System and method for determining at least one vital sign of a subject
By employing modulated electromagnetic radiation with different wavelengths and polarization channels in a remote photoelectric volumetric plethysmography system, combined with high-frequency modulation and low-resolution sensors, the stability and accuracy issues of vital sign measurement under environmental irradiation were resolved, achieving highly reliable vital sign monitoring in interference environments.
Patent Information
- Application Number
- CN201980039485.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2019-06-05
- Publication Date
- 2024-11-29
- Estimated Expiration
- 2039-06-05
AI Technical Summary
Existing remote photoplethysmography systems suffer from poor stability and accuracy under environmental irradiation or interference, making it difficult to effectively measure vital signs, especially under strenuous exercise or ambient light interference, which affects the reliability and stability of vital signs.
The system design includes an irradiation unit, a detection unit, and a processing unit. It utilizes first and second radiation sources to modulate electromagnetic radiation in different ways in different wavelength channels or polarization channels. Combined with low-resolution 2D sensors and high-frequency modulation, it extracts vital signs by demodulating detection signals and statistical measures. The system also uses a low-resolution 2D sensor array and a high-frequency modulated electromagnetic radiation source, combined with polarization filters and optical filters, to improve ambient light stability and signal-to-noise ratio.
Even under environmental irradiation or interference, it can effectively improve the reliability and stability of vital sign measurement, reduce sensitivity to ambient light, enhance the system's anti-interference ability, and is suitable for non-contact monitoring.
Smart Images

Figure CN112272538B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to a system and a method for determining at least one vital sign of a subject. Background Art
[0002] A person's vital signs, such as heart rate (HR), respiratory rate (RR) or (peripheral or pulsatile) blood oxygen saturation (SpO2; which provides an estimate of arterial oxygen saturation SaO2), serve as an indication of the person's current state and as a powerful predictor of serious medical events. Therefore, vital signs are widely monitored in inpatient and outpatient care settings, at home or in other health, leisure and fitness settings.
[0003] One method of measuring vital signs is plethysmography. Plethysmography generally refers to measuring changes in the volume of an organ or body part, and specifically refers to detecting changes in volume due to the cardiovascular pulse wave that travels through the subject's body with each heartbeat.
[0004] Photoplethysmography (PPG) is an optical measurement technique that evaluates changes in light reflection or transmission over time of a region or volume of interest. PPG is based on the principle that blood absorbs more light than surrounding tissue, so changes in blood volume with each heartbeat affect transmission or reflection accordingly. In addition to information about the pulse rate (heart rate), the PPG waveform (also called PPG signal) can include information attributable to other physiological phenomena such as respiration. By evaluating transmission and / or reflection at different wavelengths (typically red and infrared), blood oxygen saturation can be determined.
[0005] Conventional pulse oximeters (also referred to herein as contact PPG devices) for measuring a subject's pulse rate and (arterial) oxygen saturation are attached to the subject's skin, such as a fingertip, earlobe, or forehead. They are therefore referred to as "contact" PPG devices. Although contact PPG is essentially considered a non-invasive technology, contact PPG measurements are often unpleasant and obtrusive because the pulse oximeter is attached directly to the subject and the cable limits freedom of movement and can hinder workflow.
[0006] In the past decade, non-contact remote PPG devices (also called camera-based devices or video health monitoring devices) have been proposed for non-intrusive measurements. Remote PPG utilizes a light source, or generally a radiation source, arranged at a distance from the object of interest. Similarly, a detector (e.g., a camera or a photodetector) can be arranged at a distance from the object of interest. Therefore, remote photoplethysmography systems and devices are considered to be non-intrusive and well suited for medical as well as non-medical everyday applications.
[0007] Using PPG technology, vital signs can be measured. Vital signs are revealed by tiny light absorption changes in the skin caused by pulsating blood volume (i.e., changes in the periodic color of human skin caused by blood volume pulses). Since this signal is very small and is hidden in larger changes due to illumination changes and motion, there is a general interest in improving the essentially low signal-to-noise ratio (SNR). There are still demanding situations, including strenuous exercise, challenging environmental illumination conditions, or stringent accuracy requirements, which require improved stability and accuracy of vital sign measurement devices and methods, especially for more critical medical and healthcare applications.
[0008] Video health monitoring (monitoring or detecting, for example, heart rate, respiratory rate, SpO2, ECG, delirium, etc.) is a promising emerging field. Its inherent non-intrusiveness is a clear advantage for patients with fragile skin or who require long-term vital sign monitoring, such as NICU patients, patients with extensive burns, psychiatric patients who have contact sensors removed, or COPD patients who must be monitored at home during sleep. In other environments such as general wards or emergency rooms, the comfort of non-contact monitoring remains an attractive feature.
