Method and device for measuring pulse wave velocity in living tissue

Abstract

A method for determining pulse wave velocity (v) in living tissue by optically measuring a pulse wave (P(t)) is described. A measurement unit (2) having an array (4) of optical photosensors (S1... SN) distributed over a length (L) is arranged against the tissue. Repetitive measurements are performed, with each measurement including the detection, by means each photosensor (Si), of a parameter (pi). This gener- ates a time series of measurement vectors p(tj) = (p1(tj)... pN(tj)) at times tj. For at least some of the measurement vectors, the time interval delta_t = tj+1 - tj between consecutively recorded measurement vectors is no more than L / v0, with v0 being 25 m / s. These measurement vectors are then processed for determining the pulse wave velocity (v). The high rate of measurement allows tracking of a pulse feature along the measurement vectors.

Description

[0001] Method and device for measuring pulse wave velocity in living tissue

[0002] Technical Field

[0003] The invention relates to method and a device for determining pulse wave velocity in living tissue by optically measuring a pulse wave P(t).

[0004] Background Art

[0005] Pulse wave velocity (PWV) is an important parameter for assessing the cardiovascular state of a patient, in particular the state of blood vessels (such as arterial stiffness) and blood pressure, see, e.g., P. M. Nabeel et al., "Association of incremental pulse wave velocity with cardiometabolic risk factors", nature-portfolio Scientific Reports (2021)11 :15413, https: / / doi.org / 10.1038 / s41598-021-94723-2.

[0006] One class of known methods rely on non-local measurements of the propagation of blood pulses. They are based on detecting the time a pulse needs to propagate from one part of the body to another part of the body, e.g., from the chest to the limb. These methods are poorly suited for locally worn devices, nor can they be used for reliably assessing local vessel properties.

[0007] Another class of know methods rely on local measurements. For example, J.M. Meinders et al, "Assessment of Local Pulse Wave Velocity in Arteries Using 2D Distension Waveforms", Ultrasonic Imaging 23, 199 - 215 (2001) describes an ultrasonic measurement based on assessing the ratio of the temporal and spatial gradients of vessel distension.

[0008] US 10624586B2 and US10292662B2 describe methods based on video capture from a distance.

[0009] PWV can be determined optically. In this case, the pulse wave P(t) is measured. P(t) is a measure of how much light the tissue reflects or transmits at a given time, with the light being in a spectral range where reflection and / or transmission depends on the amount of blood in the tissue.

[0010] Disclosure of the Invention

[0011] The problem to be solved by the present invention is to provide a method and device with good accuracy for locally determining pulse wave velocity in living tissue by optically measuring a pulse wave P(t). This problem is solved by the method and device of the independent claims.

[0012] Accordingly, the method for measuring pulse wave velocity in living tissue by optically measuring the pulse wave P(t) comprises at least the following steps:

[0013] - Arranging a measurement unit against the tissue. The measurement unit comprises an array of optical photosensors Si ... SN distributed over a length L along a direction X.

[0014] - Performing repetitive measurements by means of the array of N photosensors Si ... SN. Each measurement includes detecting, by means each photosensor Si, a parameter pi indicative of the amount of blood within the measurement range of the photosensor, thereby generating a time series of measurement vectors p(tj) = (pi(tj) ... p\(tj)) at times tj. For at least a first set of the measurement vectors, which span a phase R of the cardiac cycle, the time interval At = tj+i - tj between consecutively recorded measurement vectors is no more than L / vO, with vO being 10 m / s, in particular vO being 25 m / s, for the reasons explained below.

[0015] - Processing, the first set of the measurement vectors for determining the pulse wave velocity.

[0016] This is based on the understanding that a better measurement can be achieved if the temporal resolution of the measurements, i.e., the time interval At, is no more than L / vO as given above. In this case, a given feature of the pulse waveform (such as a maximum, minimum, a change of slope, or a certain value) traveling past the array of photosensors at the times tj is recorded, at different positions, in several of the measurement vectors. Typical values of PWV in tissue are around 5 m / s, but they may reach values as high as 12 m / s (see, e.g., van Hout et al. J Cardiovasc Magn Reson (2021) 23:46, https: / / jcmr-online.biomedcentral.com / articles / 10.1186 / sl2968- 021-00739-y). Hence, by choosing At to be no more than L / vo with vo = 10 m / s, in particular 25 m / s, even the fastest pulses will have at least parts of the same feature visible at different positions in several consecutive measurement vectors p(tj) of said "first set". Thus, the progress of this pulse wave feature can be tracked across measurement sets, which improves accuracy.

