Device and method for the continuous and non-invasive determination of physiological parameters of a test subject

Abstract

The present invention relates to a device for the non-invasive determination of physiological parameters of a test person with a lighting unit (32), having a plurality of LED types whose emission maxima are at different wavelengths from the visible to the NIR wavelength range and include an emission maximum below 590 nm, a photo sensor for application to the skin, a data processing unit (11) for reading out the photo sensor and for controlling the illumination unit (32), so that the different types of LEDs individually in a predetermined activation sequence at successive activation start times t.sub.k (k=1, 2 . . . M) are activated for a respective predetermined activation period, and to repeat the activation sequence with a clock frequency as a result n=1, 2 . . . N, wherein the clock frequency—sufficiently high for the resolution of the pulse—is characterized in that the photo sensor is a camera sensor (34) on CCD or CMOS base, which is arranged to the illumination unit (32) in a way that it can detect by transflection light passing through the body, that the data processing unit (11) reads out the camera sensor in each activation sequence at the activation start times t.sub.k (k=1, 2 . . . M) and over the respective activation period, with the detected intensities via subregions of sensor elements being added up and assigned to the respective activation sequence and to the respective activation start time t.sub.k (k=1, 2 . . . M), and records them as time series, wherein subareas of the camera sensor are added up by reading out the lines of the camera sensor, which are parallel to the connection axis between the illumination unit and the camera sensor and along which increases the distance from the illumination unit, and parallel lines are summarized to a single averaged line, and the data processing device is arranged to evaluate the averaged line as a function of the distance from the illumination unit to the muscle oximetry.

Claims

1. Device for the continuous and non-invasive determination of physiological parameters of a test person during exercise comprising an LED-based illumination unit, designed to be able to rest on the skin of the test person at a desired measuring point, with a plurality of different, juxtaposed LED types, whose emission maxima at different wavelengths λ1, λ2 . . . λL lie from visible to NIR-wavelength range, wherein the lighting unit comprises at least one LED type with an emission maximum below 590 nm, a photo sensor, which is designed to rest on the skin of the test person in order to capture light emitted by the illumination unit and passing through the body of the person to an exit location in the area of the photo sensor, a data processing unit, which is connected to the photo sensor to read it out, and which is connected to the lighting unit and is adapted to operate the different LED types individually in a predetermined activation sequence at successive activation start times t.sub.k (k=1, 2 . . . M) for a respective predetermined activation period to activate and repeat the activation sequence with a clock frequency as a result n=1, 2 . . . N of activation sequences, the clock frequency is sufficiently high to resolve the pulse of the test person's circulation, wherein a camera sensor, based on CCD or CMOS with a two-dimensional array with rows and columns of sensor elements is used as the photo sensor, the camera sensor is arranged relative to the lighting unit so that it can rest on the skin on the same side of the body part as the lighting unit and adjacent to it to detect light passing by transflection through the body of the test person to the camera sensor, the data processing unit is set up to read out the camera sensor in each activation sequence at the activation start times t.sub.k (k=1, 2 . . . M) and over the respective activation period, and to record the detected intensities of the sensor elements, combined over subregions of sensor elements and the respective activation sequence from the sequence n=1, 2 . . . N of activation sequences and the respective activation start time t.sub.k (k=1, 2 . . . M) assigned as time series, wherein sub-areas of the camera sensor are summarized by reading out lines of the camera sensor, which lie parallel to the connection axis between the illumination unit and the camera sensor and along which the distance from the illumination unit increases, and parallel rows are combined into a single averaged row, and the data processing means is arranged to evaluate the averaged row as a function of the distance from the illumination to muscle oximetry.

2. Device according to claim 1, wherein the data processing unit is adapted to read out each line of the camera sensor, which run parallel to the connecting line between the illumination unit and the camera sensor and (for summarizing the sub-areas of sensor elements) to average all lines to an average line intensity ${\stackrel{\_}{I}\left(i\right)}_{\lambda m}\left(n,t_k\right)=\frac{1}{N_Z}\underset{a=0}{\overset{N_Z}{.Math.}}{I_a\left(i\right)}_{\lambda m}\left(n,t_k\right)$ where i=1,2 . . . N.sub.Sp is the continuous column index and N.sub.Sp is the number of columns of the camera sensor and the index λm symbolizes the wavelength of the emission maximum of the respective LED type and as averaged intensity Ī(i).sub.λm(n, t.sub.k) of the respective activation sequence n from the sequence n=1, 2 . . . N of activation sequences and the respective activation start time t.sub.k (k=1, 2 . . . M) are recorded as time series.

