonics . The peaks of the harmonics are calculated from the fundamental frequency within the spectrum, which has been determined as discussed above . Consequently, the mask covers a predetermined number of peaks corresponding to the fundamental frequencies and its harmonics of the liquid' s volume or pressure modulation, thereby identifying the AC portion.

[0045] The AC portions, as mentioned above determined by this approach, can then be separated from the spectrum. Not all harmonics are visible or contribute to the spectrum due to the noise level. Hence, it may be sufficient in most cases to consider only the fundamental frequency and its harmonics up to the seventh order and in particular up to the fifth order and in particular up to the fourth order. However this is not limited thereto, as some applications may require even higher orders, such as the 10th harmonic, or more. This may depend on where measurements are taken (e.g. , finger vs wrist) , and on the wavelengths. Depending on the position, a relatively high number of harmonics can be resolved.

[0046] In some aspects, the first and second signals are pre-processed. For example, the first and second signal may be filtered to remove transients, in particular with a frequency below 2Hz and in particular below 1 Hz and in particular below 0.5 Hz. Rate of 0.5 Hz or below for heartbeat may be quite rare. Breathing is typically in the .2 Hz range (5s per breathing cycle / 12 breaths per minute) and falls into the <.5Hz range. Those transients may be generated by breathing, for example in case of measurements with blood. In some instances, the pre-processing is performed by applying a moving mean filter with a window width of about a heartbeat, which captures the slow transients. Then, the transients are removed and DC portion added back-in to generate a PPG signal that is quite even but still retains the AC-modulation due to the heartbeat.

[0047] In some aspects, a background subtraction has been performed onto the measured signals to remove dark-counts and ambient contributions. The first and second signals may be converted into a digital signal prior to further processing. The above-mentioned pre-processing can be performed in the analogue domain (e.g. the low pass filtering and background subtraction) , while the Fourier transformation for extracting the noise portion is done in the digital domain. In some aspects, the first and second obtained signal are processed using a Fast Fourier transformation.

[0048] Some aspects concern the calculation and determination of the substance concentration or a change in the substance concentration. In some aspects, the standard deviation of the estimated portion of the noise spectrum is obtained. The standard deviation o is given by

[0049] D / N i ■ 2 . . . o = — lZji=2\noisei\ with N being the number of samples within the frequency range of interest.

[0050] Let M be the number of measurement points per data track, e.g. 2048, measured at a rate of 200Hz for a measurement time of ~10s (~10 heartbeats) . This results in a frequency scale of (-M / 2, M / 2) / 2048 x 200 Hz = ( -1 / 2 , 1 / 2 ) *200Hz = ( -100Hz , 100Hz ) in 2048 (frequency) steps.

[0051] A frequency step in this example would be then 200Hz / 2048 ~.lHz. If one wants to refine the frequency steps / resolution, then simply increasing of the sampling rate is not sufficient, rather it is necessary to measure over a longer period of time, e.g. 4096 points at 200Hz.

[0052] If one sets N = M, then one gets the standard deviation obtained by the usual formula in the time domain; this is Parseval's theorem. Uinsg this relation it is possible to evaluate the standard deviation over the entire frequency range, or only over part of it. It determines the starting point (i=2, or another number) and the end point: e.g. at 0.5 and 10Hz, N = ~102 in the example (200Hz, 2048 points) and the starting point is at ~5.

[0053] The standard deviation correlates with the substance concentration, hence with a given reference value, one can directly obtain the substance concentration or a change by simply evaluating the standard deviation. In some aspects, the signal-to-noise ratio or the inverse thereof of the estimated portion of the noise spectrum and the reference value can be used as such is also correlated with the substance concentration . In some aspects , the substance concentration or the change of substance concentration can be obtained by evaluating a quantification of deviation from a shot-noise limited system defined by the ratio of the standard deviation of the estimated portion of the noise spectrum. As mentioned before , due to the correlation, there is a systematic noise portion that deviates from a classical conventional shot noise limited system . Several signal descriptors are illustrated herein, which are suitable for the determination of the substance concentration or a change of the substance concentration in accordance with the proposed principle .

[0054] In some other aspects , the proposed method further comprises calculating a modulation to noise ratio given by the AC portion over the estimated noise . The modulation to noise ration or MNR quantifies an excess modulation of the signal due to the heartbeat over the system noise .

[0055] The proposed method can be applied particularly for the determination of substances in blood, like glucose or 02 . In such cases , where the liquid is blood, the modulation is given by the heartbeat resulting in pressure and volume changes of the blood . Obtaining such a signal refers to PPG a measurement . In some aspects , the heart rate is calculated from the identified AC portions in the spectrum. It has also been found that heart rate correlates with the blood glucose concentration, meaning that the heart rate seems to change with increasing or decreasing blood glucose concentration .