[0017] The phase R of the first set of measurement vectors may span (i.e., the first set of measurement vectors may be recorded over) only a comparatively short part of the cardiac cycle. In particular the phase R may span a time of no more than 10%, in particular no more than 1%, of a cardiac cycle. Using such short sections simplifies signal processing because the shape of the pulse waveform is (but for an offset along the vector index) substantially the same in all measurement vectors of the first set. Further, it allows to account for the fact that different sub-sections of the cardiac cycle may propagate at different velocities. For example, sub-sections where the instantaneous blood pressure is high (e.g., the systolic peak) may have higher PWV than sub-sections where the instantaneous blood pressure is low (e.g., the pulse foot (also called the systolic foot or trough)). Thus, when restricting the phase R to a small section of the cardiac cycle, more precise measurements can be achieved.

[0018] The phase R of the first set of measurement vectors may be selected to span a period where the time-derivative 5P(t) / 5t of the pulse wave P(t) has a given value DO. In particular, DO may be zero, i.e., the first set of measurement vectors is recorded over at least one of the following parts of the cardiac cycle: the systolic peak, the pulse foot, the diastolic trough, and the diastolic peak.

[0019] The phase R of the first set of measurement vectors may also be selected to span a period where the pulse wave P(t) has a given value VO. Other examples of selecting the phase R are given below.

[0020] The method may comprise the step of varying the time interval At over a cardiac cycle between values larger than L / vo, in particular larger than 10-L / vo, and values smaller than L / vo. This is based on the understanding that a high temporal resolution is only required for those sections of the cardiac cycle for which PWV is to be measured. This allows to reduce the power consumption required for the measurement.

[0021] For determining the temporal location of the phase R, i.e., the phase for recording the first set of measurement vectors at a high rate, the method may include the following steps:

[0022] - Recording measurement vectors p(tj) over one or more cardiac cycles. These measurement vectors, called reference vectors, may be recorded at lower or higher rates. To save power, a majority or all of them may be recorded at time intervals At larger than L / vo.

[0023] - Determining, from the reference vectors, the start of the phase R within a given cardiac cycle. For example, from the reference vectors, at least one parameter descriptive of the pulse shape can be determined, which allows to determine the start of the phase R in respect to the given cardiac cycle.

[0024] The method may further comprise the step of determining the blood pressure from the pulse wave velocity.

[0025] The invention also relates to a device for measuring pulse wave velocity that has a measurement unit comprising an array of optical photosensors Si ... SN distributed over a length L and a control unit adapted to execute the steps of the method. The optical photosensors Si ... SN may comprise single-photon avalanche diodes. These diodes are well-suited for recording small optical signals at high measurement rates.

[0026] The measurement unit may comprise a single substrate with the array of photosensors arranged on the substrate.

[0027] Brief Description of the Drawings

[0028] The invention will be better understood and objects other than those set forth above will become apparent when consideration is given to the following detailed description thereof. Such description makes reference to the annexed drawings, wherein:

[0029] Fig. 1 shows an embodiment of a measurement unit,

[0030] Fig. 2 shows a block diagram of an embodiment of a device for measuring pulse wave velocity,

[0031] Fig. 3 shows a pulse wave P(t), i.e., a photoplethysmogram,

[0032] Fig. 4 illustrates possible times of measurement over one cardiac cycle,

[0033] Fig. 5 shows the measured parameters pi for several consecutive measurements that were recorded during the phase R, and

[0034] Fig. 6 shows another embodiment of a measurement unit.

[0035] Modes for Carrying Out the Invention

[0036] Device

[0037] Figs. 1 and 2 show an embodiment of a device for measuring PWV.