3. Device according to claim 1, wherein the data processing unit is set up to add up the intensities of the sensor elements over all lines and columns to a via the camera sensor integrated camera sensor intensity ${\stackrel{\_}{I}}_{\lambda m}=\frac{1}{N_Z.Math.N_{\mathrm{Sp}}}\underset{a=0}{\overset{N_Z}{.Math.}}\underset{i=0}{\overset{N_{\mathrm{Sp}}}{.Math.}}{I_a\left(i\right)}_{\lambda m}$ where i=1, 2 . . . N.sub.Sp symbolizes the continuous column index, N.sub.Sp the number of columns of the camera sensor, a the continuous line index, N.sub.Z the number of lines of the camera sensor and the index λm the wavelength of the emission maximum of the respective LED type, and to record these as time series Ī.sub.λm(n, t.sub.k) of the camera sensor intensity, assigned to the respective activation sequence n from the sequence n=1, 2 . . . N of activation sequences and the respective activation start time t.sub.k (k=1, 2 . . . M).

4. Device according to claim 3, wherein the data processing unit is adapted to add up the time series Ī.sub.λm(n, t.sub.k) of the camera sensor intensity of the different wavelengths λ1, λ2 . . . λL to a wavelength-spanning individual time series Ī(n, t.sub.k) of the camera sensor intensity and to record it.

5. Device according to claim 1, wherein the illumination unit comprises at least one LED type with emission maximum in the range of 500 nm to 540 nm and an LED type with emission maximum in the range of 570 nm to 585 nm.

6. Device according to claim 5, wherein the illumination unit in the NIR wavelength range comprises at least three different LED types whose emission maxima span an NIR wavelength range from 600 nm to 1100 nm.

7. Device according to claim 6, wherein the illumination unit in the spanned NIR wavelength range comprises at least four different LED types with emission maxima distributed in the spanned NIR wavelength range.

8. Device according to claim 1, wherein the illumination unit in the NIR wavelength range has a plurality of LED types with emission maxima distributed in the wavelength range 800-1100 nm, including an LED type with emission maximum at 960 nm, an LED type with emission maximum at 930 nm, and at least two LED types with emission maxima in the range of 790-920 nm, and that the data processing unit is adapted to determine from the absorption measured at 960 nm the concentration of water, taking into account the absorption by fat at 930 nm and the absorption by hemoglobin by evaluating a plurality of absorptions in the range 790-920 nm.

9. Device according to claim 4, wherein the data processing unit is adapted to subject the time series of the wavelength-overlapping camera sensor intensity Ī(n, t.sub.k) of a Fourier transform and to search—for detection of the pulse signal—a distribution peak at a fundamental frequency, which is accompanied by one or more distribution peaks of harmonics at integer multiples of the fundamental frequency.

10. Device according to claim 9, that wherein the data processing unit is adapted to confirm the correctness of detection after detection of a pulse signal, if in the signal of the camera sensor intensity Ī.sub.λm(n, t.sub.k) with λm from the wavelength range of 570-585 nm a stronger pulsating signal with the fundamental frequency can be found than in the signal of the camera sensor intensity Ī.sub.λm(n, t.sub.k) with λm from the wavelength range of 500-540 nm.

11. Device according to claim 9, wherein the data processing unit is adapted to apply in the Fourier spectrum of the wavelength-overlapping camera sensor intensity Ī(n, t.sub.k) a dynamic band pass filter for the pulse signal detection, wherein the centre of the accepted frequency band is at the detected fundamental frequency and the width of the frequency band is predetermined, wherein the dynamic band pass filter follows the changes of the detected fundamental frequency of the pulse signal if the changed fundamental frequency is within the accepted frequency band, and wherein a hypothetically newly detected pulse signal with its fundamental frequency is discarded if it is outside the accepted frequency band.

12. Device according to claim 9, wherein the data processing unit is set up to assign in the Fourier spectrum of the wavelength-overlapping camera sensor intensity Ī(n, t.sub.k) a distribution peak with a fundamental frequency, which does not have accompanying harmonics with distribution peaks at integer multiples of the fundamental frequency and which is smaller than the frequency of the pulse signal, to the respiration and to determine therefrom the respiration rate.

13. Device according to claim 12, wherein the data processing unit is adapted to apply a dynamic band pass filter in the Fourier spectrum for respiratory signal detection, wherein the center of the frequency band is at the detected respiration rate and the width of the frequency band is predetermined, wherein the dynamic band pass filter goes along with changes of the detected respiration rate if the changed respiration rate is within the accepted frequency band and wherein a hypothetically newly detected respiratory signal with its new respiration rate is discarded if it is outside the accepted frequency band.

14. Device according to claim 3, wherein the data processing unit is set up to evaluate the time series Ī(i).sub.λm (n, t.sub.k) of the camera sensor intensity for emission maxima λm in the NIR wavelength range in relation to each other and in their dependencies on the distance from the illumination unit to determine the degree of oxygenation of the muscle.