[0009] One of the important vital signs, the measurement of blood oxygen saturation (SpO2), has recently been shown to be feasible using camera-based rPPG. SpO2 measurement requires accurate detection of relative pulsatility, i.e., the normalized amplitude of the pulsating signal (AC / DC) in different wavelength channels. There are two main threats to the accuracy of remote measurements. The first is low pulsatility compared to the noise, and the second is that the beating cardiogram (BCG) motion modifies the relative pulsatility in each channel.
[0010] In addition to oxygen saturation, saturation of other arterial blood gases and blood species, such as HBCO, MetHB, HBCO2, bilirubin can be obtained using the same technique (with slight adjustments) and are also subject to the same threats that the present invention aims to address. Arterial blood components are selected using this technique because it is only assumed that arterial blood pulsates in the rhythm of cardiac activity. It is assumed that similarly, the respiratory cycle includes volume changes in the venous system, which means that the relative strength of the pulsation at the respiratory rhythm of different wavelengths can be used to analyze venous blood components, for example to estimate venous oxygen saturation (SvO2).
[0011] While this is a promising new field, many challenges must be overcome. Disadvantages of remote PPG systems are that their stability is often compromised when strong ambient light interferes with the desired skin radiation for remote PPG, skin segmentation is often difficult, and multispectral cameras are expensive.
[0012] Therefore, there is a need for an improved system and method capable of determining at least one vital sign of a subject to obtain results with higher reliability and stability even in the presence of environmental illumination or interference.
[0013] US 2014 / 0243622A1 discloses a photoplethysmography device comprising a light source for illuminating a target object. A modulator drives the light source so that the output intensity varies as a function of a modulation signal at a modulation frequency. A detector receives light from the target object and generates an electrical output as a function of the intensity of the received light. A demodulator having a local oscillator receives the detector output and produces a demodulated output that is insensitive to any phase difference between the modulation signal and the oscillator, indicating blood volume as a function of time and / or blood composition. Many demodulators can be provided to derive signals from multiple light sources of different wavelengths or from a detector array. The plethysmograph can be operated in transmission mode or reflection mode. When in reflection mode, the device can use the green part of the spectrum and can also use a polarization filter.
[0014] Further prior art can be found in WO 2016 / 166651 A1 and WO 2011 / 117780 A1. Summary of the invention
[0015] The object of the present invention is to provide a system and a method for determining at least one vital sign of a subject, by means of which it is possible to obtain results with greater reliability and stability even in the event of environmental illumination or disturbances.
[0016] In one aspect of the present invention, a system is provided, the system comprising
[0017] an irradiation unit configured to irradiate a skin area of a subject, the irradiation unit comprising a first radiation source and a second radiation source, the first radiation source and the second radiation source being configured to emit electromagnetic radiation modulated in different ways in two or more wavelength channels and / or in two or more polarization channels,
[0018] a detection unit configured to detect electromagnetic radiation reflected from a skin area of the subject and to generate a detection signal, wherein the detection unit comprises a plurality of detection elements and wherein the detection signal is generated by each detection element or a group of two or more detection elements,
[0019] a processing unit configured to demodulate the detection signal corresponding to the modulation applied by the illumination unit to obtain a demodulated detection signal, and
[0020] - a vital sign determination unit configured to determine a statistical measure from the demodulated detection signal for each wavelength or polarization channel, wherein the statistical measure is determined from samples of the demodulated detection signal taken by different detection elements at the same time, and to extract a vital sign from the statistical measure.
[0021] In another aspect of the present invention, a corresponding method is provided.
[0022] Preferred embodiments of the invention are defined in the dependent claims. It shall be understood that the claimed method has similar and / or identical preferred embodiments as the claimed system, in particular as defined in the dependent claims and disclosed herein.
[0023] A drawback of remote PPG systems is that, for example, cameras conventionally used to detect electromagnetic radiation reflected from (or transmitted through) a subject's skin area have a relatively low temporal sampling frequency (i.e., picture rate), which prevents low-cost (frequency / phase multiplexing) multi-wavelength solutions that are very stable to ambient illumination. The stability of using HF-modulated light comes from the fact that the ambient illumination spectrum does not usually exhibit many high frequencies. Using high-frequency modulation of the light source, and demodulating the signal from the camera, makes the system almost invisible to changes in ambient light. This can only work if the camera allows the high sampling rate required to capture HF-modulated light, which is severely limited by the current state of camera technology.
[0024] Another issue that cameras have to deal with with ambient light is that they have a very limited dynamic range, typically using 8-bit or maximum 12-bit brightness quantization. This means that even with frequency modulation, if the additional non-modulated ambient light is stronger than the modulated light, the modulated light can be lost through clipping on the optical sensor, which can easily happen in sunlight.