[0038] The device comprises a measurement unit 2, which may, e.g., be formed by an integrated optics chip. It comprises an array 4 of photosensors Si ... SN distributed over a length L along a direction X. In order to provide a robust assembly with a well-defined spacing between the photosensors, the measurement unit may comprise a substrate 3, with the whole array 4 of photosensors arranged thereon or therein.

[0039] In application, direction X is advantageously aligned with the primary direction of blood flow in the tissue. For further information on some of the possible embodiments of the photosensors Si ... SN, see the section "Photosensors" below.

[0040] Measurement unit 2 further comprises a light source 6 adapted to send light pulses into the tissue to be measured. For this, light source 6 may be coupled to a light guide 8. Light guide 8 extends along array 4. The light from light source 6 travels along light guide 8. One or more outcouplers 10, such as surface gratings, are provided for coupling out the light towards the tissue.

[0041] The photosensors Si measure the amount of light returning from the tissue. For a measurement unit 2 as shown, e.g., in Fig. 1, which carries out reflection measurements, the measured light is light from light source 6 that is reflected, typically reflected in elastic and diffuse manner, by the tissue. The light measured by the photosensors may, however, also originate from non-elastic scattering processes.

[0042] The spectrum of light source 6 may, e.g., be in the range of 500 nm to 600 nm. The reflectivity of light within this range depends strongly on the amount of blood in the tissue because some strong absorption peaks of hemoglobin lie within this range.

[0043] As shown in Fig. 2, the device further comprises a control unit 12 connected to the photosensors Si ... SN and light source 6 by means of interface circuits 14 and 16, respectively.

[0044] Control unit 12 may, e.g., be a microprocessor or microcontroller programmed to carry out the method as described herein.

[0045] Alternatively to using light pulses, the illumination from light source 6 may be continuous in the sense that it spans at least several measurements. This simplifies device design. Pulsed illumination, on the other hand, reduces power consumption and / or allows for larger light intensities during the measurements.

[0046] Method

[0047] In operation, measurement unit 2 is arranged against the tissue of the patient. In this context, "arranging against the tissue" is to be understood such that the distance between the sensor array and the surface of the tissue is less than the distance L / (N - 1) between adjacent photosensors in order to spatially resolve the light from the tissue by means of the photosensors.

[0048] As shown in Figs. 3 and 4, control unit 12 performs, over each cardiac cycle CC, repetitive measurements at times tj with j = 1 to J. In each measurement, the tissue is illuminated by means of light source 6, and the photosensors Si measure the light returning from it. The parameter pi measured by a given photosensor Si represents the pulse wave P(t) and is, in most cases, a monotonous function of the amount of blood within the measurement range of the photosensor Si. Fig. 3 illustrates how the pulse wave P(t) and therefore the parameters pi may vary over time. Each parameter pi as a function of time t represents a photoplethysmogram as known to the skilled person.

[0049] The pulses shown in Fig. 3 travel, spatially, along the primary direction of blood flow with a pulse wave velocity v. In approximation, the PWV v is given by the Moens-Korteweg equation and the Hughes equation as follows: with E being the elastic modulus at pressure 0, ho, p, Ro, and ('being parameters related to the size and other properties of the blood vessels, and PB being blood pressure. For details, see Y. Ma et al., "Relation between blood pressure and pulse wave velocity for human arteries", PNAS October 30, 2018, vol. 115, No. 44, pp. 11144 - 11149 (https: / / doi.org / 10.1073 / pnas.1814392115).

[0050] Pulse wave velocity v is a function of blood pressure. As the blood pressure increases, the artery stiffens, which in turn increases v. Hence, pulse wave velocity v is not exactly constant over the cardiac cycle. Rather, the instantaneous pulse wave velocity at, e.g., the systolic peak SP, is generally higher than, e.g., at the pulse foot PF. Hence, the velocity of different sections of the pulse varies. This is one of the reasons why the waveform of the pulse will change as it propagates through the body.