[0056] Some aspects concern a detector arrangement . The detector arrangement may be implemented in a single housing , like a watch or medical instrument , but may also comprise distributed components . In some aspects , a detector arrangement is proposed for determining a substance concentration in a sample comprising liquid containing particles , in particular glucose in blood, wherein a refractive index of the liquid is dependent on a concentration of the substance dissolved therein and a density of particles in the liquid is substantially constant . The detector arrangement comprises at least one light source and at least one detecting component , wherein the at least one detecting component is optically separated from the at least one light source . The at least one light source is configured to emit light through an exit window onto a sample containing the liquid with the substance .

[0057] The detecting component is configured to detect a light component corresponding to emitted light scattered through the particles containing liquid . The detector arrangement further comprises a control circuit coupled to the at least one light source and at least one detecting component . The detector arrangement is configured to perform the above proposed method steps .

[0058] In some further aspects , the detector arrangement may comprise an evaluation circuit for extracting at last a portion of a noise spectrum from the first obtained signal , particularly in the frequency domain, and obtaining the substance concentration from the extracted portion of the noise spectrum or a change in the substance concentration from the extracted portion of the noise spectrum . The evaluation circuit can be implemented together with the other elements in a single housing , but also arranged separately and communicated using a wired or wireless interface . For example , the evaluation circuit may be a computer program executed on a mobile- , a computer- or a medical device .

[0059] In some aspects , the control circuit is configured to control the at least one light source to emit a light signal for the first time period at a plurality of different consecutive times and obtain the signals from the at least one detecting component . The control circuit can be controlled and triggered by the evaluation circuit .

[0060] Some aspects concern implementations of the detector arrangement . The detector arrangement may comprise a plurality of photodetectors arranged in a ring shape or a quadrature shape and optionally arranged, -particularly centrally- , around the at least one light source , in particular at different distances . In some other aspects , said at least one light source comprises a plurality of optoelectronic devices being arranged with different distances to the at least one detecting component ; and optionally comprising a ring shape or a quadrature shape optionally arranged, -particularly centrally- , around the at least one detecting component .

[0061] It may be suitable if the at least one light source is configured to emit light of different wavelengths . This could compensate for different s kin types and other parameters . In some other aspects , the at least one detecting component comprises a light filter comprising a low transmittance in a frequency spectrum different from a light spectrum emitted by the at least one light source .

[0062] Although the optoelectronic device presented herein is only illustrated with regard to its functionality concerning the determination of substance concentration, one may note that various implementations are possible . In this respect , the evaluation unit does not need to be implemented within the housing itself containing the emitter and the detector , but can be located separately therefrom . In some aspects , the evaluation unit is implemented in a separate device distances from the housing itself . Communication between the emitter and detectors and the evaluation unit as described above is facilitated for example by a wireless communication . This will allow for example to realize a master slave configuration, in which the evaluation unit request measurement to be taken on regular basis . Furthermore , the evaluation unit may be implemented largely in Software , for example as an app executed on a mobile device , whereas the remaining portion of the optoelectronic device are implemented in a separate housing .

[0063] In some aspects , the housing ( or optoelectronic device in more general terms ) is implemented as a ring, earbud, watch or any other wearable , that may fit in some aspects into a user' s usual environment and can be carried continuously . Said ring, earbud, watch or any other wearable is in communicative connection with the evaluation unit or a mobile or any other device implementing the evaluation unit . The ring , earbud, watch or any other wearable may cover a larger portion of the user' s skin, e . g . , wrap around the finger, clipped to- or inserted into the ear, and measuring at one- , opposite- , or multiple sites , and the like ; it is also conceivable to measure on opposite sides of the wrist , such as the dorsal ( top ) and palmar side near the clasp of the wrist strap . They may contain several emitters and detectors at various locations, thus allowing them not only to measure at one spot but at several at the same instant or sequentially. As a result thereof, skin irregularity or other issues can be overcome, and the overall measurement quality may be improved.

[0064] Apart from wearable- and handheld devices, other applications are possible. For example, the optoelectronic device can be implemented in medical devices or laboratory equipment, e.g. for test and measurement purposes. Those devices again can be stationary or mobile.

[0065] Some more aspects concern mobile displays in which the detectors are directly implemented. In such applications, the display LEDs e.g. for the red and green color can be used as emitter in accordance with the proposed principle. A finger is placed directly on the display surface and then illuminated by the display for obtaining the first and / or second signal. Likewise, the proposed principle can be implemented in VR or AR glasses and devices. Another application concerns safety issues, e.g. during certain laborious- or dangerous work, while driving a motor vehicle, and the like. It is possible to implement such optoelectronic devices in accordance with the proposed principle in a car, e.g. on the steering wheel to obtain the perfusion index and from there the noise spectrum during driving. This enables, for example to warn drivers of potential health threats while driving.

[0066] Implementation as a remote PPG sensor is also possible and has been validated: in such systems, the light source at a distance from the specimen is imaged with the help of an optical lens. Likewise, the detector is in a plane conjugated to the specimen in an imaging configuration. The size of the illumination spot is controlled with the source size and the NA can be controlled with an aperture in collimated space, i.e., by the size of the clear aperture of the lens. The same considerations apply to the detection side, where the size the collection spot is defined by the area of the detector. With such a setup, enhancement of sensitivity by negative angles of illumination- and detection may be realized in a trivial manner by illumination and collection light paths crossing over at some distance above the specimen .