[0025] Another issue associated with the use of cameras relates to privacy, which the present invention can also address.
[0026] The proposed system for remote PPG can effectively eliminate one or more of these disadvantages by selecting a detection unit for detecting electromagnetic radiation and for generating a detection signal from which statistical information can be derived that allows to deal with intrinsic contamination of skin areas with light reflections from non-skin surfaces. Preferably, a low-resolution 2D sensor is used, such as a photodiode arranged in a small array, making very high sampling rates and a large dynamic range feasible. This allows for a HF modulated radiation source and can provide a strong ambient light stability in combination with its very large dynamic range.
[0027] The illumination unit comprises two radiation sources for emitting differently modulated electromagnetic radiation in two or more wavelength channels and / or two or more polarization channels.The illumination unit may be implemented, for example, by separate radiation (eg light) emitting elements such as LEDs.
[0028] Each detection signal is thus demodulated twice (corresponding to the modulation applied by the first radiation source and the second radiation source) (respectively), i.e., two demodulated detection signals are obtained for each measured detection signal. In other words, each pixel of the detection unit is demodulated with a different modulation frequency to obtain a pixel value for each wavelength or polarization channel.
[0029] Extracting vital signs from statistical measures. As an example, if there are 16 photodiodes in the detection unit, each generating a detection signal, and if there are three wavelength channels, preferably, at least two statistical measures are generated per wavelength channel from the detection signal, i.e., a total of six statistical measures are derived. In an embodiment, a centripetal trend measure (e.g., the mean or median of all pixels for each wavelength) and a dispersion measure (e.g., the variance of all pixels for each wavelength) are obtained for each wavelength channel. Vital signs are then extracted from the six statistical measures (two measures per three wavelength channels).
[0030] As used herein, a statistical measure (also referred to as a statistical parameter value) can relate to a measured value indicating the value of a pixel of a detection unit. Statistical dispersion (also referred to as variability, scattering or spread) can indicate the degree to which a distribution is stretched or compressed. Common examples of measured values of statistical dispersion are variance, standard deviation and interquartile range. Candidate signals can be determined by concatenating statistical measures of corresponding detection signals that vary over time. For example, at each moment, the standard deviation value of the pixel value at that moment is determined, and subsequent standard deviation values form candidate signals over time.
[0031] Therefore, according to the present invention, spatial statistics (i.e., statistical measures in the spatial domain) are used (e.g., summary statistics of samples of a 2D detection unit comprising an array of detection elements). Such spatial statistics are different from temporal statistics (e.g., which can be viewed as a demodulation that is temporally correlated with a demodulated periodic signal), which aim to summarize a set of samples obtained by combining temporally consecutive samples from the same detection element (e.g., a pixel), i.e., temporal domain statistics of a temporal signal. Instead, the present invention uses spatial statistics of samples of multiple detection signals acquired at the same time by different detection elements or groups of detection elements (i.e., for example, pixels of a detection unit of a low-resolution 2D sensor array).
[0032] Optionally, additional steps such as normalizing the candidate signal may be performed, for example by dividing it by its (moving window) time average, taking its logarithm and / or removing an offset. Furthermore, (pre-)processing steps such as weighting pixels by a weighting map, resizing, rescaling, resampling or cropping the (weighted) detection signal. The (weighted) detection signal used herein may therefore also relate to such a (pre-)processed (weighted) detection signal.
[0033] It should be noted here that the terms "light" and "radiation" as used herein are generally understood to have the same meaning, specifically referring to electromagnetic radiation in the visible and infrared spectra. Preferably, according to the present invention, radiation in the range of 400nm to 1000nm is used.
[0034] According to an embodiment, the statistical measure indicates at least one of a standard deviation, a variance, a mean absolute difference, a median absolute difference and / or an interquartile range. Thus, instead of the conventional approach of averaging pixels in the region of interest (i.e., evaluating a centripetal trend metric), a statistical measure indicative of statistical dispersion is evaluated.
[0035] According to an embodiment, the first radiation source and the second radiation source are configured to emit electromagnetic radiation of different wavelengths or wavelength ranges or mixed wavelengths.The (pseudo)-color signal can then be stably demodulated according to these modulated wavelength channels.
[0036] In another embodiment, the first radiation source and the second radiation source are configured to emit electromagnetic radiation with different polarizations. It is also possible to implement both options in one embodiment. This may further improve ambient light stability and / or reduce sensitivity to specular reflections.