[0051] To measure the PWV of a desired section of the cardiac cycle CC, repetitive measurements are carried out over the cycle, at times ti, t2, ... tj as illustrated in Fig. 4. At each time tj, a measurement vector p(tj) = (pi(tj) ... p\(tj)) is calculated from the signals from the photosensors Si, ... SN. The parameters pi may correspond to the raw signals n of the photosensors Si, or they may be derived therefrom as, e.g., described in the section "Raw signal processing" below.

[0052] In order to save processing power, the time interval At = tj+i - tj between the measurement times ti, t2, ... tj is, in an embodiment, not constant. This is illustrated in Fig. 4. Here, the PWV at the systolic peak SP is to be measured. To do so, control unit 12 analyses the shape of the pulse wave P(t) and identifies, within the cardiac cycle CC, a phase R that includes the systolic peak PS but extends over only part of the cardiac cycle CC, in particular over less than 10% thereof. Details on how to determine this phase R can be found in the section "Determining the phase R" below.

[0053] During the phase R, control unit 12 performs measurement at a higher rate than outside phase R and records a "first set" of measurement vectors at times tjo.tjo+i, etc. The time interval Atabetween consecutive measurement vectors of the first set is no more than L / vo, with vo being 25 m / s, in particular with vo being 100 m / s.

[0054] The number of measurement vectors in this first set is at least 4, in particular at least 8, in order to have a rich dataset to perform the analysis as, e.g., described in the following.

[0055] Assuming, e.g., that L = 1 cm or less, Atais 100 ps or less, corresponding to a sampling rate of at least 10 kHz. For vo being 100 m / s, Atawould be 25 ps or less.

[0056] On the other hand, outside the phase R, the time interval Atb between consecutive measurements is larger than L / vo, in particular larger than 10-L / vo, thereby generating a "second set" of measurement vectors recorded at a lower rate of measurements.

[0057] Fig. 5 illustrates the values of consecutively measured measurement vectors p(tj) during the phase R, with curve Ml representing a first measurement vector p(tjo), M2 representing a second measurement vector p(tjo+i), M3 representing a third measurement vector p(tjo+2), M4 representing a fourth measurement vector p(tjo+s), and M5 representing a second measurement vector p(tjo+4). The vertical axis shows the values of the parameters pi, and the horizontal axis shows the index i of the sensors Si.

[0058] In Fig. 5, it is assumed that the phase R spans the systolic peak SP. Hence, each measurement vector p(tj) comprises the peak value, at locations fi, fz, ... Since Atais no more than L / vo, the peak is visible in several consecutive measurement vectors p(tj) but at different locations thereof.

[0059] In order to determine how a pulse feature, e.g., the systolic peak SP, travels through the measurement vectors p(tj) of the first set, various techniques may be used. Some of these are described in the following section.

[0060] Determining PWV

[0061] PWV is determined from the first set of measurement vectors p(tj) by tracking a pulse feature as it propagates through the vectors. The skilled person knows of methods how to do this. The present section describes some techniques that have specific advantages. In one of these techniques, the shape of the pulse wave P(t) as recorded by the photosensors is approximated by a function P(t) = F(t, ai ... ax), with the ai ... ax being curve parameters. In other words, function F approximates the pulse wave at least in region R.

[0062] In this case, the parameters pi(tj) in the measurement vectors of the first set can be written as pi(tj) = F(t(i, j), ai ... aK) (2) with t(i,j) given by the times tj when the measurement vectors were recorded, by the position of the photosensors Si along direction X, and by the value v the PWV. For equally spaced times tj (at intervals Ata) and equally spaced photosensors Si (at distances L / (N - 1)), we have t(i, j) = tO + Ata • (j —jO) - i • L / (N - 1) / v. (3)

[0063] Eq. (3) accounts for the time when a given part of the pulse wave reaches photosensor Si in the measurement at time tj. tO is a parameter that may be estimated or obtained by fitting. jO is the known index of the first measurement vector p(tjo) in the "first set" of measurement vectors, i.e., it is the index of the first measurement vector within phase R.

[0064] In general, the parameters ai ... ax, v, and (optionally) tO are unknown. Given the measured parameters pi(tj) of the first set of measurement vectors in phase R, they may be obtained by means of curve fitting, i.e., by minimizing the errors between the predictions from function F (as calculated with Eqs. (2) and (3)) and the parameters in the measurement vectors.