15. Method for the continuous and non-invasive determination of physiological parameters of a test person during exercise, in which An LED-based illumination unit, having a plurality of different, juxtaposed LED types whose emission maxima at different wavelengths λ1, λ2 . . . λL are from the visible to the NIR wavelength range, is brought in contact with the skin of the test person to a desired measuring point, A photo sensor is placed in contact with the skin of the test person in proximity of the illumination unit in order to detect light, emitted by the illumination unit and passing through the body of the person to an exit location in the area of the photo sensor, A data processing unit, which is connected to the reading unit of the photo sensor for reading it out and which is in communication with the lighting unit, which activates each of the different LED types individually in a predetermined activation sequence at successive activation start times t.sub.k (k=1, 2 . . . M) for a respective predetermined activation period and repeats as a result the activation sequence at a clock frequency n=1, 2 . . . N of activation sequences, wherein the clock frequency is sufficiently high to dissolve the pulse of the test person's blood circulation, wherein in each activation sequence at least one LED type is activated with an emission maximum below 590 nm, wherein A camera sensor based on CCD or CMOS with a two-dimensional array of sensor elements is used as the photo sensor, The camera sensor is positioned towards the illumination unit so as to rest on the skin on the same side of the body portion as the illumination unit and adjacent thereto to capture the light which gets by transflection through the body of the test person to the camera sensor, The data processing unit reads out the camera sensor in each activation sequence at the activation start times t.sub.k (k=1, 2 . . . M) and over the respective activation period and combines the detected intensities of the sensor elements via subregions of sensor elements and assigns the respective activation sequence from the sequence n=1, 2 . . . M of activation sequences and the respective activation start time t.sub.k (k=1, 2 . . . M) and records them as a time series, wherein subareas of the camera sensor are combined by reading lines of the camera sensor which are parallel to the connection axis between the illumination unit and the camera sensor, and along which the distance from the illumination unit increases, and parallel lines are combined into a single averaged row, and the data processing device is arranged to evaluate the averaged line as a function of the distance from the illumination unit to the muscle oximetry.

Description

(1) The invention is explained in more detail below with reference to exemplary embodiments in the drawings, in which

(2) FIG. 1 is a schematic perspective view of the device according to the invention from a point of view within the observed body region,

(3) FIG. 2 is a block diagram showing the structure and operation of the apparatus,

(4) FIG. 3 shows a schematic sectional view through a device according to the invention and through the tissue underlying the device of the examined body site of the test person,

(5) FIG. 4 shows a plan view of a preferred compact, multi-spectral LED module with seven individually controllable LED semiconductors and a corresponding position matrix with the emission maxima of the LED types used,

(6) FIG. 5 a) shows an integrated module, in which the optical unit with the electronic unit is integrated in a housing and arranged and attached as a whole unit on the skin at the observable body part of the test person,

(7) FIG. 5 b) shows an embodiment of the device with an optical sensor module which is mounted directly on the skin and connected to a digital data processing unit via a cable to the optical sensor module and which can be placed at another location,

(8) FIG. 6 shows VIS/NIR absorption spectra of the muscle-relevant substances for spectroscopic assessment, namely oxygenated hemoglobin (O.sub.2Hb), deoxygenated hemoglobin (HHb), water (H.sub.2O), melanin, fat, and tissue scattering with significant wavelengths marked, which are well suited for a data recording,

(9) FIG. 7 shows schematically constant and pulsatile fractions of body components which contribute to the absorption,

(10) FIG. 8 shows a schematic plan view of a lighting unit and a camera sensor as well as read-out intensity signals and their summary,

(11) FIG. 9 shows the wavelength-overlapping camera sensor intensity signal Ī(τ) as a function of time as well as the components contributing thereto, resulting from respiration, pulse, and movement of the test person,

(12) FIG. 10 shows the Fourier spectrum of the wavelength-overlapping camera sensor intensity signal with distribution peaks contained therein, which originate from the pulse, respiration, and movement of the test person,

(13) FIG. 1 shows a schematic diagram of the structure of a device according to the invention. The body surface on which the device is mounted is stated at 31. Below this, the relevant measuring medium or tissue 33 is located, which may be, for example, the muscle that is mainly stressed during training, but in principle also any other body location, such as the breast, shoulder, or forehead. The use on the wrist can also be interesting, whereby the integration of the measuring device into a so-called “wearable” (for example fitness tracker or smart watch) is also conceivable.

(14) The multispectral illumination unit 32 with a matrix of LEDs sequentially emits light pulses with emission maxima at different wavelengths into the tissue in a defined time sequence (activation sequence), wherein the incident light is immediately scattered there after entry and homogenized in the propagation direction. As it passes through the tissue, the light is also absorbed by the substances, contained in the tissue, with fixed known extinction coefficients.

(15) The light comes out at different points of the tissue again. The area in which the CMOS optical camera sensor 34 is placed is important. There, the light is measured as a function of the distance from the point of irradiation and recorded as an analogue signal. By this arrangement, in which the illumination unit 32 and the camera sensor 34 lie in one level, the light is measured in the so-called transflection, since the light partly passes through the tissue (transmission), but also partly is diffusely scattered or reflected (reflection).

