System and method for sensing physiological parameters
阅读说明：本技术 用于感测生理参数的系统和方法 (System and method for sensing physiological parameters ) 是由 L·J·辉基布雷格茨 G·德哈恩 M·J·H·范加斯特尔 于 2018-11-13 设计创作，主要内容包括：本发明提供了生理参数感测系统(50)和方法,其中,导出指示至少一个生理参数的生理信息。本发明的方法基于根据至少两个检测信号的不同加权组合来构建多个脉搏信号,所述至少两个检测信号是根据检测到的被引导到对象的皮肤区域上的或被引导通过对象的皮肤区域的电磁辐射导出的。所述加权基于各种血容量脉搏向量的集合中的不同的血容量脉搏向量。针对每个生成的脉搏信号导出质量指标值,其中,这基于针对患者的获得的心率信号与脉搏信号之间的导出关系。使用得到具有最高质量指标值的脉搏信号的血容量脉搏向量和/或根据所导出的脉搏信号本身来导出生理参数信息。(The invention provides a physiological parameter sensing system (50) and method, wherein physiological information indicative of at least one physiological parameter is derived. The inventive method is based on constructing a plurality of pulse signals from differently weighted combinations of at least two detection signals derived from detected electromagnetic radiation directed onto or through a skin area of the subject. The weighting is based on different blood volume pulse vectors in the various sets of blood volume pulse vectors. Deriving a quality indicator value for each generated pulse signal, wherein this is based on a derived relationship between the obtained heart rate signal and the pulse signal for the patient. The physiological parameter information is derived using the volume pulse vector that resulted in the pulse signal with the highest quality index value and/or from the derived pulse signal itself.)
1. A physiological parameter sensing system (50), comprising:
a sensing interface (52) adapted to obtain at least two detection signals derived from detected electromagnetic radiation reflected from or transmitted through a skin area of the subject's body;
a heart rate sensing module (54); and
a processor (58) operatively coupled with the sensing interface and the heart rate sensing module and adapted to:
deriving at least two pulse signals, each pulse signal being formed from a weighted combination of the detection signals, wherein the weighting for each pulse signal is based on components of a different one of a set of at least two blood volume pulse vectors;
deriving, for each derived pulse signal, a quality indicator value based on a characteristic of a derived relationship between the pulse signal and a heart rate signal of the subject, the heart rate signal being sensed by the heart rate sensing module; and
deriving physiological information indicative of at least one physiological parameter from the blood volume pulse vector which resulted in the pulse signal with the highest quality indicator value and/or from the derived pulse signal itself.
2. Physiological parameter sensing system (50) according to claim 1, wherein the sensing interface (52) comprises a photoplethysmography, PPG, sensing module.
3. Physiological parameter sensing system (50) according to claim 1 or 2, wherein the heart rate sensing module (54) comprises an ECG sensing module and/or an accelerometer.
4. Physiological parameter sensing system (50) according to any preceding claim, wherein the heart rate sensing module comprises a photoplethysmography, PPG, sensing module.
5. Physiological parameter sensing system (50) according to claim 2 or 4, wherein the heart rate sensing module is integrated with the sensing interface and both the heart rate sensing module and the sensing interface comprise the same photoplethysmography, PPG, sensing module.
6. Physiological parameter sensing system (50) according to any preceding claim, wherein the sensing interface (52) comprises a chest-mountable sensing unit, and preferably wherein the chest-mountable sensing unit comprises a PPG sensor and the heart rate sensing module (54).
7. Physiological parameter sensing system (50) according to any preceding claim, wherein the physiological parameter is blood oxygen saturation, SpO2And the system is an oxygen saturation sensing system (50).
8. Physiological parameter sensing system (50) according to any preceding claim, wherein the quality indicator value for each derived pulse signal is derived based on values of one or more frequency components of the pulse signal corresponding to frequency components of the heart rate signal.
9. Physiological parameter sensing system (50) according to any preceding claim, wherein the quality indicator value for each pulse signal is considered to be the value of the highest maximum in the frequency spectrum for the pulse signal.
10. The physiological parameter sensing system (50) according to claim 9, wherein deriving the quality indicator value comprises: prior to determining the quality indicator, enhancing one or more frequency components corresponding to frequency components of the heart rate signal in each pulse signal.
11. Physiological parameter sensing system (50) according to claim 9 or 10, wherein deriving the quality indicator value for each pulse signal comprises: suppressing or eliminating frequency components of the pulse signal that do not correspond to frequency components of the heart rate signal prior to determining the quality indicator.
12. Physiological parameter sensing system (50) according to any of claims 1-7, wherein deriving the quality indicator value for each derived pulse signal is based on determining a strength of correlation between the pulse signal and the heart rate signal or a signal derived from the heart rate signal.
13. The physiological parameter sensing system (50) according to claim 12, wherein deriving the quality indicator value for each pulse signal comprises: applying a Hilbert transform to the pulse signal to derive an analysis signal, and subsequently deriving a strength of correlation between the analysis signal and the heart rate signal for the subject.
14. A physiological parameter sensing method, comprising:
obtaining at least two detection signals derived from detected electromagnetic radiation reflected from or transmitted through a skin area of the subject's body;
obtaining a heart rate signal for the subject;
deriving at least two pulse signals, each pulse signal being formed from a weighted combination of the detection signals, wherein the weighting for each pulse signal is based on components of a different one of a set of at least two blood volume pulse vectors;
deriving a quality indicator value for each derived pulse signal, the quality indicator value being based on a characteristic of a derived relationship between the pulse signal and the heart rate signal of the subject; and is
Deriving physiological information indicative of at least one physiological parameter from the blood volume pulse vector which resulted in the pulse signal with the highest quality indicator value and/or from the derived pulse signal itself.
15. A computer program product comprising computer program code which, when run on a computer, is adapted to cause the computer to perform the method according to claim 14.
Technical Field
The present invention relates to a system and method for sensing a physiological parameter of a subject, in particular a system and method utilizing reflection of electromagnetic radiation from or transmission of electromagnetic radiation through a skin area of a subject's body.
Background
A physiological parameter of the person (e.g. Heart Rate (HR), Respiratory Rate (RR) or arterial oxygen saturation (SpO)2) As an indicator of the current state of the person and as a strong predictor of serious medical events. For this reason, physiological parameters are widely monitored in hospitalized and outpatient care settings, at home or in additional health, leisure and fitness settings.
One method of measuring physiological parameters is plethysmography. Plethysmography generally refers to the measurement of volume changes of an organ or body part, in particular to the detection of volume changes due to a cardiovascular pulse wave travelling through the body of a subject with each heartbeat.
Photoplethysmography (PPG) is an optical measurement technique that evaluates the time-varying changes in the light reflection or transmission of a region or volume of interest. PPG is based on the principle that blood absorbs more light than the surrounding tissue, and therefore the change in blood volume with each heartbeat affects transmission or reflection accordingly. In addition to information about heart rate, PPG waveforms can also include information attributable to other physiological phenomena (e.g., respiration). By evaluating the transmittance and/or reflectance at different wavelengths (typically red and infrared), the blood oxygen saturation can be determined.
A particularly important physiological parameter that can be measured using certain variants of PPG sensors is arterial oxygen saturation (SpO)2). In the present disclosure, SpO can be measured2Will be referred to as "SpO2A sensor ". This may be understood as referring to being able to measure SpO2Of the particular type of PPG sensor.
Arterial oxygen saturation is a vital parameter that needs to be continuously monitored in intensive care and operating rooms. It is also a useful parameter for patient monitoring in the general ward. SpO2The sensor may be placed on a finger. It is also possible to place the sensor at other locations (e.g. forehead, toe or earlobe) and to perform contactless monitoring.