[0067] SHORT DESCRIPTION OF THE DRAWINGS

[0068] Further aspects and embodiments in accordance with the proposed principle will become apparent in relation to the various embodiments and examples described in detail in connection with the accompanying drawings in which

[0069] Figure 1 shows a detector arrangement in accordance with some aspects of the proposed principle ;

[0070] Figure 2 illustrates an exemplary PPG measurement that can be obtained by a detector arrangement of Figure 1 ;

[0071] Figure 3 is a reference curve showing the glucose concentration over time as well as the corresponding PI measurement obtained with light in the near infrared spectrum, with two points selected to illustrate some aspects of the proposed principle ;

[0072] Figure 4 illustrates the two PPG signal for point #1 and #30 from Figure 3 ;

[0073] Figure 5 is the corresponding frequency spectrum of the previous two PPG measurements ;

[0074] Figure 6 shows a detailed view of the enhanced spectrum for data point #1 and the fundamental frequency;

[0075] Figure 7 illustrates an exemplary mask generated from the information obtained from the fundamental shown in the previous figure in accordance with some aspects of the proposed principle ;

[0076] Figure 8 shows the application of the generated mas k on the enhanced spectrum of data point #30 , respectively; Figure 9 illustrates the spectrum for point #30 with the mas ked portion corresponding to the fundamental and harmonics separated from the noise in accordance with some aspects of the proposed principle ;

[0077] Figure 10 illustrates the estimated noise under the respective peaks of the fundamental and harmonics separated from the noise in accordance with some aspects of the proposed principle ;

[0078] Figure 11 shows the fundamental and harmonics with the estimated noise subtracted in accordance with some aspects of the proposed principle ;

[0079] Figure 12 shows the spectrum of points #1 and #30 , respectively;

[0080] Figure 13 illustrates the extracted noise spectrum for point #1 and point #30 in accordance with some aspects of the proposed principle ;

[0081] Figures 14 to 21 illustrate several descriptors and their respective correlation of a change in glucose concentration in accordance with several aspects of the proposed principle .

[0082] DETAILED DESCRIPTION

[0083] The following embodiments and examples disclose various aspects and their combinations according to the proposed principle . The embodiments and examples are not always to scale . Likewise , different elements can be displayed enlarged or reduced in size to emphasize individual aspects . It goes without saying that the individual aspects of the embodiments and examples shown in the figures can be combined with each other without further ado , without this contradicting the principle according to the invention . Some aspects show a regular structure or form . It should be noted that in practice slight differences and deviations from the ideal form may occur without , however , contradicting the inventive idea .

[0084] In addition, the individual figures and aspects are not necessarily shown in the correct size , nor do the proportions between individual elements have to be essentially correct . Some aspects are highlighted by showing them enlarged . However , terms such as "above" , "over" , "below" , "under" "larger" , "smaller" and the like are correctly represented with regard to the elements in the figures . So it is possible to deduce such relations between the elements based on the figures .

[0085] Figure 1 illustrates a detector arrangement in accordance with some aspects of the proposed principle . In this embodiment , the detector arrangement is facilitated for glucose measurements in blood vessels of human or animal body . However , it is understood that the present invention and the proposed method is not limited to such measurements . Rather , it is possible for example to replace the human tissue by an artificial pipe to measure a substance concentration in a liquid flowing through the pipe . The various elements of such a detector arrangement are similar to the arrangement presented in Figure 1 . This enables the proposed method for a variety of measurements in different applications .

[0086] The arrangement is used for a variety of measurements including for example PPG measurement or SPO2 measurements . To this extent , one can implement the proposed method as explained herein by software in already existing hardware , provided the detector arrangement is suitable to provide the PPG signal including the noise portion . In the present embodiment , the detector arrangement 1 comprises a housing 11 including at least one light source 12 , said light source 12 configured to emit light pulses of at least one wavelength in the visible or IR spectrum .

[0087] The light source 12 may also comprise a plurality of light sources ( not shown here ) that are configured to emit light of different wavelengths or located at different positions . This enables a compensation due to skin irregularities and also provides an adj ustment to the signal-to- noise ratio , as the penetration of light into the deeper layer strongly depends on the wavelength . For example , typical wavelengths suitable for PPG measurements and measurements for determining the glucose concentration include light in the green and red visible spectrum as well as in the near infrared spectrum. The housing 11 further comprises a detector device 13 that is arranged separated from the light source 12. A light locking element 14 is arranged between the light source 12 and the photodetector. In this example, the photodetector 13 comprises several detector areas 13.1, 13.2 etc. The detector areas are located at different distances from the at least one light source. The various distances between the light source 12 and the detector areas 13.1, 13.2 can compensate skin irregularities, but also provide distance dependent signals. Those signals can also be used to determine the substance concentration, because the overall amount of scattering is dependent from distance.