[0065] In more general terms, therefore, the method comprises the steps of

[0066] - approximating the pulse wave P(t) at least within the phase R by a function F(t, ai ... ax), with the ai ... ax being curve parameters, and

[0067] - determining the pulse wave velocity by varying the pulse wave velocity and the curve parameters to make the function F(t, ai ... ax) fit the measurement vectors of the first set.

[0068] In this case, the time t in function F may be calculated using the times tj of the measurements of the measurement vectors of the first set as well as using the positions of the photosensors along direction X. In particular, if the measurement vectors have equal temporal spacing Ataand the photosensors have equal spatial spacing L / (N - 1), the relation of Eq. (3) may be used. Advantageously, function F(t, ai ... an) is nonlinear in time t, which, e.g., allows to track a maximum, a minimum, or a bent section of the pulse wave P(t) as it propagates through the measurement vectors.

[0069] For example, function F may be a polynomial of second order, i.e.,

[0070] F(tij, ai ... ax) = ai + a? • tij + as • tij2(4) with tij as given in Eq. (3).

[0071] In this case, linear curve fitting can be used to find the parameters ai, a2, as, v, and (optionally) tO.

[0072] The above is only one of the ways for calculating the velocity v from the measurement vectors p(tj).

[0073] In another embodiment, the data in each measurement vector may be preprocessed, e.g., by first calculating the numerical derivative in respect to the index i, i.e.

[0074] P'(tj) = (p2(tj) - pi(tj), (p3(tj) - p2(tj) ... pN(tj) - pN-l(tj)) (5) and then using a function F that approximates the derivative of P(t).

[0075] In yet another embodiment, the discrete cross-correlations

[0076] CCjlj2 = p(tji) * p(tj2) (6) between pairs j 1, j2 of measurement vectors p(tji), p(tj2) obtained within the phase R can be calculated. From the (real-valued) position jji j2 of the maximum of ccjij2, an estimate Vjij2 of the velocity v can be obtained from

[0077] Vjlj2 =jjlJ2 - L / (N - l) / (j2 -jl) / Ata. (7) v can, e.g., be estimated from the average of the values Vjij2 of different pairs j 1, j2 of measurement vectors.

[0078] In the example of Figs. 4 and 5, the tracked pulse feature is the systolic peak SP. In other embodiments, or in addition to this, one or more other pulse features may be tracked as they propagate over the measurement vectors p(tj), in particular: a) The diastolic peak DP; b) The diastolic trough DT; c) The pulse foot PF; d) The location where the pulse wave P(t) has a given slope, i.e., the time-derivative 5P(t) / 5t has a given value DO. e) The location where the pulse has a maximum positive or minimum negative slope, such as at the maximum rising slope RS; f) The location where the pulse wave P(t) has a given value VO.

[0079] As to d), this represents a more general case than a), b), and c) (where the tracked pulse feature is a location where the slope DO is zero).

[0080] As to f), for example, R can, e.g., be positioned at the middle point between the maximum value of P(t) at the systolic peak SP and the minimum value of P(t) at the pulse foot PF.

[0081] The tracked pulse feature is selected by suitably positioning the phase R during which the first set of measurement vectors is recorded, i.e., those measurement vectors that are then combined for determining the PWV. Methods for positioning (determining) the phase R are described in the next section.

[0082] As mentioned, function F(t, ai ... ax) may be nonlinear in time t, i.e. it may have a second-order derivative that is non-zero. This is of particular importance when tracking a feature (such as the systolic peak SP, the diastolic peak DP, the diastolic trough DT, or the pulse foot PF) where P(t) has a maximum or a minimum.

[0083] Determining the phase R

[0084] The positioning of the phase R, within the cardiac cycle defines the part (i.e., the feature) of the pulse wave P(t) that is to be tracked and for which the PWV is to be measured.