SpO2The output of the sensor (i.e., oxygen saturation) is defined as:
wherein HbO2Is the concentration of hemoglobin bound to oxygen, and Hb is the concentration of hemoglobin not bound to oxygen.
SpO2The sensor (as a PPG sensor) optically measures the blood volume change; the light traveling through the skin (and possibly the underlying tissue) is detected, from which changes in blood volume can be determined. SpO2The sensor utilizes absorption of light at least two wavelengths. The amount of light detected depends on the light absorption. The absorption spectrum of deoxyhemoglobin is different from that of oxyhemoglobin, allowing the derivation of SpO2。
This situation is illustrated in fig. 1. FIG. 1(A) shows the molar extinction coefficients (y-axis; cm) of oxyhemoglobin and deoxyhemoglobin as a function of the wavelength (x-axis; nanometers) of the incident light-1·Mole-1).
The ripple of the sensor signal is defined as the time-varying AC portion of the signal divided by the constant (or slowly varying) DC component of the signal. The light absorption of (skin) tissue and venous blood contributes to the constant (or slowly varying) "DC" component of the detected light signal, while the pulsations in arterial blood contribute to the portion of the signal that varies with the cardiac cycle ("AC" component). SpO2The sensor utilizes pulses of at least two wavelengths of light.
This method is illustrated in fig. 2, fig. 2 showing the method in PPG (SpO)2) The light signal detected at the sensor for each of the red light source (line 28) and the infrared light source (line 30).
In the reflective mode, the ripple as a function of wavelength (see fig. 1(B)) is determined not only by the molar extinction coefficient (and possibly specular reflection), but also by the penetration depth of the wavelength. For example, blue light penetrates the skin only very shallowly, and therefore reaches only relatively few blood vessels with pulsating blood. In contrast, red light penetrates deeper.
In conventional SpO2In sensors, the so-called "ratio of ratios" R is used to calculate SpO2The value is obtained. The ratio R of the ratios is the relative (AC/DC) ripple of one wavelength of the sensor light divided by the relative ripple of the second wavelength:
usually, red light is usedAnd infrared light (840nm or 940 nm). SpO2Calculated according to an empirically determined formula, e.g. SpO2110-25R (%) (seeDepending on the wavelength used).
One known method is to determine R beat-by-beat using the maxima and minima in a single pulse cycle to derive the AC component amplitude and the DC component amplitude (e.g., based on an average of the maxima and minima). Further alternatives include e.g. averaging different light signals over multiple heart beats or processing the signals after fourier transformation.
Recently, a new approach has been proposed to derive a measure of a physiological parameter using the PPG signal. This is known as the "adaptive PBV method" (APBV) and is described in detail in WO 2017/055218.
In this method, a so-called blood volume pulse vector ("P") is usedbvVector ") of the two or three different wavelength components, the components of which represent possible relative pulsatility of the optical signal of the two or three different wavelength components, as a basis for extracting the pulse signal from the measured optical signal. In particular, a plurality of pulse signals may be derived using a weighting related to one of a set of blood volume pulse vectors, wherein each pulse signal is formed as a combination (e.g., a linear combination) of the sensed optical signals of different wavelengths.
Then, for each generated PbvThe vector derives a quality index value.
Then, P is selected to generate the best quality pulse signalbvVector (e.g., pulse signal characterized by the highest peak in the normalized power spectrum) and use this PbvThe vector is used to derive the arterial oxygenation. This is because the presence of the highest peak is an indication of the highest signal-to-noise ratio (SNR), i.e., the highest quality signal.
Can be assisted by, for example, adding PbvVector sum SpO2Predetermined look-up tables for value association or by allowing P-basedbvVector to determine SpO2According to the identified highest quality PbvVector derived SpO2. The calibration equation may employ, for example, SpO2＝100–C1Form of x k, wherein C1Is from SpO2The value of the detection signal received by the sensor, and kIs the sum of P which produces the highest quality signalbvThe value associated with the vector (e.g., P listed in a table or listbvThe value of the vector).
Essentially, the method is an indirect means of identifying the blood volume pulse vector that most closely matches the "true" blood volume pulse vector that characterizes the pulse of a subject without any noise.
The method can robustly resist artifacts in signals with different relative intensities in the wavelength channels, e.g. the contact of the sensor with the skin of the subject is changing (for a contact sensor) or the body part is moving close to or away from the camera (for remote monitoring).
To accurately determine the SpO2It is important to select the "best" blood volume pulse vector PbvI.e. the blood volume pulse vector P which produces the highest quality signalbv. For this reason, for P according to differencebvEach of the vector-derived pulse signals determines a quality indicator value.
A simple and advantageous quality indicator is the height value of the highest peak in the frequency spectrum for each pulse signal.
Each extracted pulse signal (using a particular P)bvVector) can be normalized in the fourier domain so that the sum of the spectral energies equals 1. Thereafter, the pulse signal exhibiting the highest peak in the frequency plot is identified as the pulse signal of the highest quality. This is illustrated in fig. 3, which fig. 3 shows in (a) and (B) two pulse signals derived from different respective blood volume pulse vectors, each signal being transformed into the frequency domain. The x-axis represents frequency and the y-axis represents the magnitude of the frequency component.
It can be seen that the pulse signal (a) is characterized by a highest peak 42. Thus, in this simple example, the signal (a) will be selected as the highest quality signal, and the physiological information is therefore derived using the blood volume pulse vector from which the pulse signal (a) was derived.
Although using this quality criterion generally results in good motion robustness, it is possible for a moving patient or a patient with a strict center of gravityProblems may arise for the patient with irregularities. For such patients, motion artifacts in the generated signal remain a significant problem. In addition, if SpO is used2The sensor is placed on the chest and the signal may contain a significant component corresponding to breathing. These respiratory fluctuations are often misinterpreted as contributions to the pulse signal, resulting in a contribution to the SpO2Is measured.
Improved methods for deriving physiological parameters are sought that can utilize the APBV methods outlined above, but that can provide improved accuracy and improved robustness against motion artifacts.
Disclosure of Invention
The invention is defined by the claims.
According to an aspect of the present invention, there is provided a physiological parameter sensing system comprising:
a sensing interface adapted to obtain at least two detection signals derived from detected electromagnetic radiation reflected from or transmitted through a skin area of a subject's body;
a heart rate sensing module; and
a processor operatively coupled with the sensing interface and the heart rate sensing module and adapted to:
deriving at least two pulse signals, each pulse signal being formed from a weighted combination of the detection signals, wherein the weighting for each pulse signal is based on components of a different one of a set of at least two blood volume pulse vectors;
deriving, for each derived pulse signal, a quality indicator value based on a characteristic of a derived relationship between the pulse signal and a heart rate signal of the subject, the heart rate signal being sensed by the heart rate sensing module; and
deriving physiological information indicative of at least one physiological parameter from the blood volume pulse vector which resulted in the pulse signal with the highest quality indicator value and/or from the derived pulse signal itself.
The invention is based on the advantageous integration of sensor information from an auxiliary heart rate sensing module into a physiological parameter sensing method for improved physiological parameter sensing. In particular, a more robust quality assessment of the plethysmography vector can be performed, thereby enabling a more reliable selection of the plethysmography vector that yields the most accurate determination of the physiological parameter information.
The general idea regarding determining the quality indicator value may be understood as determining the extent to which each derived pulse signal corresponds to or contains a component of the heart rate of the subject as represented by the heart rate signal derived from the heart rate sensing module. Greater correspondence may indicate a greater signal-to-noise ratio, which may indicate less distortion caused by motion artifacts.
The characteristic of the derived relationship may refer to the strength of the relationship between the derived pulse signal and the measured heart rate signal.
The blood volume pulse vector may be otherwise referred to as a blood volume vector or a feature vector.