[0088] In the present invention, in which systematic noise is added due to scattering the emitted light within the skin of a user, a plurality of detector areas 13.1, 13.2 at various distances from the light source 12 enables to determine an optimal distance, in which the ratio of systematic noise (used in the proposed method) to other noise portions is maximal .

[0089] Finally, the detector arrangement 1 also comprises an evaluation and control device 10. The evaluation device 10 can be placed within the hosing 11 of the arrangement or locates separately. For example, the housing 11 can be part of a smartwatch or a wearable device communicating with another device via a wireless interface. The other device contains the evaluation and control device 10. The evaluation and control device 10 is implemented as hardware, software, or a combination of both.

[0090] The evaluation and control device 10 is controlling the measurement and can be thought of as an integrated data acquisition system, which may provide control signals to the current drivers for the at least one light source 12, the analogue to digital converter (s) for sampling of the current from the detector device 13 and the areas 13.1, 13.2, as well as clocks for triggering and timing of measurements.

[0091] During a measurement, a user places its finger with tissue 30 on surface 20, that is part of a glass interface 21 or another transparent interface 21. The interface 21 is transparent for light emitted by the light source, but may be opaque or at least comprises a reduced transparency for another wavelength. This reduces the dark current at the photodetector 13 and its areas 13.1, 13.2.

[0092] The light emitted by the source 11 propagates through tissue 30 along various light paths 32 in tissue 30, where interaction with various blood-carrying and non-blood carrying layers occur. Due to the resulting, periodic blood flow, the heartbeat modulates the fraction of blood within a measurement volume and thus encodes a modulation on the luminous flux re-emitted by the tissue and accessible for detection .

[0093] The measured modulation, corresponding to the heartbeat, contains various information accessible for determination and is generally referred to as PPG measurement. For the purpose of this application, three constituents are explained in greater detail.

[0094] Figure 2 illustrates such PPG signal. The PPG signal comprises a signal level or DC component, about 5.2xl04counts in the example shown, a primary modulation AC of the signal due to the heartbeat, which is commonly of direct interest for various applications, and (3) a noise component, which is most easily recognized in the example shown by the profound point-to-point variability. The heartbeat is given by the maxima in the signal and lies in the range of about 60 beats / min (1 Hz) . The perfusion index or PI is defined by the ratio AC over DC, that is, the amplitude of the AC portion of appr O.lxlO4counts divided by the DC portion. In the given example, the PI is approximately 1.9.

[0095] The measurement for a PPG measurement as well as for the proposed method are usually acquired over several heartbeats, e.g., for 5 to 20 seconds, and with acquisition rates substantially greater than the heart rate. As the proposed method requires a profound knowledge about the noise imposed on the actual signal, it is suitable to acquire the PPG measurement with a rate that is at least 20 times higher than the actual heartbeat. As the heartbeat is in itself subject to variation between app 60 beats / min to appr . 180 beats / min, it is suitable to use data acquisition rate between 50 Hz and 200 Hz. In the example depicted in Figure 2 , an acquisition rate of 100 Hz with 1024 data points per trace was used .

[0096] In the following, a method is illustrated to decompose one or more PPG measurements into DC , AC , and noise components and utilize the noise component for determination of the glucose concentration . The use of the AC and DC components for glucose determination has been shown and is subj ect to other application filed by the applicant . While several different methods are possible to extract noise from the signal , both in the time and frequency domain, the proposed method applies an approach in the frequency domain .

[0097] Figure 3 shows a typical Oral Glucose Tolerance Test is shown, where the reference data referred to as Reference is taken from an Abott Freestyle Libre 3 system . It constitutes the reference measurement during the period of the test . Food uptake takes place at about 10 : 20 , indicated by the blue vertical bar and the absence of data points . Shortly afterward, both optical and reference measurements are performed each minute . The optical measurements are taken for 10 seconds , which are long enough to obtain a suitable PPG measurement but short enough that the change in glucose concentration during a single measurement is not overly large .

[0098] Following ingestion of sugar containing food, the Glucose concentration increases with time , peaking about 40 minutes after eating, and decaying over the next 40 minutes . The optical measurement in Fig . 3 ( referred to by IR PI ) is the perfusion index for IR light at 940 nm . The response of the PI to blood Glucose concentrations is subj ect to a different application already filed by the applicant .

[0099] For the purpose of extracting noise , PPG traces from indicated points #1 and #30 will be used in the exemplary embodiment . Of course , the proposed method can be performed with each data point , that is each PPG measurement . The exemplary points correspond to low and high blood

[0100] Glucose concentrations , respectively . Fig . 3 shows the actual PPG traces of data points #1 and #30 obtained by detecting the light scattered by the human tissue . The measured signals are pre-processed by using a low pass filter to remove slow transients , e . g . caused by breathing . The transient removal is suitable to suppress end-point discontinuities , which typically would increase the 1 / f contribution ( s ) in the frequency domain . Likewise , the dark current is at least partially removed . As visible from the two traces , the increasing blood sugar concentrations cause an increase in the AC portion as well as a drop in the DC portion . Hence , point #1 with the low glucose concentration comprises a higher DC component of appr . 6xl04and lower AC portion, that is also quite noisy . The trace for point #30 comprises a lower DC value of appr . 4 . 8 xlO4with a higher AC portion . The noise may be similar to the other, at least from the noise visible to the eye .