(16) This signal is amplified directly in the camera sensor 34, digitized and transmitted in parallel or serially for each pixel to a processor unit.

(17) The camera sensor 34 is preferably a high-resolution, two-dimensional CMOS digital camera sensor array. The array must be a so-called monochrome sensor, that is a sensor without a structured colour filter matrix. Thus, the sensor can receive light radiation of different wavelengths, depending on its specific sensitivity curve.

(18) FIG. 2 shows a block diagram of a preferred embodiment of the device. In the device, a data processing unit (arithmetic unit) 11 takes over the entire control, data synchronization, data readout, and buffering. Firstly, the data processing unit assumes the parameterization of the camera sensor 34 (parameterization here means the setting of the registers of the camera sensor such as gain, skipping, binning, exposure time, ROI, pixel rate, etc.), the digital data acquisition of the camera sensor 34, and the control of the multispectral illumination unit 32. Furthermore, the data processing unit can optionally also perform a data reduction, data buffering or data preprocessing.

(19) The multispectral illumination unit 32 emits light 36 onto the measurement area 35 of the muscle. This light exits the muscle in a distance-dependent manner as analysis light 37 as described above and strikes the CMOS camera sensor 34. The camera sensor 34 is a two-dimensional CMOS digital camera sensor. In one direction (connecting axis of LED illumination unit 32 and camera sensor 34—this direction is referred to in the present application as the row direction), the distance dependence of the exiting light is detected due to the growing distances to the entry point into the muscle.

(20) The sensor is already amplifying and digitizing the photoelectric signals. These are then transmitted via a connecting line 38 in parallel or serially to the data process.

(21) The intensity adaptation between the illumination unit and the sensitivity of the camera sensor is to be controlled particularly favourable over the length of the switch-on time of the LEDs of the illumination unit. Thus, a very precise constant current source 9 can keep the light intensity from one light pulse to the next very stable, but the intensity can be set very precisely in small steps.

(22) With the help of a rechargeable battery 14, the sensor module can be operated during training without a fixed supply connection.

(23) Optionally, during the acquisition process the current data may be communicated to other devices via a digital radio link 12 with an antenna 13 (incorporated into the housing). Another possibility is the transmission of data during training, but especially after the completion of data acquisition by wire via an output unit 15 to other devices. This cable can be used simultaneously to power and charge the battery 14.

(24) In a memory 10, the recorded physiological parameter values can be buffered.

(25) On other devices the data can be further processed and evaluated, but above all they can be displayed and stored in a larger context.

(26) FIG. 3 shows the optical arrangement in a section through the device and the underlying muscle. Thus, light 17 enters the tissue 16 from the respectively controlled monochrome LED of the multispectral LED illumination unit 32 and is many times diffusely scattered and—on its way through the tissue—absorbed, depending on the substance concentrations and the wavelength. Because this is a random process, the light is scattered and reflected in all directions in an equally distributed manner. However, individual light paths, which also impinge on the camera sensor, have only been drawn by way of example. As described, this principle is called transflection.

(27) FIG. 4 shows a preferred embodiment of the multispectral LED illumination unit with seven separate wavelengths; more specifically, the LEDs have seven separate emission maxima, with the half-widths of the emission spectra being about 15 nm. The size of the module is typically 4.9 mm×3.2 mm×1.2 mm (length×width×height). The seven different dyes are arranged together on a carrier so that they can be controlled and switched separately via a common cathode. The spectral assignment of the LED is shown in the following table, whereby the predominant function of the measuring signal at the wavelength for the physiological evaluation is indicated shorthand.

(28) TABLE-US-00001 520 nm Strongly pulsatile, but Pos 4 almost isosbestic 575 nm For strongly pulsating Pos 5 O.sub.2Hb band 660 nm For pulse oximetry and Pos 1 SmO.sub.2 720 nm SmO.sub.2 Pos 2 760 nm SmO.sub.2, significant band in Pos 8 the HHb signal 805 nm SmO.sub.2, isosbestic point Pos 3 910 nm Pulse oximetry and SmO.sub.2 Pos 6

(29) The LED 4 (520 nm) is used to differentiate the pulsatile components as a function of the oxygen saturation. At 575 nm (LED 5), the high absorption coefficient of O.sub.2Hb determines the signal through the oxygenated hemoglobin, while at 520 nm, the absorption coefficients of O.sub.2Hb and HHb are very similar (isosbestic). Thus, in an analytical comparison, the two signals can distinguish between the delivery of highly oxygenated blood in the arteries and desaturated blood to the microcirculation and veins, respectively.

(30) The five wavelengths in the NIR range (1, 2, 8, 3, 6) are mainly used for data generation for NIRS technology, that is the determination of muscle oxygenation. For this purpose, a somewhat lower sampling frequency is sufficient since no higher-frequency signals (in particular pulsatile signal components) must be detected.