Blood volume pulse vector (' P)bvVectors ") are similar to those mentioned above with respect to the APBV method. The components of these vectors represent the possible relative pulsatility of the optical signal at two or three different wavelength components. In particular, the method involves deriving a plurality of pulse signals using a weighting related to one of a set of blood volume pulse vectors, wherein each pulse signal is formed as a combination (e.g., a linear combination) of detection signals. Thus, in this context, the vector is not a vector in a spatial coordinate system, but a vector in a self-constructed "blood volume pulse" space. The volume pulse vector may be a predetermined set of vectors, or the volume pulse vector may be constructed by the processor in real-time. This will be described in more detail below.
The heart rate sensing module can sense the heart rate or the pulse rate without affecting the efficacy of the method. For simplicity and brevity, in the following disclosure, embodiments will be described with reference to using a "heart rate signal" derived from a heart rate sensing module. In all cases, however, the pulse rate signal may be measured and used instead.
Furthermore, for the following disclosure, the term "heart rate signal" should be understood to refer to a signal derived from a heart rate sensing module, i.e. a direct measurement of the "true" heart rate or pulse rate of a subject. In contrast, unless otherwise specified, "pulse signal" is understood to mean a derived pulse signal derived based on one of a set of blood volume pulse vectors according to the method of the present invention.
The term "reflect from … …" should be understood to cover the situation where light is scattered from an area of the skin of a subject (e.g., from blood within such an area of skin) in addition to being reflected from a surface or substance in a classical sense.
Each blood volume pulse vector may contain n components. The n components of each blood volume pulse vector represent possible relative pulsations in each of n detection signals Cn, which may e.g. be derived from different frequency (or color) channels of the sensing interface, each frequency (or color) channel sensing a specific part of the spectrum of reflected or transmitted electromagnetic radiation.
Relative pulsatility generally refers to the relative intensity of the blood pulse in each of the three detection signal channels, i.e., the effective relative intensity in each of the three detection signals.
For a given subject, there is typically a relatively stable blood volume pulse vector P when making measurements, without motion or other artifacts or distortionsbvThe relatively stable blood volume pulse vector PbvThe intensity of each detection signal in the set of n detection signals is accurately represented, e.g. representing the relative pulsatility in each of the n color channels. From which physiological parameter information can be derived, e.g. SpO2Pulse rate, CO and CO2。
However, motion of the subject, physical displacement of the sensing interface, or other disturbances can cause distortion of the obtained detection signal, resulting in deviations from the "true" non-distorted (characteristic) blood volume pulse vector for the subject. Some of the detection signals will increase, for example, in relative intensity, resulting in a distortion of the ratio between the individual channel intensity values. However, these changes in intensity are due to noise in the signal(s), rather than a true reflection of changes in the underlying physiological parameter. The problem is how to identify the changes in the color channel signal caused by noise as well as the changes caused by real physiological changes.
So-called adaptive P developed by the inventionbvThe solution of the (APBV) method is: effectively constructing a plurality of prospective pulse signals, each prospective pulse signal being formed from a weighted combination of the detection signals; each pulse signal is then tested using a quality assessment to determine which pulse signal reaches the maximum "quality index value". The quality indicator value may for example directly or indirectly indicate the signal to noise ratio.
For each generated prospective pulse signal, the detection signals may be weighted based on the relative magnitude of the components of one prospective plethysmographic vector of the set of prospective plethysmographic vectors. In this way, a set of potential pulse signals is generated, each potential pulse signal being associated with one of a different set of blood volume pulse vectors. The derived pulse signal with the highest quality indicator value indicates: the blood volume pulse vector on which the pulse signal is based is the "true" (non-distorted) blood volume pulse vector that most accurately represents for the subject without noise and distortion that may affect the original detected signal.
The innovation of the present invention is that additional heart rate sensor data is advantageously incorporated into the quality assessment to enable a more (motion) robust and thus more reliable quality assessment of various prospective pulse signals. According to the method, each derived pulse signal is evaluated with respect to the physiological parameters acquired in real time, thereby improving accuracy and robustness.
More specifically, for each pulse signal, a quality indicator is derived based on characteristics of a derived relationship between the measured heart rate signal and each derived pulse signal.
The characteristic of the derived relationship may refer to the strength of the relationship between the pulse signal and the heart rate signal.
The characteristic of the derived relationship may refer to a degree of correlation between the pulse signal and the heart rate signal.
The characteristic of the derived relationship may refer to the strength of the frequency component(s) in the pulse signal that correspond to the frequency components of the heart rate signal of the subject.
As indicated above, a greater correspondence with the measured heart rate signal may indicate a greater signal-to-noise ratio of the pulse signal (relative to the physiological pulse), which may indicate less distortion due to motion artifacts, and thus may more accurately represent the true, undistorted pulse signal.
The sensing interface may include a photoplethysmography (PPG) sensing module. The sensing interface may be a PPG sensor.
The sensing interface may comprise a chest-mountable sensing unit. This may be, for example, a chest patch. The chest-mountable sensing unit may incorporate or comprise a PPG sensor. In an example, the chest-mountable sensing unit may also conveniently comprise or incorporate a heart rate sensing module. This is convenient both for the subject (who does not need additional sensing elements to be coupled to them) and the clinician (who only needs to attach or mount one sensing unit to the subject).
In an example, the heart rate sensing module may be a separate (separate from the sensing interface) heart rate sensing element or device, or may be included by or integrated with the sensing interface.
According to a set of embodiments, the heart rate sensing module may comprise an ECG sensing module and/or an accelerometer.
The ECG sensing module refers to an electrocardiogram module or device. The ECG sensing module may comprise one or more sensors adapted to be applied to the skin for detecting an electrical signal generated by the heart of the subject at each heartbeat. In this way, the heart rate can be derived.
ECG has the advantage of high motion robustness.
Accelerometers provide a very convenient way of measuring heart rate and are based on the use of motion data (e.g. vibrations caused by the heartbeat). In an example, motion-based heart rate sensing may be advantageously applied to patients with severe arrhythmias. Here, the object of the invention is less directed to the improvement of the motion robustness, but more directed to the improvement of the performance of patients suffering from such arrhythmias.
Determining the highest quality pulse signal is particularly difficult for patients with severe arrhythmias. This is because the standard method of identifying the quality indicator is simply to take the height of the highest peak of the spectrum of each pulse signal as the value of the quality indicator. However, for patients with severe arrhythmias, the frequency spectrum of each pulse signal is typically very spread out, making it difficult to identify a single "highest" peak or at least making an indicator of signal quality unreliable.
Thus, in these cases, the accuracy of the results is improved by using an auxiliary heart rate sensing module (as is the case in the present invention) which allows determining the true pulse frequency, thereby enabling a more reliable determination of the quality indicator as a peak corresponding to the specific heart rate frequency(s).
According to one or more embodiments, the heart rate sensing module may comprise a photoplethysmography, PPG, sensing module. Measuring the heart rate signal using a PPG sensor is a procedure well known in the art.
In these cases, in a particular example, the heart rate sensing module may be integrated with the sensing interface, and wherein the heart rate sensing module and the sensing interface both comprise the same photoplethysmography, PPG, sensing module. In these cases, the heart rate sensing module is comprised by the sensing interface, and the same PPG sensor facilitates the functionality of both the sensing interface and the heart rate sensing module.
In these examples, the PPG sensor may be adapted to irradiate and sense electromagnetic radiation of a plurality of wavelengths, and wherein one subset of the wavelengths is used to obtain the at least two detection signals and a further subset of the wavelengths is used to obtain the heart rate signal. In a particular example, the two subsets of wavelengths may overlap to some extent.
In a particular example, the heart rate sensing module is integrated with the sensing interface, and both the heart rate sensing module and the sensing interface may comprise the same photoplethysmography, PPG, sensing module.