[0101] After pre-processing the signals , a Fourier Transformation is performed and normalized to their respective DC value ( zero-frequency) to provide a relative comparison of the modulation amplitudes . As visible from Figure 5 , the fundamental frequency from the heartbeat at appr . 1 . 3 Hz is clearly visible in both traces . However , the amplitude for point #1 is reduced, and the harmonics are in contrast to the trace of point #30 almost not visible . The amplitude of the fundamental frequency at about 1 . 2 Hz for point #30 is about 3x greater than for point #1 corresponding to a lower glucose concentration .

[0102] As indicated in Figure 5 , the noise floor discernible in between the harmonics is also higher when the Glucose concentration is increased . This is observation is new, and it can provide additional means for detecting blood Glucose non-invasively by optical measurement and determine a change and even it absolute value when properly calibrated .

[0103] The following Figures describe an approach to extract the noise level from the spectrum and correlate it with the glucose level to determine the concentration from the noise and more particularly the change in noise in the PPG measurement over time . In a first step , some further pre-processing is performed to enhance the peak of the fundamental frequency as well as the harmonics over the noise . The enhancement is achieved by taking the absolute of the FFT to the power of M, where M can be an integer such as 2 , 3 , 4 and so forth . The enhancement is necessary when the actual signal from the heartbeat is rather small or the noise level is large , such that harmonics are only marginally protruding over the adj acent noise . Furthermore , a smoothing function is applied, in the exemplary embodiment a 3-point smoothing function, in which 3 points of the PPG measurement are moving averaged . The result of this addition preprocessing is visible for the fundamental of point #1 in Figure 6 with M=6 and a 3-point average .

[0104] The amplitude of the peak is significantly larger than the noise level . In a first aspect , one determined the frequencies fmlnand fmax, in which the amplitude reaches certain level , for example in the present case 5% of the overall maximum. Other peak values can be chosen as well and may depend on the enhancement factor . The level detection determines a dynamic peak width as fmax- fmin; to arrive at the actual peak width a static offset may be added to the dynamic width, but the advantage is that the offset needs to be determined only once for all measurements . Further, the fundamental peak is not symmetric , hence the center of gravity has to be determined in a next step using the dynamic width, but shifted by an offset given between the difference of the middle frequency ( fmax- fmin) / 2 and the center of gravity .

[0105] This can be done by summing the product of frequency and FFT amplitude over the previously determined dynamic width of the fundamental frequency . The result is then divided by the sum of the amplitudes . In the example depicted in Figures 6 and 7 the center pixel is at a location 13 . 93 ( arbitrary value ) , which corresponds roughly to frequency 1 . 27 Hz . Even though the center of gravity approximately coincides with the maximum in this example , it does not have to , and that it depends on the distribution of amplitudes . The final mas k for the peak boundaries of the fundamental is shown on Figure 7 as applied to the original FFT of point #1 with a base level of peak of appr . 5 % The mask, which has been determined both in width as well as its center frequency, is then applied to the spectrum . More particularly, one may adj ust the overall number of mas ks to be applied to the spectrum, usually between 3 and 10 . The reason for the limited number of masks being used to mask the harmonics and the fundamental lies in the fact that the harmonic decay quite fast and will soon be completely buried in the noise . Hence , it is not necessary to ask for more than 10 harmonics most of the time , as there will not be any further contribution .

[0106] For the process , the mas k is placed with the respective width of fundamental peak as obtained in the previous step and with a center frequency corresponding to the multiple of the fundamental . If needed, an additional offset can be applied . The process is repeated for all higher harmonics up to the maximum number of harmonics that shall be resolved and analyzed .

[0107] Figure 8 illustrates the mask obtained for the fundamental of point #30 applied up to the ninth harmonics . It is noted that the mask width as well as its center of gravity is determined for each measured point over time separately . This is necessary as heart rate may change during the overall time .

[0108] Now the mas k is multiplied with the spectrum of point #30 thereby separating the fundamental and the peaks (plus the noise portion beneath the peaks ) from the noise between the various mas ks . The result is presented in Figure 9 with Curve KI being the peaks of the fundamental and the harmonics up to the ninth harmonics . Curve K2 provides the noise portion of the spectrum between the mas ks .

[0109] Figure 10 shows the next steps of estimating the noise beneath the masked peak of the fundamental and the harmonics . For this purpose it is assumed that the noise is frequency independent and constant over the mas ked range , that is , between fmlnand fmaxfor each mas k . A slight frequency dependence can also be assumed, as the noise generally decays with increasing frequencies . In order to estimate the noise under the peaks , the gaps in the noise spectrum caused by the mas ks are filled with an average value of the adj acent points , corresponding to the limits of the mas ks on the lower and upper frequencies . The result is illustrated in Figure 10 with the noise level in between the peaks of the fundamental and harmonics removed for viewing purposes .