[0085] Phase R covers only a small part of the cardiac cycle CC. Assuming, for example, that PWV is 5 m / s and the cardiac cycle has a length of 1 s, the spatial length of a single pulse wave is 5 m. If we have a photosensor array with a length L = 1 cm, the array can record 0.2% of the pulse wave in a single measurement vector p(tj). If it is important to track a specific feature of the pulse wave P(t) exactly, an accurate placement of phase R is required. On the other hand, it must be noted that the techniques described in the previous section can also be used if R is not positioned exactly at the relevant feature but merely close to it. For example, when using a second-order polynomial approximative function F(t, ai ... ax), tracking will also work in R is not located exactly at a peak or trough of the wave.

[0086] However, even if phase R need not be placed with an accuracy of less than 1% in all cases, it should be advantageously be placed with an accuracy of 10% or less in order to make a measurement of a defined section of the pulse wave P(t). In particular, as mentioned above, PWV is a function of pressure, and, therefore, different values for PWV may, e.g., be measured at the systolic peak SP and the pulse foot PF. On the other hand, measurements at the maximum rising slope RS may be of interest because the high derivative of P(t) at that point increases measurement accuracy.

[0087] To determine the start of the phase R, the pulse wave P(t) may be observed over one or more cardiac cycles, e.g., by recording measurement vectors p(tj) over one or more cardiac cycles. In the following, these measurement vectors are called "reference vectors". The reference vectors need not be recorded at a high rate, but they should cover the whole cardiac cycle.

[0088] From the reference vectors, the start time of phase R can then be determined.

[0089] In one example, estimates Pe(tj) can be calculated as the average value of the parameters pi(tj) of the reference vectors p(tj) recorded at the times tj. Spline interpolation may be applied to the estimates Pe(tj) for identifying the time of the feature of interest, and therefore of the phase R, within the cardiac cycle, e.g., relative to the start of the cardiac cycle. Alternatively to spline interpolation, an analytical or heuristic mathematical model of the pulse wave P(t) may be used, and the parameters of the model can be obtained by means of fitting the model to the estimates Pe(tj)-

[0090] Once an estimate of the start time of phase R in the cardiac cycle is knows, a next phase R can be started at this time within a cardiac cycle.

[0091] Raw signal processing

[0092] As mentioned, the length L of array 4 is typically short as compared to the spatial length of the pulse wave. Hence, the signal variation over the photosensors Si ... SN due to the spatial change of the pulse wave are small. On the other hand, the raw signals n(tj) measured by the photosensors Si ... SN may be affected not only by the pulse wave but also by local tissue properties, such as skin unevenness, hairs, pigmentation, vessel distribution, etc. In addition or alternatively thereto, the raw signals n(tj) may also depend on the brightness of the light from the light source at the locations measured by the photosensors, and this brightness may differ between these locations. The effects of these tissue properties and / or light-source properties may be larger than the variation due to the spatial change of the pulse wave.

[0093] Hence, advantageously, the method comprises the steps of - measuring raw signals n(tj) by means of the array of N photosensors Si ... SN, and

[0094] - processing the raw signals n(tj) to compensate local tissue properties in order to calculate the parameters pi(t).

[0095] The second step may, e.g., be performed by measuring the raw signals n(tj) over one or more cardiac cycles and calculating, from these raw signals n(tj), an average raw signal Si for each sensor Si. The average Si may, e.g., be the arithmetic mean, weighted arithmetic mean, median, or modulus of the raw signals n(tj). The average Si may then be used for scaling subsequently measured raw signals for calculating the parameters pi(tj), e.g., by

[0096] Pi(tj) = n(tj) / Si. (8)

[0097] Hence, in one embodiment, the processing of the raw signals n(tj) comprises the scaling of the raw signals n(tj) measured by a given sensor Si with an average Si of the raw signal n from the sensor Si averaged over a plurality of measurements, in particular over at least 100 measurements.

[0098] Photosensors

[0099] For the reasons mentioned above, the photosensors Si ... SN may comprise single-photon avalanche diodes (SPADs). SPADs provide a high sensitivity and can be operated at large frame rates in the order of several kHz.

[0100] Alternatively, though, other types of detectors may be used in the photosensors, such as avalanche photodiodes or generic photodiodes.

[0101] In addition to the detector itself, each photosensor may comprise optics, such as a light guide and / or a lens to carry the light arriving at the measurement unit to the detector.