"integrated with … …" may mean, for example: the heart rate sensing module is comprised by the sensing interface being the same item as the sensing interface, and/or the heart rate sensing module and the sensing interface are facilitated by the same device or component (i.e. in this case the same PPG sensing module).
The system may be used to sense a blood analyte concentration parameter.
The physiological parameter sensed by the system may be blood oxygen saturation SpO2. In this case, the system may be an oxygen saturation sensing system. SpO2Is one example of a blood analyte concentration parameter.
In this case, as discussed above, the indicator physiological parameter (i.e., SpO) is derived2) The physiological information of (a) may include using PbvVector sum SpO2A predetermined look-up table with associated values. Alternatively, a calibration equation may be used, allowing for P-basedbvVector to determine SpO2. The calibration equation may, for example, employ SpO2＝100–C1Form of x k, wherein C1Is from SpO2The value of the detection signal received by the sensor, and k is the sum of P producing the highest quality signalbvThe value associated with the vector (e.g., P listed in a table or listbvThe value of the vector).
However, other physiological parameters may also be derived, such as the concentration of carboxyhemoglobin, methemoglobin and bilirubin. These are also examples of blood analyte concentration parameters.
In each case, the physiological parameter is preferably derived from the determined blood volume pulse vector. Can lead out SpO2This operation is performed in a similar manner. However, the pulse signal with the best quality index valueItself also physiological parameters of interest which can be processed and output by the system according to the invention. Still further, a pulse rate, a heartbeat interval or a heart rate variability can be derived from the pulse signal. At SpO2Under varying conditions, the pulse signal may be more robust than a pulse signal obtained with a fixed blood volume pulse vector.
According to an example, the set of blood volume pulse vectors may be predetermined.
For example, the system (e.g., a controller of the system) may include a memory, and the blood volume pulse vector may be pre-stored on the memory. In other examples, the blood volume pulse vector may be stored remotely and accessed via an established communication channel with a remote data storage device.
In this case, the processor may be configured to use a set of fixed volume pulse vectors, i.e. each volume pulse vector corresponds to a discrete physiological parameter value, e.g. SpO2The value is obtained. For example, a fixed set of vectors may represent SpO in the range of 60% to 100%2Values, where for example 10 different vectors cover the range.
According to any embodiment, the processor may be adapted to generate output information indicative of or representing the derived physiological information, the output information being indicative of the at least one physiological parameter. The processor may also be adapted to transmit the generated output information to a computer, processor or data storage device, for example, either locally or remotely, via a communications channel.
According to an example, the quality indicator value for each derived pulse signal may relate to the strength of one or more frequency components of the heart rate signal within the frequency spectrum of the pulse signal.
Intensity is intended to be broadly interpreted to mean any measure indicative of the extent to which frequency components are present in a pulse signal.
The quality indicator value for each derived pulse signal may be indicative of a signal-to-noise ratio of the respective pulse signal relative to a real heart rate of the patient. This gives a good indication of the quality of the blood volume pulse vector used to derive the physiological parameters of the patient based on the particular optical signal readings given by the patient.
In an example, a quality indicator value for each derived pulse signal may be derived based on values of one or more frequency components of the pulse signal corresponding to frequency components of the heart rate signal.
A value means, for example, the magnitude, amplitude or height of a frequency component (rather than, for example, the frequency it represents).
For example, each pulse signal may be transformed into the frequency domain, e.g. by a fourier transform. The value of the relevant frequency component can then be determined.
In an example, the quality indicator value may be considered as the height of a spectral peak substantially or just at the measured heart rate frequency, i.e. the value of the signal when transformed in the frequency domain just or substantially corresponds to the measured heart rate frequency (or e.g. the frequency at which the heart rate spectrum was obtained). In this case, other frequency components may be ignored.
In an example, the heart rate signal may include only one frequency component, or may include multiple components. In an example, only the strongest (i.e., largest) frequency component of the heart rate signal may be selected, and only that component identified and selected within the pulse signal spectrum.
In an example, the quality indicator value for each pulse signal may be considered to be the highest maximum value in the spectrum for that pulse signal, i.e. the value or height of the highest peak in the spectrum for that pulse signal. In this context, a value means an amplitude or "peak height".
The method is based on the following assumptions: the highest peak in the frequency spectrum of each pulse signal corresponds to or is caused by the physiological blood pulse of the subject. The value or height of the frequency component (i.e. its strength in the frequency domain) gives an indication of the signal-to-noise ratio. Therefore, by using the height of the highest peak as the quality index value, an index representing the SNR of each pulse signal corresponding to the physiological pulse of the subject can be selected.
According to one or more examples of the method, deriving the quality indicator value may comprise: prior to determining the quality indicator, enhancing one or more frequency components corresponding to frequency components of the heart rate signal in each pulse signal. Then, the height of the highest peak in the spectrum of the pulse signal is selected as a quality index value for each pulse signal.
This may better ensure that the highest peak does correspond to the subject's pulse, since one or more frequency components matching the heart rate signal may emphasize one or more components.
Additionally or alternatively, in an example of the method, deriving the quality indicator value for each pulse signal may comprise: suppressing or eliminating frequency components of the pulse signal that do not correspond to frequency components of the heart rate signal prior to determining the quality indicator. Then, the height of the highest peak in the spectrum of the pulse signal is selected as a quality index value for each pulse signal.
This may better ensure that the highest peak does correspond to the subject's pulse, as components that do not match components known to be present in the heart rate signal will be reduced or removed.
According to an alternative set of embodiments, deriving the quality indicator value for each derived pulse signal may be based on determining a strength of a correlation between the pulse signal and the heart rate signal or a signal derived from the heart rate signal.
This represents an alternative way of determining the strength of the relationship between the pulse signal and the heart rate signal for each derived pulse signal of the subject. This gives a way to evaluate the signal-to-noise ratio of the pulse signal with respect to the true pulse of the subject. The stronger the correlation, the greater the strength of the signal component(s) in the derived pulse signal that correspond to the subject's pulse.
The "strength" of the correlation generally means the degree of the correlation, i.e. the value of a correlation coefficient determined, for example, between the respective pulse signal and the heart rate signal.
In an example of the method, deriving the quality indicator value for each pulse signal comprises: applying a Hilbert transform to the pulse signal to derive an analysis signal, and subsequently deriving a strength of correlation between the analysis signal and the heart rate signal for the subject.
When the pulse signal itself is used to derive the correlation, the correlation may be sensitive to the phase shift between the heart rate signal sensed by the heart rate sensing module and the PPG-derived pulse signal. By alternatively using the derived analysis signal, a correlation can be determined that is independent of (or not affected by) the phase shift between any of the above two. In particular, the method allows both the magnitude of the correlation and the phase difference to be derived, wherein the phase difference can then be ignored.
According to one or more sets of embodiments, the processor may be adapted to perform a filtering step in which the obtained detection signals are adaptively filtered to reduce the contribution of the signal components caused by the motion, prior to deriving the at least two pulse signals.
In an example, this may include filtering based on a motion signal obtained from a motion sensor coupled to the object. In particular, the detection signal may be filtered to reduce or eliminate frequency components identified in the frequency spectrum of the motion signal. In other examples, the filtering may be based on a heart rate signal derived from a heart rate sensing module. In this case, the detection signal may be filtered to reduce or eliminate frequency components that are not identified in the heart rate signal.
This reduces the potential contribution of subject motion in deriving physiological parameter information and thereby reduces potential inaccuracies in the measurement results.
In a specific example, a band pass filter may be used, wherein the passed frequency range is selected based on detected frequency components in the spectrum of the heart rate signal derived by using the heart rate sensing module.