[0110] The frequency spectrum comprises complex numbers , hence , as can take an additional step by fitting on the phase of the noise spectrum in the peak regions . The obj ective is to minimize the magnitude of the difference between the complex data amplitude and the complex noise amplitude for each frequency position, using the magnitude of the noise amplitude determined above without alteration . This phase fitting step is an approximation that leads to a conservative estimate by assuming that noise will always reduce the modulation amplitude . However , statistically, noise could also artificially enhance the modulation, but since no knowledge of the actual phase difference between modulation and noise is available , it has been observed that the conservative approach provides satisfying results .

[0111] The advantage of the proposed methodology becomes apparent when comparing the two curves KI and K2 of Figure 9 . There is no need for knowledge of the number of harmonics present or discernible in the frequency spectrum, because higher and weaker harmonics become simply indistinguishable from noise and hence will not contribute to the modulation spectrum. This allows as stated previously to specify how many harmonics should be considered for analysis with the knowledge that only those harmonics will contribute that protrude above the noise floor . An otherwise challenging problem is thereby defused .

[0112] Figure 11 illustrates the peak level of the fundamental and the harmonics with the estimated noise level subtracted . In this example , only the first two harmonics add significantly to the modulation spectrum, which is consistent with the above methodology and its advantages . Figure 12 shows the overall spectrum of points #1 and #30 , respectively . The Figure 13 is the corresponding noise spectrum for both points , with the estimated noise filling the gaps in the removed peaks . Hence , the AC portion is separated in the spectrum of Figure 13 , leaving only noise and DC ( at frequency 0 )

[0113] The spectrum is obtained for each of the PPG measurements and then subsequently processed . Several descriptors are available to process the noise spectrum, as well as a combination of one of the other portions with the noise spectrum. All these descriptors show a correlation with the glucose concentration, hence they are affected by a substance concentration or a change thereof . One useful description is the standard deviation o given by wrth DC be ng the DC portion and N being

[0114] Another description is given by the absolute signal modulation ( that is without the noise beneath as illustrated in the spectrum of Figure 11 . The modulation AC is given by :

[0115] The perfusion index PI is already known and given by the ratio of AC / DC .

[0116] It has been observed that without any effect of the substance concentration on the noise , the system would resemble a shot-noise limited system . However, it has been observed that a deviation occurs , which in itself also correlates to the glucose or more general substance concentration . A shot-noise limited system is given by sigma o equal to the square root of the DC portion . Hence , the signal-to-noise ratio SNR can be expressed by

[0117] Finally, the modulation to noise ratio MNR is given by

[0118] The following figures 14 to 21 illustrate the correlation of the various descriptions with the glucose concentration or the change thereof .

[0119] The obtained data are not smoothed of averages and may thus appear noisy . However , the correlation is clearly visible and a further averaging of the various points on the PPG measurement , i . e . a moving average of 3 to 5 individual measurement points would greatly smooth the curve . Further, the reference curve also shown in the following figure has been shifted (with the original curve shown for reference ) . It seems that the optical measurement is more sensitive . Hence , it is affected approximately 5 min to 10 min earlier than the reference curve . This can be explained by the delay until glucose increases in the interstitial fluids , but also by other causes , e . g . change of the concentration in blood occurring earlier than in the interstitial fluid or other changing parameters in the tissue .

[0120] For improved readability and possible reference with already existing correlation, Figure 14 illustrates the perfusion index PI for light of 940 nm together with a reference measurement . The references curve is also used for the following figures . The perfusion index PI is calculated using the above ratio from the pure AC spectrum as shown in figure 11 for all measurement during the oral glucose tolerance test (OGTT ) . For changes of the reference measurement between 82 and 155 mg / dL, which corresponds to an ~89% rise of the peak over the minimum, the PI scales from ~ . 0065 to ~ . O31 , which is a rise of almost 3 . 8x over the same period . This proves the sensitivity of the approach, comprising PPG measurement and the analysis algorithm presented here .

[0121] Figure 15 presents the DC-level ( 1 / DC ) during the same OGTT . As discussed already in a previous yet unpublished application by this applicant , the DC level drops in a reflective PPG configuration when the glucose concentration increases , and vice versa . The peak rises some 20% over the value during normal glucose levels observed at the beginning and at the end of the OGTT .

[0122] Referring now to Fig . 16 . The figure illustrates the standard deviation o as calculated above . Apart from the inherent variability, there is a strong correlation with the glucose reference measurement . This observation was not expected, because the DC level drops with increasing glucose . The decrease in DC means that o is expected to drop as well . While this holds true for a shot-noise limited system, it appears that with increasing glucose concentration, the behavior deviates from shot noise limitation .