(31) Since the Pleth curve is known via the evaluation of the pulsations in the VIS spectral range (LEDs 5 and 4), the arterial oxygen saturation (SpO.sub.2) can be determined in conjunction with the signals of LED 1 and LED 6, using the conventional pulse oximetry technology.

(32) The assignment of the wavelengths of the illumination unit to the pulse-based and the absorption-based parameters makes it possible to operate the individual LEDs with different clock frequencies. This is achieved by defining an activation sequence of the LEDs. A preferred embodiment of an LED activation sequence is: D,5,4,2,5,6,1,5,4,8,5,3

(33) D stands for image acquisition without activation of an LED (dark image). Other criteria for the sequence are, that LED 5 is placed equidistantly in the sequence, that LED 4 may only follow LED 5 (which does not imply that 5 must follow the 4) and the pulse oximetry detection LEDs Signals (6,1) are directly consecutive. The LEDs for evaluating the NIRS signals and the dark image are distributed to fill the sequence, as they are integrated over a longer time range.

(34) For the realization of the pulse resolution, the wavelengths intended for this purpose must be activated with the highest individual frequency since, as explained, the motion artifacts and the pulse must be sampled at a higher frequency.

(35) The effective frame rates thus result from the camera frame rate f.sub.s and the occurrence in the sequence: f.sub.D,1,2,3,6,8=1/12.Math.f.sub.s, for the LEDs 1, 2, 3, 6, 8 as well as for the dark image D, f.sub.4=1/6.Math.f.sub.s, for LED 4 and f.sub.5=1/3.Math.f.sub.s, for LED 5.

(36) If, for example, a frame rate of 300 fps (frames per second) is assumed, the individual LED frequencies f.sub.D,1,2,3,6,8 result in each case at 25 Hz, f.sub.4 at 50 Hz, and f.sub.5 at 100 Hz.

(37) FIG. 5 a) shows a possible embodiment in which the measuring device illustrated as a block diagram in FIG. 2 is completely integrated in a compact housing 18. The housing is shaped so that it can be comfortably attached to the body surface over a larger human muscle and can also be integrated into possible fastening systems which hold the device stably on the surface of the muscle (1) even under heavy muscle stress. It is advantageous, if the housing is formed in the manner that it is protected against moisture (sweat, water). FIG. 5 b) shows a further embodiment in which the optical unit of the device is kept very small and can also be attached directly to the muscle at the measuring point. The advanced unit for data processing, communication, and power supply is housed in a second compact housing 19, which is connected via a thin cable to the optical sensor unit. Thus, for example, the optical sensor unit can be firmly integrated into a garment and the second housing for data exchange, power supply, and other functions can be arranged separately next to the actual device for measuring value acquisition.

(38) FIG. 6 shows the individual spectra of pure, oxygenated hemoglobin, deoxygenated hemoglobin, water, the skin pigment melanin, fat, and the typical scattering properties of human tissue. The plot of absorption in the Y-axis is logarithmic. The presentation should mainly show the typical curves of the absorption curves and the characteristic absorption bands. Since the concentrations of the individual chemical components in the muscle can vary greatly and unknown substances contribute in addition to the spectra, in practice a combined evaluation with as many spectral bases as possible by means of absorption spectra makes sense. For this purpose, a multivariate statistical analysis method is used to determine the contributions of the single substances to the recorded spectra, and in particular to determine the relative contribution of oxygenated hemoglobin to deoxygenated hemoglobin (oxygenation), the tissue hemoglobin concentration, and concentrations of the other substances, especially of water. Illustratively, in such an analysis method, the measured spectrum is represented as a linear combination of the individual spectra of the pure substances, and the coefficients of the individual spectra are determined in that linear combination of the individual spectra, which provides the best match with the measured spectrum. For example, when determining the concentration of water, the pronounced absorption of water at 960 nm is used. Since at this wavelength, as can be seen from FIG. 6, absorption by fat and by hemoglobin also occurs, their proportions must be taken into account in the multivariate statistical analysis method. For this, the absorption at 930 nm is used, where fat has a high absorption, and several support points in the wavelength range 800-910 nm, which are characteristic of the contributions of oxygenated and deoxygenated hemoglobin. In the combined statistical evaluation of the absorptions at all wavelengths, therefore, the contributions of the individual substances can be determined, in this example in particular the water.

(39) With the embodiment described above in connection with FIG. 4 with an LED illumination unit with LEDs at positions 1-6 with emission maxima as indicated in the table reproduced above, an evaluation and detection of the water concentration is not possible. If evaluation is also desired in relation to water, at least two more LEDs must be provided: one having an emission maximum in the range of 930 nm to be sensitive to the absorption of fat and an emission maximum in the range of 960 nm to be sensitive to the strong absorption of water there.

(40) FIG. 7 schematically shows the proportions of the measured absorption of hemoglobin when measured in the muscle. The measured optical signal has a constant component and a pulsatile component.