Examples according to further aspects of the invention provide a physiological parameter sensing method comprising:
obtaining at least two detection signals derived from detected electromagnetic radiation reflected from or transmitted through a skin area of the subject's body;
obtaining a heart rate signal for the subject;
deriving at least two pulse signals, each pulse signal being formed from a weighted combination of the detection signals, wherein the weighting for each pulse signal is based on components of a different one of a set of at least two blood volume pulse vectors;
deriving a quality indicator value for each derived pulse signal, the quality indicator value being based on a characteristic of a derived relationship between the pulse signal and the heart rate signal of the subject; and is
Deriving physiological information indicative of at least one physiological parameter from the blood volume pulse vector which resulted in the pulse signal with the highest quality indicator value and/or from the derived pulse signal itself.
An example according to a further aspect of the present invention provides a computer program product comprising computer program code adapted to, when run on a computer, cause the computer to perform any of the embodiments of the physiological parameter sensing method outlined above.
Drawings
Preferred embodiments of the present invention will now be described in detail, by way of example only, with reference to the accompanying drawings, in which:
FIG. 1 shows the molar extinction coefficient and pulsation for partially oxygenated blood and fully oxygenated blood;
FIG. 2 schematically illustrates SpO2An AC component and a DC component of the sensor signal;
fig. 3 shows the frequency spectra of two sample pulse signals derived from different blood volume vectors;
FIG. 4 schematically depicts a sensing system according to one or more embodiments of the invention;
FIG. 5 schematically illustrates an example breast patch that integrates a sensing interface with a heart rate sensing module that may be used in accordance with one or more embodiments of the invention;
fig. 6 illustrates a filter circuit for removing noise from an obtained detection signal based on a reference signal for noise; and is
Fig. 7 illustrates a filter circuit for removing noise from an obtained detection signal based on a reference pulse signal.
Detailed Description
Physiological parameter sensing systems and methods are provided in which information indicative of at least one physiological parameter is derived. The inventive method is based on constructing a plurality of pulse signals from differently weighted combinations of at least two detection signals derived from detected electromagnetic radiation directed onto or through a skin area of the subject. The weighting is based on different blood volume pulse vectors in the various sets of blood volume pulse vectors. Deriving a quality indicator value for each generated pulse signal, wherein this is based on a derived relationship between the obtained heart rate signal and the pulse signal for the patient. The physiological parameter information is derived using the volume pulse vector that resulted in the pulse signal with the highest quality index value and/or from the derived pulse signal itself.
FIG. 4 shows a block diagram of an example physiological parameter sensing system in accordance with one or more embodiments of the invention. The
Different detection signals Cn correspond to different wavelengths of radiation. In some examples, two or three detection signals may be obtained for two or three wavelengths of light. Advantageously, these wavelengths may be (selected from) for example 660nm, 810nm and 940 nm.
The
In a preferred example, the heart rate sensing module comprises an ECG sensing device or a motion sensing device (e.g., an accelerometer).
The ECG sensing device may include electrodes or other sensors for applying or attaching to the body of a subject to sense electrical signals generated by the heart. These signals allow detecting the beating of the heart and thus the heart rate.
In the case of an accelerometer, the heart rate may be detected based on motion sensing (e.g., sensing vibrations around the chest area of the patient or sensing movement of the chest caused by the heart beating). More details of motion-based heart rate sensing can be found in document US 9510775.
Alternatively, other heart rate sensing modules may be used, for example, ultrasound-based sensing modules and inductive sensing modules.
According to some examples, the heart rate sensing module may include or be facilitated by a PPG sensor. Sensing heart rate (or pulse rate) using a PPG sensor is a well-known procedure and has been discussed above.
In one set of advantageous embodiments, at least one of the
An example is schematically illustrated in fig. 5, which shows a subject 62 wearing a breast patch 64, the breast patch 64 incorporating by way of example an integrated PPG sensor and ECG sensor. Additionally or alternatively, in an example, the patch may incorporate one or more of: thermometers and motion sensors (e.g., accelerometers). In an example, a motion sensor may be used to measure respiration rate and/or background information (e.g., body posture) and whether the subject is walking.
A chest-mountable sensor device is particularly useful for patients in e.g. a general ward of a hospital or in a professional care facility. A breast patch based sensing unit as shown in fig. 5 is convenient for the subject, as he or she can keep moving sufficiently without stopping the collection of physiological data.
Although in the examples of fig. 4 and 5 the sensing interface comprises a PPG sensor, any sensing device operable to apply electromagnetic radiation to a skin region of the subject's body and detect reflections of radiation from or transmissions of radiation through the subject's body may be used.
In general, any combination of electromagnetic radiation emitters (e.g., LEDs) and electromagnetic radiation sensors (e.g., photodiodes or sensor arrays) may be used. A plurality of different electromagnetic radiation emitters (emitting different wavelengths) may be used. These electromagnetic radiation emitters may be used, for example, in a time-multiplexed or frequency-multiplexed manner. In other examples, a broadband emitter may be combined with a sensor array having optical filters applied to individual cells to simultaneously sense radiation reflected from or transmitted through the skin.
Although in the examples outlined above, a
In case of a moving subject, there is an increased risk of having motion artifacts in the derived physiological parameter information.
The present invention aims to provide a more robust measurement method to derive physiological parameter information.
As discussed above, the present invention is a development of previous methods for deriving physiological information indicative of a physiological parameter. This method is described in detail in WO 2017/055218 and will be referred to as the "adaptive PBV" method (APBV method).
Embodiments of the present invention improve the robustness of the APBV method by exploiting additional sensor data related to the subject's cardiac activity (i.e., by the heart rate sensing module 54).
The processor is configured to process the signal data from the
For a better understanding of the present invention, the principles of the known APBV method will now be briefly discussed. More detailed information can be found in WO 2017/055218, "Improved movement debug of remote-PPG by using the block volume pulse signature" (Physiol.Meas., Vol.35, 1913) of G.de Haan and A.van Leest and "New print for measuring specific block generation, initialization-debug removal monitoring" (No. 6 of science report: 38609, 2016).
Commonly used to measure physiological parameters (e.g. blood oxygen saturation, SpO)2) The method of (3) uses the ratio of the PPG sensor signal amplitudes of red light and infrared light (IR). The PPG signal, in particular for red light, is very small. With this known method, the amplitude of the red PPG signal would be overestimated, since it includes noise. In contrast, the APBV method does not make physiological parameter measurements directly based on sensed light frequency amplitudes, but rather identifies a pulse blood volume ("signature") vector (explained below) that produces a derived pulse signal exhibiting the best signal-to-noise ratio (SNR).
The PPG signal is caused by changes in blood volume in the skin. In particular, the beating of the heart causes pressure changes in the arteries, as the heart pumps blood against the resistance of the vascular bed. Since the artery is elastic, the diameter of the artery changes synchronously with the pressure changes, thereby increasing the local volume of blood present in a given area. These diameter changes occur even in the smaller skin vessels where the volume changes cause the light absorption to change.
These variations cause characteristic pulsating "signatures" in the relative intensities of the different spectral components of the reflected/transmitted light. This feature results from the comparison between the absorption spectrum of blood and the absorption spectrum of bloodless skin tissue.
If the detector(s) is (are for example,camera or sensor) has a discrete number of color channels Cn, each sensing a particular portion of the spectrum, then can be mathematically expressed as a "feature vector" (also referred to as a "normalized blood volume vector" P)bv) In the form of relative pulsations of these channels (as indicated by signal strength). It has been shown in "Improved motion robustness of remote-PPG by using the blood volume pulse signal nature" of g.de Haan and a.van Leest that motion robust pulse signal extraction can be performed based on color channels and feature vectors if the feature vectors are known.
Is defined as PbvNormalized blood volume pulse vector (also referred to as feature vector) indicates the relative PPG intensities in the red, green and blue optical signal channels, i.e.:
where σ denotes the standard deviation.