[0123] The behavior becomes particularly apparent in view of Figure 17 illustrating the standard deviation o plotted together with the square root of the DC portion and DC portion itself . As expected, the square root of DC drops as the DC drops , but the standard deviation o increases with the glucose concentration and then decreases again . One can conclude that systematic noise is increasing when the glucose level rises . If this finding is corroborated as the understanding of lighttissue interactions improves , it can provide a further insight for PPG measurement and especially for optical , non-invasive blood Glucose measurement .

[0124] If the standard deviation o increases while the DC portion decreases , then SNR-1= o / DC should also correlate to blood Glucose concentration and a change thereof . This is shown in Fig . 18 . In other words , the noise-to-signal ratio , or the inverse of the signal-to-noise as defined herein can be a measure for a substance concentration in accordance with the proposed principle .

[0125] Figure 19 shows the behavior of o2 / DC as a measure of how closely the data adheres to shot-noise limited performance . For a pure shot noise limited system, the noise level should result in a value of 1 , said value equal to shot-noise , or sqrt ( DC ) . However , even for low glucose concentrations , e . g . at the beginning of the OGTT , the overall system noise is above the shot-noise limit at a value of proximately 1 . 25 . The intake and increase of the glucose concentration also pronounces a further deviation from shot-noise performance , averaging at about 2 . 25 at the maximum level of the glucose concentration .

[0126] Figure 20 shows another description, namely the absolute modulation AC over the standard deviation o . The descriptor is also referred to as the modulation to noise ratio (MNR) . The descriptor may act as a measure of quantity and an indicator for the performance of the proposed approach with regard to standard vital sign monitoring applications , such as heart rate monitoring or SPO2 measurement . The modulation to noise ratio quantifies the preponderance of modulation due to the heartbeat over the noise in the system. The higher the excess modulation, the easier it will be to perform the measurement and extract the desired piece ( s ) of information for some given application .

[0127] Finally, Figure 21 illustrates the heart rate during an oral glucose tolerance test by extraction of the pure modulation spectrum shown earlier in Figure 11 and summation of the various harmonics using this algorithm. Overall , twelve channels for measurement in the system are available , and the results are overlaid . Although there is some significant point-to-point variability, the heart rate at rest slightly increases during the oral glucose tolerance test , indicating a correlation with the glucose concentration . A similar observation is made with the sPO2 measurement . It also correlates with the glucose concentration .

[0128] Overall , the observation that noise is affected by a substance concentration may support other applications and measurements , including SPO2 measurement . The proposed method is in any case a useful supplement to further enhance non-invasive blood Glucose measurement . One of the benefits for this method is that it can be easily added and supplement already existing measurements and other application currently based on PPG measurements .

[0129] The method extracts the noise from a PPG measurements and utilizes it to determine a substance concentration in a liquid . It is based on the fact that the underlying system of liquid, particles and substance as well as light source and the optical detector is deviating from a shot noise limited system . The deviation is caused by the introduction of systematic noise introduced into the system.

[0130] Although a comprehensive theory that would explain the detailed cause and origin of the observation may currently not exist , it is reasonable to assume that the liquid becomes more transparent due to the decreasing difference in the refractive index . This is evidenced by the decreasing DC level during the same period when using a configuration in reflection mode . Consequently, light can be detected from deeper layers , which has interacted more with the tissue and accumulated more noise . Such additional accumulated noise may originate from various physiological mechanisms , such blood flow, nerve and muscle activity . Such a hypothesis is corroborated by the fact that the PPG modulation amplitude (AC, PI ) increases substantially when the glucose levels increase . This increase is explainable by the portion of the signal modulated by the pulsating blood is gaining in weight .