[0102] In the embodiment above, the photosensors are equally spaced long direction X. Alternatively, non-equal spacing may be used. In this case, Eq. (3) has to be adapted accordingly.

[0103] Fig. 1 shows the photosensors to be arranged in a one-dimensional array 4. This reduces the number of photosensors to be used and reduces the data that needs processing, but it makes it necessary or at least advisable to properly align array 4 along the body tissue, with direction X extending substantially along the flow direction of the blood. This may, e.g., be easily implemented by mounting measurement unit 2 in device with a wristband, with direction X extending perpendicularly to the wristband. In an alternative embodiment, though, as shown in Fig. 6, a two-dimensional array 4' of photosensors Sij may be used. A two-dimensional array provides various advantages that may be used individually or in combination:

[0104] - The two-dimensional array provides information for the blood flow not only along a single direction but in two directions. Hence, by analyzing the measurement vectors p(tj) recorded by such an array, the primary direction of blood flow and its velocity along this direction can be calculated. The alignment between the array with the direction of blood flow does not matter.

[0105] - The two-dimensional array allows a measurement at a larger number of positions. Hence, for example, if first part of the tissue beneath it provides a weaker signal than a second part, e.g., due to variations is skin reflectivity or blood vessel density, the second part will provide a better-quality measurement.

[0106] - Combining several measurements along a direction perpendicular to the primary direction of blood flow generates data redundancy that can be used for obtaining a measurement with better signal-to-noise ratio.

[0107] For illumination, the rows of a two-dimensional array 4' as shown in Fig. 6 may, e.g., be interspersed with linear light guides 8 of the type shown in Fig. 1. Alternatively, e.g., a light guide 8 with suitable couplers may be arranged to cover the photosensors Sij.

[0108] For one- or two-dimensional arrays, for good spatial resolution, the number N of photosensors, along direction X, is at least 4. However, in many embodiments, it is advantageously at least 16, in particular at least 128, for tracking pulse features with high resolution.

[0109] For recording a measurement vector p(tj), light source 6 is operated to send a light pulse into the tissue at time tj. Each photosensor Si records a raw signal n(tj), which then may be processed to generate the parameter pi(tj) as described above.

[0110] Measuring blood pressure

[0111] As mentioned in the context of Eqs. (la, lb), blood pressure and PWV are correlated. However, these equations are not necessarily accurate in all circumstances. PWV often provides a more robust biomarker.

[0112] However, the present method and device can also be used for measuring a patient's blood pressure.

[0113] In one embodiment, after calibration, the parameters E , ho, p, Ro, and of Eq. (la, lb) are relatively constant. Hence, both systolic and diastolic blood pressure may be estimated due to the linear relationship between the logarithm of PWSV squared and blood pressure, see details as described in the context of Eq. (18) of Y. Ma et al. (cited above).

[0114] Notes

[0115] In the embodiments above, the photosensors Si and light source 6 are arranged to perform a reflective measurement on the tissue, i.e., the photosensors Si detect light reflected from the tissue. Alternatively, though, the device may be adapted to perform a transmission measurement.

[0116] As mentioned, measurement unit 2 may be part of a wearable device, such as a device worn on a limb, i.e., all the photosensors 4 are arranged locally on the body. The methodology described herein allows to measure pulse wave velocity v even by means of a locally worn device.

[0117] In the embodiment above, pulse wave velocity v was measured during one phase R of the cardiac cycle. In other embodiments, several phases R in a single cardiac cycle can be used to measure PWV v at various sections of the pulse wave, e.g., with one phase R at the systolic peak SP and another one at the pulse foot PF.

[0118] The average of PWV over long time periods on the order of days may provide a measure of long-term cardiovascular risk while the average of PWV over short time periods on the order of minutes may provide a measure of peak (systolic) blood pressure and trough (diastolic) blood pressure for the assessment of acute events.

[0119] While there are shown and described presently preferred embodiments of the invention, it is to be distinctly understood that the invention is not limited thereto but may be otherwise variously embodied and practiced within the scope of the following claims.