To demonstrate this concept, a remote PPG sensor including a camera was utilized in the test to derive a sample blood volume pulse vector.
In order to first quantify the expected PbvVector components measuring the signal responses H of the red, green and blue channels of a global shutter color CCD camera for obtaining optical detection signals Cn, respectivelyred(w)、Hgreen(w) and Hblue(w) which are a function of the wavelength w. The model takes into account the skin reflection ρ of the subjects(w) is carried out. These values are based on the absolute PPG amplitude curve PPG (w).
From these curves, the theoretical blood volume pulse vector P is shown in FIG. 2 of the de Haan and van Leest article cited abovebvThe calculation formula of (2) is as follows:
which is normalized by using white halogen illumination spectrum I (w)bv＝[0.27，0.80，0.54]. When a noisier curve is used, the result may be Pbv＝[0.29，0.81，0.50]。
These values are for the SpO known for them2Values of 98 +/-2% were obtained for healthy subjects.
The blood volume pulse predicted by the model used corresponds reasonably well to the SpO known for it under white lighting conditions2Experimentally measured normalized blood volume pulse vector P found after averaging measurements on the same plurality of healthy subjectsbv＝[0.33，0.78，0.53]. Given this result, the following conclusions can be drawn: the observed PPG amplitude (in particular in the red channel, to a lesser extent in the blue channel) can be explained by crosstalk in the wavelength interval between 500nm and 600 nm. As the model shows, the exact blood volume pulse vector depends on the color filter of the camera or optical sensor, the spectrum of the light and the skin reflection. In practice, the vector will be very stable for a given set of wavelength channels (the vector will differ in the infrared compared to an RGB-based vector).
It was further found that under white illumination, the relative reflection of the skin in the red, green and blue channels is not strongly related to the skin type.
Therefore, the following conclusions can be drawn: for a given constant SpO2Normalizing the blood volume pulse vector P under constant (e.g. white) illuminationbvIs very stable.
Blood volume pulse vector PbvMakes it thus effective for distinguishing color changes in the acquired signal caused by blood volume changes from changes due to alternative causes such as motion of the object or movement of the sensor, i.e., PbvThe vector can be used as a "feature" of blood volume changes to distinguish such changes from color changes of other causes.
Known relative pulsing P of color channelsbvAnd can therefore be used to discriminate between components of the derived pulse signal that represent the true physiological pulse signal and components caused by distortion.
According to the APBV method, each derived pulse signal S can be written in an example as a linear combination (or another kind of "mix") of the respective DC-free normalized color channels:
S＝W Cn
therein, WWT1 and wherein each of the three rows of the 3 × N matrix Cn contains N red, green and blue channel signals R without DC normalizationn、GnAnd BnRespectively, they are:
the operator μ corresponds to the mean value.
The APBV method uses the blood volume pulse vector to obtain the blending coefficient W, as described in US 2013/271591 a1 and the de Haan and van Leest article cited above. If band-pass filtered versions of R are usedn、GnAnd BnThe best results are obtained. According to this method, P is knownbvThe "direction" (in the detection signal channel space) of (a) can be used to distinguish between pulse signals and distortion.
The pulse signal is derived as a linear combination of normalized color signals. Since the relative amplitudes of the pulse signals in the red, green and blue channels are known to be represented by PbvGiven, therefore searching for the weight W given to the pulse signal SPBVWith a color channel Rn、GnAnd BnIs equal to Pbv. In other words, a weighted combination of at least two detection signals is sought using the selected weights such that the resulting pulse signal is combined with the original detection signal Cn and the pulse blood volume (P)bv) And (5) vector correlation.
Thus, the weight of the blend is determined by the following equation:
wherein the scalar k is determined such that WPBVHas a unit length. The following conclusions can be drawn: such as in normalizing the blood volume pulse PbvThe characteristic wavelength dependence of the PPG signal reflected in (1) can be used to estimate the pulse signal from the time-sequential RGB pixel data averaged over the skin area.
Thus, as explained above, the pulse signal (S1, S2) may be derived as a weighted sum of at least two detection signals Cn.
However, in order to obtain a "correct" (i.e. noise-free) pulse signal, it is important that the features used to create the pulse signal must be "correct", i.e. accurately representative of the true blood volume changes, otherwise the applied weighting would mix noise into the generated pulse signal.
To this end, in the APBV method, a quality index value is calculated for each of the derived pulse signals, and physiological information indicative of at least one physiological parameter is derived using only the feature vector resulting in the pulse signal having the best quality index value and/or from only the pulse signal (i.e. from only the pulse signal having the best quality index value).
It is known that if the detection signals are simply combined according to an arbitrary ratio, the resulting pulse signal will exhibit a relatively poor signal-to-noise ratio (SNR) (i.e., a poor quality indicator). Thus, in an advantageous embodiment, the blood volume pulse vector P giving the best SNR pulse signal (i.e. the best quality indicator) is selectedbvThe algorithm will most closely achieve the correct physiological parameter information (e.g., blood oxygen saturation SpO)2)。
Thus, the APBV method uses quality criteria to manipulate the optimal pulse blood volume vector PbvAnd (4) selecting a vector.
For example, the determination of the best quality P is illustrated above with reference to FIG. 3bvExample implementation of vectorsThe method is as follows.
The extracted pulse signal is transformed into the fourier (frequency) domain and the resulting spectrum is normalized (sum of spectral energies equals 1).
The height of the highest peak is used as the quality index value. For this method to work, it is necessary that the highest peak in the spectrum is indicative of the subject's pulse, rather than noise, e.g. caused by motion. Thus, while this method works well for static patients under controlled conditions, for moving patients, there is a motion that causes a significant noise component in the signal and thus PbvThe risk of distortion of the selection of the vector.
In the present invention, this problem is solved by improving the accuracy of the determined physiological parameter information by using heart rate information obtained from another (more motion robust) sensor.
According to one set of embodiments of the present invention, a spectrum of cardiac activity of a subject is measured using a heart rate sensing module 54 (e.g., an ECG sensor or accelerometer). Thereafter, frequency components identified within the obtained heart rate signal spectrum are enhanced or emphasized within the derived pulse signal spectrum before the highest spectral peak is identified. In this way, it can be better ensured that the highest peak within the spectrum actually corresponds to the subject's pulse. Therefore, the value (i.e., height) of the highest peak becomes a more robust quality indicator metric.
According to a further set of embodiments, the heart
According to one set of embodiments, the heart
For example, the quality indicator value may be considered to be the value of the highest peak within the fourier transformed pulse signal corresponding to the frequency component of the heart rate signal. Alternatively, the quality indicator value may be considered as the average height of all peak heights in the pulse signal spectrum corresponding to the frequency components of the heart rate signal.
According to an alternative set of embodiments, the quality indicator value may be based on a strength of a correlation between the pulse signal and a heart rate signal derived using a heart rate sensing device. In an alternative example, the quality indicator value may be based on the strength of the correlation between the respective spectra for the pulse signal and the heart rate signal. In other words, the quality indicator value may be based on the strength of the correlation between the pulse signal and the heart rate signal derived in the time domain or the frequency domain.
According to this set of embodiments, a correlation between each derived pulse signal and the obtained heart rate signal may first be derived or calculated, e.g. by deriving a standard pearson coefficient between the two signals. The quality indicator value for each pulse signal may then for example simply be taken as the derived correlation coefficient itself.
According to an advantageous set of examples, the analysis signal may first be derived from the heart rate signal using, for example, a hilbert transform. The quality indicator value for each pulse signal may then be based on the magnitude of the correlation between the pulse signal and the derived analysis signal. For example, the quality indicator may simply be taken as the magnitude of the correlation itself.