(41) The pulsatile component is generated by the pumping of the arterial blood by the heart. The strength of absorption in arterial blood is not equal at systole and diastole. This allows a differentiation. In a heartbeat, highly oxygenated blood is pumped into the measured part of the body. Here, the oxygen saturation of hemoglobin in a healthy man lies in the range of 95% to 99%. The pulsating signal component is wavelength-dependent and depends on the measuring point and the detected spectral range. From this pulsating signal, for example, the heart rate (HR), the heart rate variability (HRV), and the pulsation index (PI) could be determined as parameters.

(42) The tissue-derived signal is divided into two parts. One part depends on the constituents of the tissue, the other part depends on the scattering properties of the tissue, which influence the real light paths. From the range of the substances' contents, the muscle oxygenation, the tissue hemoglobin index, but principally also the fat and glucose portion (energy supply) can be calculated. The scattering properties of the tissue can be detected via the distance-dependent evaluation on the CMOS sensor array.

(43) On the one hand, the respiratory rate, on the other hand, the movements of the athlete can be detected by the shift of the venous blood portion. During a breathing cycle, a larger amount of blood is shifted from the central area of the body to the peripheral areas and back again.

(44) FIG. 8 schematically shows the generation of the various signals for the calculation of the different parameters. For the individual images of the camera, the spatial dimensions N.sub.Sp is evaluated as a number of columns and N.sub.Z as a number of lines, the row direction (with the column number i as a running variable), running parallel to the connection axis of the LED illumination unit and the camera sensor, and the column direction orthogonal thereto. In the first step, the individual lines, whose intensity decreases with increasing number of columns i and thus increasing distance from the light source, are averaged together. The result averaged over all rows represents an intensity function in the row direction i, with i=1, 2 . . . N.sub.Sp:

(45) ${\stackrel{\_}{I}\left(i\right)}_{\lambda m}=\frac{1}{N_Z}\underset{a=0}{\overset{N_Z}{.Math.}}{I_a\left(i\right)}_{\lambda m}$

(46) Since the wavelength is different for each image within the activation sequence, the index λm is carried along. Since there is an activation start time for each activation within the activation sequence, the average intensity Ī(i).sub.λm also represents a time series as a function of the time r given below. The individual activation phases within the activation sequence are each assigned an activation start time t.sub.k with k==1, 2, . . . M (with M as the number of activation phases within a sequence) and an illumination duration, wherein it must be noted that the illumination duration varies per LED type individually but of course always must be smaller than the time period between two consecutive start times t.sub.k-1 and t.sub.k.

(47) The time duration T.sub.A between the individual activation start times is predetermined by the camera frame rate f.sub.A so that the activation start times of the LEDs are arranged equidistant from each other—but the switch-off times may be individually different:

(48) $T_A=\frac{1}{f_A}=t_k-t_{k-1}=\mathrm{const}.$

(49) The time duration T.sub.S between two start times of successive activation sequences in turn results from the time T.sub.A between the individual activation start times within the activation sequence and the number M of activations of the activation sequence, it being assumed that the activation sequences are repeated continuously without pause:

(50) $T_S=\frac{1}{f_S}=M.Math.T_A=\mu .Math.\left(t_k-t_{k-1}\right)=\mathrm{const}.$

(51) The individual times at which the same LED is switched on give the signal Ī.sub.λm[n,t.sub.k] as a time series with λm as the index for the emission maxima of the LED types D,1,2,4,5,6,8, where D is not an LED type in the true sense, but represents the absence of illumination (dark phase). Instead of the sequence number n of the activation sequence and the activation start time t.sub.k within the activation sequence, the absolute time τ(n,t.sub.k) of the respective activation start time can also be indicated τ(n,t.sub.k)=n.Math.T.sub.s+t.sub.k. The corresponding signals (e.g., Ī[τ] and Ī[n] may therefore be considered equivalent.

(52) The averaged intensity distributions Ī(i).sub.λm of the different wavelengths partially already serve as input data for the calculation of the spectral absorption-based parameters (e.g., SmO.sub.2). To obtain the pulsatile (pulse-based) parameters, the mean intensity distribution Ī(i).sub.λm is averaged over the columns i. Thus, the intensity distribution Ī(i).sub.λm becomes an intensity value Ī.sub.λm, wherein each of these intensity values a time τ(n,t.sub.k)=n.Math.T.sub.s+t.sub.k is assigned to, according to the respective activation, which is not listed in the following formula:

(53) ${\stackrel{\_}{I}}_{\lambda m}=\frac{1}{N_{\mathrm{Sp}}}\underset{i=0}{\overset{N_{\mathrm{Sp}}}{.Math.}}{\stackrel{\_}{I}\left(i\right)}_{\lambda }=\frac{1}{N_Z.Math.N_{\mathrm{Sp}}}\underset{a=0}{\overset{N_Z}{.Math.}}\underset{i=0}{\overset{N_{\mathrm{Sp}}}{.Math.}}{I_a\left(i\right)}_{\lambda m}$

(54) In fact, for each wavelength λm, this is a time series Ī.sub.λm(τ), as shown on the right side in FIG. 8.