[0131] LIST OF REFERENCES

[0132] 1 detector arrangement

[0133] 10 control and evaluation circuit 11 housing

[0134] 12 light source , LED

[0135] 13 detector

[0136] 13 . 1 13 . 2 detector areas

[0137] 14 optical barrier 21 glass interface

[0138] 20 surface

[0139] 32 light path

[0140] 30 tissue

[0141] 32 light path

Claims

CLAIMS1 . Method for determining a substance concentration in a sample comprising liquid containing particles , in particular glucose in blood, wherein a refractive index of the liquid is dependent on a concentration of the substance dissolved therein and a density of particles in the liquid is substantially constant , wherein the liquid is modulated in its volume or pressure comprising the steps of :Obtaining a first signal during a first time period with an acquisition rate that is at least two times larger than a periodicity of the volume or pressure modulation of the liquid; Extracting at last a portion of a noise spectrum from the first obtained signal particulary in the frequency domain;Obtaining one of : o the substance concentration from the extracted portion of the noise spectrum and a reference value ; and o a change in the substance concentration from the extracted portion of the noise spectrum and a reference value .2 . The method according to claim 1 , further comprising :After obtaining a first signal , obtaining a second signal during a second time period with an acquisition rate that is at least two times larger than a periodicity of the volume or pressure modulation of the liquid;Extracting at last a portion of a noise spectrum from the second obtained signal in particular the frequency domain; wherein optionally, the reference value is based on the extracted portion of the noise spectrum of the second obtained signal .3 . The method according to any of the preceding claims , wherein the reference value is given by a pre-determined extracted portion, or a value derived by such extracted portion .4 . The method according to any of the preceding claims , wherein the step of extracting at least a portion of a noise spectrum comprises the step of :Identifying AC portions in the spectrum of the obtained signal , said AC portions corresponding to the fundamental frequencies and its harmonics of the liquid' s volume or pressure modulation; Removing the identified AC portion from the spectrum to obtain a noise portion .5 . The method according to claim 4 , wherein the step of identifying the AC portions in the spectrum of the first and second obtained signal comprises at least one of the steps of :Applying a smoothing function onto the spectrum of the first and second obtained signal ;Enhancing at least the peak of the fundamental frequencies , in particular by taking the absolute values of the spectrum to the power of M with M being an integer larger than 2 ;Determining a mas k with a width based on a pre-determined amplitude level of at least the peak of the fundamental frequencies ;Determining a center of at least the peak of the fundamental frequencies , optionally by using the width based on a predetermined amplitude level of at least the peak of the fundamental frequencies ;Applying a mask, in particular the determined mask over a predetermined number of peaks corresponding to the fundamental frequencies and its harmonics of the liquid' s volume or pressure modulation, thereby identifying the AC portion .6 . The method according to claim 4 , wherein the step of removing the identified AC portion from the spectrum comprises the step of : extracting the noise portion within the spectrum at the fundamental frequencies and its harmonics of the liquid' s volume or pressure modulation based on the values of adj acent points in the spectrum; or extracting the noise portion of the masked portions of the spectrum based on points in the spectrum at the masked portion' s borders .7 . The method according to claim 4 , wherein said AC portions correspond to the fundamental frequencies and harmonics up to the seventh order and in particular up to the fifth order and in particular up to the fourth order .8 . The method according to any of the preceding claims , wherein the step of obtaining the first and / or second signal comprises at least one of :Low-pass filtering the first and / or second obtained signal to remove transients , in particular with a frequency below 2Hz and in particular below 1 Hz and in particular below 0 , 5 Hz ;Removing dark counts from the first and / or second obtained signal ; andConverting the first and / or second obtained signal into a digital signal .9 . The method according to any of the preceding claims , wherein the step of extracting at least a portion of a noise spectrum from the first and / or second comprises the step of :Conducting a Fourier transformation or a fast Fourier transformation upon the first and / or second obtained signal .10 . The method according to any of the preceding claims , wherein the step of obtaining the substance concentration or the change of substance concentration comprises at least one of :Obtaining the standard deviation of the extracted portion of the noise spectrum and the reference value ;Obtaining a signal-to-noise ratio or the inverse thereof of the extracted portion of the noise spectrum and the reference value ;11 . The method according to any of the preceding claims , wherein the step of obtaining the substance concentration or the change of substance concentration comprises a quantification of deviation from a shot-noise limited system defined by the ratio of the standard deviation of the estimated portion of the noise spectrum.12 . The method according to any of the preceding claims , further comprising :Calculating a modulation to noise ratio given by the AC portion over the estimated noise , thereby quantifying an excess modulation of the signal due to the heartbeat over the system noise .13 . The method according to any of the preceding claims , wherein the liquid is blood and the modulation of the blood' s volume and / or pressure is defined by the heartbeat .14 . The method according to claim 13 , further comprising :Calculating the heart rate from the identified AC portions in the spectrum; and / orObtaining a blood glucose correlation from a change of the heart rate .15 . Detector arrangement for determining a substance concentration in a sample comprising liquid containing particles , in particular glucose in blood, wherein a refractive index of the liquid is dependent on a concentration of the substance dissolved therein and a density of particles in the liquid is substantially constant , said detector arrangement comprising : at least one light source and at least one detecting component , wherein the at least one detecting component is optically separated from the at least one light source ; wherein said at least one light source is configured to emit light through an exit window onto a sample ; and said detecting component is configured to detect a light component corresponding to emitted light scattered through; a control circuit coupled to the at least one light source and at least one detecting component ; wherein the detector arrangement is configured to perform the method according to one of the preceding claims .16 . Detector arrangement according to claim 15 , wherein the control circuit is configured to control the at least one light source toemit at light signal for the first time period at a plurality of different consecutive times and obtain the signals from the at least one detecting component .17 . Detector arrangement according to claim 15 or 16 , wherein said detector arrangement comprises a plurality of photodetectors arranged in a ring shape or a quadrature shape and optionally arranged, -particularly centrally- , around the at least one light source , in particular at different distances .18 . Detector device according to one of claims 15 to 17 , wherein said at least one light source comprises a plurality of optoelectronic devices being arranged with different distances to the at least one detecting component ; and optionally comprising a ring shape or a quadrature shape optionally arranged, -particularly centrally- , around the at least one detecting component .19 . Detector device according to any of claims 15 to 18 , wherein the at least one light source is configured to emit light of different wavelength20 . Detector device according to any of claims 15 to 19 , wherein the at least one detecting component comprises a light filter comprising a low transmittance in a frequency spectrum different from a light spectrum emitted by the at least one light source .

Images