Once a quality index value has been derived for each derived pulse signal, these values are compared and the pulse signal with the highest quality index value is identified. Thereafter, the blood volume pulse vector P from which the highest quality pulse signal is derived is used as the basisbvOr derive physiological information indicative of at least one physiological parameter from the highest quality pulse signal itself.
According to a set of embodiments, the indication of blood oxygen saturation SpO is derived2The information of (1).
As described aboveIllustratively, P may be aided in examplesbvVector sum SpO2Value-associative predetermined look-up table to identify the highest quality PbvVector derived SpO2Or in the alternative by allowing for P-basedbvVector determination of SpO2To derive SpO2. The calibration equation may, for example, employ SpO2＝100–C1Form of x k, wherein C1Is from SpO2The value of the detection signal received by the sensor, and k is the sum of P producing the highest quality signalbvThe value associated with the vector (e.g., P listed in a table or listbvThe value of the vector).
Additionally or alternatively, the highest quality pulse signal and/or corresponding blood volume pulse vector P may be based onbvInformation indicative of other physiological parameters is derived. Physiological parameters that can be derived include, for example, carboxyhemoglobin (a combination of CO and hemoglobin), bilirubin, and methemoglobin.
Can be reacted with SpO2In a similar manner (by allowing P to be removedbvA dedicated predetermined look-up table in which the vectors are associated with physiological parameters or by means of a dedicated correlation equation) based on the identified highest quality PbvThe vector derives this information.
As discussed above, in the APBV method, the pulse signals are derived from a weighted combination of the detection signals Cn, each relating to a different wavelength of radiation (e.g., different color channels of detected light), wherein the weighting is based on PbvDifferent P in vector setbvThe components of the vector. The same method is applied in the present invention.
In particular, to calculate each pulse signal, the
For calculating different blood volume vectors P for each pulse signalbv1、Pbv2 provides the expected relative strengths of the pulse signals S1, S2 in the at least two detection signals Cn. The calculation of the pulse signals S1, S2 involves a weighted combination of at least two detection signals Cn using weights selected such that the resulting pulse signals S1, S2 are as corresponding feature vectors Pbvl、Pbv2 are related to the original detection signal Cn, e.g. by PbvThe scale indicated by the vector components is related to the original detection signal.
In more detail, the
Wherein k is selected such that
In a preferred example, different blood volume pulse vectors P are usedbv1、Pbv2. In an example, the vectors may be predetermined.
Blood volume pulse vector Pbv1、PbvThe fixed set of 2 covers the range of physiological parameters to be measured. For example, for SpO2The set of blood volume pulse vectors may correspond to a range of SpO between 60% and 100%2The value of (c). The vector is the same for each object and may, for example, beIs stored in a look-up table.
A given set of blood volume vectors typically involves the measurement of only one physiological parameter. In this example, SpO is being considered2. However, if another blood gas is to be determined, a different set of blood volume vectors is required (as this other gas will give a different blood absorption spectrum depending on its concentration).
In this illustrative explanation, it is assumed that only one blood component is to be determined, but at a sufficiently large number of wavelengths, more than one blood component can in principle be measured, as long as the set of characteristic vectors covers all possible gas combinations. For example, a plurality of different sets of blood volume vectors may be stored in a look-up table.
The above-described processing may be completed within a time window (e.g., 10 seconds), and a detection signal may be obtained in each wavelength channel. The window may be a sliding window, i.e. the next measurement is again from a 10 second window registered at some later time. From the measurements obtained in each window, a physiological parameter (e.g. SpO) is derived (derived based on the found "best quality" blood volume pulse vector)2) And estimating the result. The resulting series of physiological parameter estimates (from different time windows) may then be (temporally) filtered in order to obtain smoother, higher resolution measurements.
This filtering can be understood as follows.
As discussed, the sensing system may be adapted to repeatedly obtain the detection signal, e.g. at regular time intervals. The processor may be adapted to determine from these detection signals a corresponding time series (series) of "highest quality" blood volume pulse vectors, each "highest quality" blood volume pulse vector being a blood volume pulse vector that for a given point in time (based on the relevant set of detection signals obtained for that point in time) derives the highest quality pulse signal. Additionally or alternatively, may be according to a series of PbvVectors are derived to derive a corresponding series of physiological parameter measurements (e.g., SpO)2)。
In this case, the processThe processor may filter the obtained time series of blood volume pulse vectors to obtain filtered feature vectors from which physiological information is derived (or the processor may filter the SpO2Is filtered to obtain a filtered SpO2Value of (d).
In particular, the time series of vectors (or values) may be temporally filtered in order to eliminate outliers or to smooth the results, to obtain more reliable results or to e.g. improve the derived physiological parameter (e.g. SpO)2) The resolution of (2).
For example, SpO configured to retain only no more than 5% of separation may be applied2P with corresponding valuebvA filter of the vectors. In this case, the temporal filtering may be to SpO2The resolution of (2) is improved to 1%.
In accordance with one or more embodiments, a processor may be configured to: when determining the blood volume pulse vector which yields the pulse signal with the highest quality, the blood volume pulse vector (P) is adjusted in a direction which depends on which pulse signal yields the best quality indicator valuebv1、Pbv2) One or more blood volume pulse vectors of the finite set.
In these embodiments, the obtaining of the blood volume pulse vector (P) of the highest quality pulse vector may be performed by means of an iterative methodbv) Thereby testing PbvAn initial finite set of vectors, and based on the results, improved P is selected or derivedbvThe vectors are tested until convergence to the highest quality PbvVector to vector.
In such a method, not all possible P's are testedbvVector (corresponding to relevant physiological parameter value (e.g. SpO)2Value) but only a limited set (e.g., two) is tested. In this context, "test" means according to the relevant PbvThe vector derives a pulse signal and determines a quality indicator value for the pulse signal.
For example, if when testing an initial set (e.g., two) of pulse volume vectors, corresponding to the lowest SpO2Value PbvVector achievementStarting the second test when the highest quality index value is obtained, wherein the low SpO is tested2Vector of values and even lower value SpO2PbvAnd (5) vector quantity. This process continues until convergence at P which achieves the highest quality pulse signalbvUp to the vector.
In other words, a dynamic iterative testing process is followed, wherein the results of each round of testing are used to guide the direction in which further tests are performed, i.e. whether higher or lower physiological parameters corresponding to P should be testedbvAnd (5) vector quantity.
The result is a recursive process that involves fewer computations and is therefore more efficient.
According to one or more sets of embodiments, the processor may be adapted to perform a filtering step in which the obtained detection signals are adaptively filtered to reduce noise, before deriving the at least two pulse signals. In particular, the detection signal may be filtered to reduce the contribution of the signal component caused by the motion.
In one set of examples, the sensing system may be adapted to adaptively filter the detection signal based on a reference noise signal. For example, the reference noise signal may be a signal corresponding to a (motion) disturbance affecting the sensing system.
For example, in a particular set of examples, the sensing system may include a motion sensing module (e.g., an accelerometer) attached or coupled to the object for sensing motion of the object. The
Fig. 6 illustrates an example filter circuit for performing such a filtering process.
According to an exemplary circuit, an adaptive Finite Impulse Response (FIR)
An adaptive filter is a dynamic filter that iteratively changes its filter characteristics to achieve an optimal output. In particular, the adaptive filter is adapted to adjust its parameters based on an algorithm to minimize a difference function between the desired output of the filter and the actual output at each iteration.
In this example, the filter parameters of the
the circuit iteratively processes the input detection signal I using an
The output of the adaptive filter is coupled to the input detection signal I via a
Then, at each iteration, the error signal e (n) is fed back to the NLMS algorithm and the parameters of the
The final convergence result is output from the filter circuit. This is labeled as an enhancement signal ("enhancement signal") in fig. 6.