(55) If the LED is connected in the already described activation sequence D,5,4,2,5,6,1,5,4,8,5,3, first, the eight time series Ī.sub.D(τ), Ī.sub.5(τ), Ī.sub.4(τ), . . . Ī.sub.3(τ), result, which are shown among another. These can also be combined into a time series Ī[τ], as shown at the bottom right in FIG. 8.

(56) The respective signals Ī(i).sub.λm, Ī(i).sub.λm[τ], and Ī[τ] can be used separately in the evaluation, but also in combination with each other.

(57) In the present invention, this combination is used to allow the evaluation of the individual parameters even during intensive exercise, which is also characterized by strong regularity.

(58) FIG. 9 shows the components contributing to the signal Ī[n]. The periodic signal (20) produced by muscle contraction and vibration (movement) will cause a shift of the hemoglobin in the muscle tissue. The pulsatile signal (21), caused by the heartbeat, will affect mainly the oxygenated arterial blood as already shown and is additionally characterized by its distinctive shape by diastole, systole, and dichrotic notch (by aortic valve closure) of the cardiac cycle. The frequency (22) caused by the respiration is characterized by a shift mainly of the venous part (high HHb part). The respiration rate can be assumed to be always lower than the pulse rate, especially when using the athletic performance examination.

(59) The following properties, which are used by the present invention in order to enable as precise a separation as possible, apply to the three influences shown: pulse, respiration, and (muscle) movement.

(60) 1. The pulse signal is indicated by:

(61) 1.1. A characteristic shape caused by systole and diastole of the heart as well as the closure of the aortic valve and the flexible nature of the arteries, and through Fourier transformation has a specific frequency spectrum with multiple harmonic waves, 1.2. An oxygenation increase per pulse, as the signal is caused by arterial pulse wave (arterial saturation SpO.sub.2 typically at 98%), 1.3. Gradients/rate changes in a defined time range (heart rate variability) are physiologically limited.
2. The respiratory signal is indicated by: 2.1. An almost sinusoidal course, 2.2. A lower frequency than the pulse signal, 2.3. Because the signal is caused by a mainly venous blood shift: 2.3.1. No oxygenation increase per pulse, 2.3.2. In the NIR range relative signal change, since the ratio of constant (veins/microcirculation) to variable proportion (arteries) changes due to increased penetration depth. 2.4. Rate gradients (respiratory rate variability) are physiologically limited.
3. The motion signal is indicated by: 3.1. A periodic but not sinusoidal waveform with many (disordered/undefined) harmonics, 3.2. Because the signal is mainly caused by venous/microcircular blood shift: 3.3. No oxygenation increase per pulse 3.3.1. In the NIR range relative signal change, since the ratio of constant (veins/microcirculation) to variable proportion (arteries) changes due to increased penetration depth. 3.4 Movement patterns can be determined for different sports and extracted by appropriate filtering (e.g., wavelet analysis with special wavelets) and predictive models (e.g., autoregressive moving average).

(62) With this knowledge and the embodiment of the invention (selection and arrangement of the LED wavelengths, use of a fast 2D camera sensor), all physiological parameters of interest can be determined: those related to the pulse and those related to the respiration, as well as those related to the oxygenation of the tissue. Thereby, all of the test person's physiological parameters of interest can be detected with a single sensor device, which can also be made so compact that it can be attached to a location of the test person's body without obstructing the person's exercise. In the prior state of the art, different sensors or measuring devices were necessary for the different areas (pulse, respiration, oxygenation of the muscles), which then had to be attached to different locations of the test person's body. With the present invention, with a single compact optical sensor, the entire physiological parameter set can be detected without obstruction of the exercise.

(63) The described characteristics of the signals pulse, respiration, and movement can be evaluated as follows. The numbering of the properties is included below and used for better representation.

(64) 1. Determination of the Pulse Signal:

(65) 1.1. Characteristic Form of the Pulse Wave:

(66) In each pulse cycle, highly oxygenated blood is supplied to the vascular system, which can not drain immediately due to the resistance of the smaller vessels. This creates a maximum in the blood pressure curve that corresponds to the systolic blood pressure. The conclusion of the aortic valve causes an incision in the blood pressure curve. The blood pressure falls due to the flow of blood to the periphery, and falls to the next heartbeat to a minimum, the so-called diastolic blood pressure. The course of the blood pressure curve, which is composed of systole, incisor, and diastole, is “low pass filtered”, as the distance from the heart increases and the characteristic dicrotic pulse wave is created by the flexibility of the vessels.