Claims

Claims1. A method for determining pulse wave velocity (v) in living tissue by optically measuring a pulse wave P(t) comprising the steps of arranging a measurement unit (2) against the tissue, with the measurement unit (2) comprising an array (4) of N optical photosensors Si ... SN distributed over a length L along a direction X, performing repetitive measurements at times tj by means of the array (4) of N photosensors Si ... SN, wherein each measurement includes detecting, by means each photosensor Si, a parameter pi indicative of an amount of blood within a measurement range of the photosensor, thereby generating a time series of measurement vectors p(tj) = (pi(tj) ... pN(tj)) at the times tj, wherein, at least for a first set of the measurement vectors spanning a phase R of a cardiac cycle (CC), a time interval At = tj+i - tj between consecutively recorded measurement vectors is no more than L / vo, with vo being 10 m / s, in particular vo being 25 m / s, and processing the first set of the measurement vectors for determining the pulse wave velocity (v).

2. The method of claim 1 wherein vo = 100 m / s.

3. The method of any of the claims wherein the phase R spans no more than 10% of the cardiac cycle (CC).

4. The method of any of the preceding claims wherein the phase R spans a period where a time-derivative 5P(t) / 5t of the pulse wave P(t) has a given value DO.

5. The method of claim 4 wherein DO = 0.

6. The method of any of the preceding claims wherein the phase R spans a period where the pulse wave P(t) has a given value V0.

7. The method of any of the preceding claims comprising the steps of- approximating the pulse wave P(t) in the phase R by a function F(t, ai ... ax), with the ai ... ax being curve parameters and- determining the pulse wave velocity (v) by varying the pulse wave velocity (v) and the curve parameters ai ... aK to make the function F(t, ai ... an) fit the measurement vectors of the first set.

8. The method of any of the preceding claims comprising the step of calculating cross-correlations of the measurement vectors p(tj) at the consecutive times.

9. The method of any of the preceding claims comprising the step of varying the time interval At over a cardiac cycle (CC) between values larger than L / vo and values smaller than L / vo.

10. The method of claim 9 comprising the step of varying the time interval At over a cardiac cycle (CC) between values larger than 10-L / vo and values smaller than L / vo.

11. The method of any of the claims 9 or 10 comprising the steps of recording, over one or more cardiac cycles (CC), measurement vectors p(tj) as reference vector(s), determining, from the reference vector(s), a start of the phase R in a given cardiac cycle (CC).

12. The method of any of the preceding claims further comprising the step of determining a blood pressure from the pulse wave velocity (v).

13. The method of any of the preceding claims wherein the photosensors comprise single-photon avalanche diodes.

14. The method of any of the preceding claims further comprising the step of sending, for each measurement, a light pulse into the tissue.

15. The method of any of the preceding claims wherein N is at least 4, in particular at least 16, in particular at least 128.

16. The method of any of the preceding claims comprising the steps of- measuring raw signals n(tj) by means of the array (4) of N photosensors Si ... SN, and- processing the raw signals n(tj) to compensate local tissue properties in order to calculate the parameters pi(t).

17. The method of claim 16 wherein processing the raw signals n(tj) comprises scaling the raw signals n(tj) measured by a given sensor Si with an average Si of the raw signal from the sensor Si averaged over a plurality of measurements.

18. The method of any of the preceding claims wherein a number (K) of measurement vectors in said first set is at least 4, in particular at least 8.

19. A device for measuring pulse wave velocity (v) having a measurement unit (2) comprising an array (4) of optical photosensors Si ... SN distributed over a length L and a control unit adapted to execute the steps of the method of any of the preceding claims.

20. The device of claim 19 wherein the optical photosensors Si ... SN are single-photon avalanche diodes.

21. The device of any of the claims 19 or 20 further comprising at least one pulsed light source (6) adapted to send light pulses into the tissue.

22. The device of claim 21 further comprising a light guide (8) extending along the array (4) of optical photosensors, wherein the light source (6) is coupled to the light guide.

23. The device of any of the claims 19 to 22 wherein the array is a one-dimensional array.

24. The device of any of the claims 19 to 22 wherein the array is a two- dimensional array.

Images