According to a further set of examples, the
In particular, in some examples, the detection signal may be filtered according to a heart rate signal detected by a heart rate sensing module of the sensing system. In these examples, the controller is adapted to: the heart
Fig. 7 illustrates an example filter circuit for performing such a filtering process.
Similar to the circuit of fig. 6, the circuit includes an adaptive Finite Impulse Response (FIR)
The adaptive filter receives a reference signal I corresponding to the pulse of the subject2As an input. For the present example, the reference signal is considered to be a heart rate signal obtained from a heart rate sensing module of the sensing system. As described in the examples above, in particular examples, the heart rate sensing module may comprise a PPG sensor or an ECG sensing device. The
The output of the
According to a further set of examples, a band-pass frequency filter may be applied to the detection signal (e.g., a narrow band-pass frequency filter) prior to determining the pulse signal. In these cases, the frequency range passed by the filter may be determined from the detected frequency components of the heart rate signal obtained using the heart rate sensing module. In this case, after the filter is applied, only those frequency components of the detection signal are left for which a correspondence is found in the frequency spectrum of the heart rate signal. This represents a simple method, for example, for implementing the pre-filtering method outlined in the above example.
By filtering the detection signal before deriving the pulse signal, the influence of motion artifacts can be reduced, thereby enabling an improved accuracy of the physiological parameter measurement.
As described above, embodiments of the present invention may advantageously utilize the inclusion or integration of a sensing interface (e.g., SpO)2Sensors) and a heart rate sensing module (e.g., ECG or accelerometer). In this way, all sensors can be conveniently integrated in a single patch. Such a patch should be able to give an accurate SpO even for moving objects (e.g. general ward patients and professional care facility patients or households)2And (6) reading.
More broadly, in the examples, any SpO2Sensors (whether based on a breast patch or not) can be derived from the solution provided by the inventionBenefits are realized in the solution.
Another example of an application for which the invention may provide benefits is the example of contactless physiological parameter monitoring (e.g. using a camera (vital signs camera)), e.g. remote PPG sensing.
As briefly noted above, the heart rate sensing module may be facilitated in a number of different ways.
According to various embodiments, the heart rate sensing module may comprise or may be facilitated by a PPG sensor. In some cases, the PPG sensor may be applied to a different body location than the sensing interface used to derive the detection signals used to determine the physiological parameter information.
For example, in one set of embodiments, the sensing system may comprise a single secondary PPG sensor for sensing the heart rate or pulse rate of the subject. As a simple example, the patient may be continuously monitored, for example, with an auxiliary PPG sensor (e.g., a finger PPG sensor) adapted to continuously or repeatedly derive a heart rate signal (in the form of a pulse rate). A physiological parameter sensing system (e.g., a bilirubin meter for optically sensing bilirubin levels) may include a sensing interface and a processor, and may be adapted to be communicatively connected to a PPG sensor for receiving a sensed heart rate signal output. The communication connection may, for example, comprise a Near Field Communication (NFC) connection, a bluetooth connection, or any other wired or wireless communication channel or link.
The processor uses the sensing interface of the bilirubin meter, and the method according to the invention derives physiological information indicative of bilirubin using heart rate signal information obtained from the PPG sensor.
According to a further set of example embodiments, the sensing system may comprise a plurality of PPG sensors for sensing the heart rate signal. In some examples, the intensity of the heart rate signal obtained from each PPG sensor may be determined, and a single sensor implementing the highest intensity heart rate signal (e.g. highest signal-to-noise ratio) may be used to obtain the heart rate signal used in deriving the physiological parameter information according to the method of the invention.
For example, in a simple example, the patient may be suffering fromSimultaneous application of multiple SpOs at different locations on a person's body2A sensor (as an example of a PPG sensor). Each SpO2The sensor is operable to sense a heart rate and obtain a detection signal Cn for deriving the SpO according to the method of the invention2. In this case, each SpO2The sensors perform the functions of both the
The sensor identified as generating (in the form of the measured pulse signal) the strongest measured heart rate signal (e.g., highest signal-to-noise ratio) is then identified. Then, SpO is derived according to the invention2(or any other physiological parameter) the sensor is used as a heart rate detection module for all other sensors.
According to at least one set of embodiments, the heart rate sensing module and the sensing interface may comprise or be facilitated by a PPG sensor. In such a case, the two may be integrally combined, i.e., both may include or be facilitated by the same PPG sensor.
In these cases, a single PPG sensor may be adapted to irradiate and sense a plurality of different wavelengths of electromagnetic radiation, and wherein one subset of wavelengths is used to directly detect heart rate signals (or pulse rate signals) and another subset of wavelengths is used to obtain detection signals for obtaining physiological parameter measurements according to the method of the invention.
In one set of examples, a single PPG sensing module may be utilized that is adapted to emit and sense electromagnetic radiation of at least three different wavelengths (e.g., 520nm (green), 660nm (red), 830nm (infrared), and optionally also 940nm (infrared)). In a particular example, the PPG sensing module may be incorporated into a chest patch or wristband, or may be provided by a camera (i.e., a remote PPG sensing module).
The green light alone may be used to measure the heart rate signal used in the method of the invention, while the red and infrared light are used to obtain information for deriving physiological parameters (e.g. SpO)2) The detection signal of (1). The green PPG signal typically has more than the red and infrared PPG signalsA much higher signal-to-noise ratio (stronger pulse, higher AC/DC) and is therefore more suitable for determining the heart rate or pulse rate. Furthermore, green light has a shallower penetration depth into tissue than red and infrared light, and is therefore less suitable for use, for example, in determining SpO2。
According to a further set of examples, a single PPG sensing module may be utilized, which is adapted to emit and sense electromagnetic radiation of at least two different wavelengths, e.g. 520nm (green) and one or more red and/or infrared wavelengths.
A green light alone may be used to measure the heart rate signal used in the method of the invention, while a green light signal may be used together with red/infrared light to obtain information for deriving physiological parameters (e.g. SpO)2) The detection signal of (1). Also, in an example, the PPG sensing module may be incorporated into a chest patch or wristband, or may be provided by a camera (i.e., a remote PPG sensing module).
According to a further set of examples, a single PPG sensing module may be utilized, which is adapted to emit and sense electromagnetic radiation of at least three different wavelengths, e.g. 520nm (green), 660nm (red), 830nm (infrared) and optionally also 940nm (infrared). In an example, the PPG sensing module may be incorporated into a chest patch or wristband, or may be provided by a camera (i.e., a remote PPG sensing module).
A combination of wavelengths (green together with red and/or infrared) may be used to obtain a heart rate signal for use in the method of the invention. This can be achieved using the method described in de Haan, G and van Leest, A, article "Improved mobility of remote-PPG by using the volume pulse signature". The method is similar to the method used in the present invention and comprises deriving a plurality of different prospective pulse signals based on different weighted combinations of different color channel signals (green, red and/or infrared), the weights being determined based on components of a set of pulse volume vectors. A quality indicator value (e.g. a signal-to-noise ratio) is determined for each derived pulse signal, and the pulse signal with the highest quality indicator value is used as an auxiliary heart rate signal in determining the physiological parameter according to the invention.
Other variations to the disclosed embodiments can be understood and effected by those skilled in the art in practicing the claimed invention, from a study of the drawings, the disclosure, and the appended claims. In the claims, the word "comprising" does not exclude other elements or steps, and the word "a" or "an" does not exclude a plurality. Although some measures are recited in mutually different dependent claims, this does not indicate that a combination of these measures cannot be used to advantage. Any reference signs in the claims shall not be construed as limiting the scope.
- 上一篇：一种医用注射器针头装配设备
- 下一篇：运动负荷估计方法、运动负荷估计装置和记录介质