(67) This characteristic of the pulse wave can be used to allow the determination of the pulse frequency by analysing the time course Ī[n]. Since the pulse wave is not sinusoidal, through transformation of the time signal into the frequency domain (by Fourier transformation), in addition to the fundamental frequency other upper frequencies result, which are formed as integer multiples of the fundamental frequency. This is given in FIG. 10 for the distribution peak 25, which corresponds to the fundamental frequency of the pulse and to which harmonics 27 (at twice the frequency) and 29 (at three times the fundamental frequency) are identifiable. Thus, the pulse signal can be delimited from the respiration in the frequency domain as well, since this is similar to a sine and occurs without or with hardly pronounced harmonics. This is the case in FIG. 10 for the distribution tip 24, which can therefore be associated with respiration.

(68) 1.2. Oxygenation Increase Per Pulse:

(69) To test the oxygenation changes per pulse, the signals Ī.sub.λ.sub.5[n] and Ī.sub.λ.sub.4[n] are evaluated. The signal Ī.sub.λ.sub.4[n] at 525 nm serves as a reference, since it scans an isosbestic point whose intensity does not depend on the oxygenation. Using the signal at 575 nm Ī.sub.λ.sub.5[n], which is characterized by a large absorption distance between oxygenated and deoxygenated hemoglobin, allows the oxygen saturation to be evaluated for each individual pulse, thus providing a separation between signals of the pulse and due to movement.

(70) 1.3. Limited Change of the Pulse Rate:

(71) In addition, due to physiological limits of heart rate variability, the pulse wave can be filtered by a matched dynamic filter (band pass filter in the frequency domain) with the heart rate as the (dynamic) centre frequency and the allowed variability as the bandwidth, such that the respiration or movement are not included, provided that the corresponding rates (or frequencies) do not run into the defined frequency window here. This will never be the case for the respiration rate, since the pulse signal always has a much higher frequency than the respiration signal. The separation of the arterial pulse signal and the movement is achieved in this case by the examination of the oxygenation.

(72) 2. Detection of Respiration:

(73) 2.1, 2.2 and 2.4: Sinusoidal, Lower Frequency than the Pulse and Limited Change in the Respiration Rate:

(74) The principle of the dynamic filter can also be applied to the signal of breathing. Here, a frequency spectrum is expected that has only a significant peak, since the signal of the breathing is sinusoidal-like and has no or hardly pronounced harmonics (property 2.1), which is the case in the frequency spectrum in FIG. 10 for the distribution peak 24. The filter can be parameterized on the assumption that the respiration rate is significantly lower than the pulse rate (property 2.2) and the variability of the respiration rate within a defined time window is clearly limited (property 2.4).

(75) 2.3. By Venous Blood Shift:

(76) 2.3.1. Low Oxygenation Per Pulse:

(77) By evaluating the oxygenation with the signals Ī.sub.λ.sub.4[n] and Ī.sub.λ.sub.5[n], a delineation of the (arterial) pulse signal and the respiratory signal can be achieved. As a result, a decision can be made as to whether a found frequency can be assigned to the respiration or the pulse.

(78) 2.3.2. Relative Signal Change in the NIR Range:

(79) It should also be noted that the venous shift results in a change of the signal in the NIR range, since in the NIR range, relatively higher signal fluctuations are generated than in the VIS range. This is caused because higher penetration depths into the tissue are achieved in the NIR range and thus the proportion of venous or microcircular signals changes with respect to the arterial signals.

(80) 3. Movement:

(81) Although the motion signal contains no information from which the physiological performance parameters are calculated directly, it must be decided for a found significant frequency, whether this was the pulse, the respiration, or caused by the movement. The examination of the motion signals is therefore to secure the pulse and respiration detection and the exclusion of rhythmic movements. This can be achieved again by examining the oxygenation (property 3.2.1) and changing the signal components in the NIR range (property 3.2.2).

(82) In addition to the determination of oxygenation, the characteristics of the time course during exercise can be used to separate the motion signals. The signals that are generated by the contraction of the muscle or the vibration of the measuring device are characterized by a periodicity, which may differ depending on the performed sport (property 3.1). For example, a stronger influence of shocks is expected for carrying out a running training than for bike training. The different exercises produce specific waveforms, which may require the use of adapted analysis forms, e.g. enable wavelet analysis. The motion pattern can be stored as a wavelet and the signal can be examined. The wavelet can be considered as a special band pass filter, whose shift in the frequency domain produces a maximum where the frequency of the motion occurs.

(83) FIG. 10 shows the frequency spectrum that can be generated from the discrete time signal Ī[n]. The respiration 24 is characterized by a single peak, the pulse is described by the fundamental frequency 25 and the harmonics 27 and 29. The signal of the movement has the fundamental wave 26 and harmonics 28 and 30. It can be seen, that the examination of the pulse rate by the query of the harmonics is not necessarily sufficient to decide whether one of the fundamental frequencies 25 and 26 is about pulse or movement. For this, the query of the oxygenation can be carried out.