[0037] Optionally, the illumination unit is configured to modulate the emitted electromagnetic radiation applying a modulation frequency in a range above 1 kHz, in particular in a frequency range from 10 kHz to 100 MHz. This enables them to be clearly distinguished from ambient illumination variations. In offices, current fluorescent lamps emit 100 / 120 Hz with potential harmonics, while recent LEDs typically emit strong frequencies in the range up to kHz. Some dimmable fluorescent lamps can even emit frequencies up to 20 kHz. Therefore, in order to prevent all conceivable interferences, it is preferred to use rather high frequencies. These frequencies are obviously not obtainable with current camera technology. Therefore, this embodiment provides a good stability to ambient light variations.
[0038] The first radiation source may be configured to apply a first modulation frequency for modulating the emitted electromagnetic radiation, the first modulation frequency being at least 8 Hz away from a second modulation frequency applied by the second radiation source for modulating the emitted electromagnetic radiation. This enables good separation of wavelength channels and also helps improve stability to ambient light.
[0039] The detection unit preferably comprises a plurality of detection elements, in particular a plurality of photodiode arrays, a CCD array or a CMOS array, a single photodiode or a photodiode array, which represents a cheap but effective solution. For example, a relatively small photodiode array may be used, i.e. an array just large enough to capture a single polarization direction or an array equipped with a single filter to further improve the ambient light stability.
[0040] The system may further comprise a polarization unit configured to apply polarization to the electromagnetic radiation reflected from the skin area of the subject. For example, a single (shared) polarizer may be arranged in front of all pixels of the detection unit. After demodulation, cross- and parallel polarization channels can be obtained simultaneously. In another embodiment, some detection elements may have the same polarizer as the first polarization of the first radiation source, and other detection elements may have the same polarizer as the second polarization of the second radiation source.
[0041] The system may further comprise an optical filter unit configured to filter the electromagnetic radiation reflected from the skin area of the subject to allow only the modulation frequency used by the illumination unit to pass. In a cost-effective embodiment, a common filter is provided that transmits only the wavelength used in the modulated radiation source. This renders the sensor "blind" to all other wavelengths, thereby improving ambient light stability (typically ambient light has a very wide spectrum (e.g., sunlight)). In another implementation, a filter may be provided for a single pixel that passes less than all of the wavelengths used (e.g., only one or two wavelengths in the case of 3 different wavelengths emitted by the radiation source). However, this reduces the number of effective pixels per wavelength, while further improving ambient light stability. Even in the case where the number of different filters used is equal to the number of wavelengths (e.g., three LEDs and three groups of pixels with different matched filters), modulation is helpful because most changes in the wavelength interval (e.g., caused by motion) may not fall in a narrow band near the modulation frequency.
[0042] The processing unit may also be configured to apply synchronous demodulation to the detection signal to obtain a demodulated detection signal.
[0043] The extraction of physiological parameters (e.g., vital signs) of an object based on candidate signals can be performed using known techniques. Exemplary techniques include, but are not limited to, evaluating a fixed weighted sum on candidate signals of different wavelength channels (RGB, IR), blind source separation techniques advantageously involving two candidate signals (e.g., blind source separation based on selecting independent signals of maximum pulsation), principal component analysis (PCA), independent component analysis (ICA), CHROM (chrominance-based pulse extraction), POS (where pulses are extracted in a plane orthogonal to the skin color vector), PBV method (using a predetermined signature of the blood volume pulse vector), or APBV (an adaptive version of the PBV method that also allows detection of blood oxygen saturation). It should be noted that optional pre-processing can be applied to the candidate signals, such as correcting the gain of one or more channels, to further improve performance.
[0044] Therefore, in an embodiment, the vital sign determination unit is configured to determine the vital sign based on the demodulated detection signal by linearly combining the demodulated detection signal through weighted combination, wherein the weights of the weighted combination are determined by blind signal separation, specifically by principal component analysis or independent component analysis, and by selecting component channels of the combined detection signal according to predetermined criteria.
[0045] In another embodiment, the vital sign determination unit is configured to determine the vital sign based on the demodulated detection signal by linearly combining the demodulated detection signal using a weighted combination of weights to produce a pulse signal, for which the product with the original detection signal is equal to the relative pulsation represented by each signature vector, and the signature vector provides the expected relative strength of the detection signal in the original detection signal.
[0046] At least one of the first radiation source and the second radiation source is configured to emit a controllable narrow radiation beam. For this purpose, a control unit may be provided. Thus one or more of the direction, beam width and intensity of the radiation beam emitted by the radiation source may be controlled. This may, for example, be used to irradiate primarily or only a skin portion of the subject without irradiating other portions of the subject or her / his environment.
[0047] Most effectively, the illumination unit is configured to illuminate the skin area of the subject with electromagnetic radiation in the wavelength range of 300 nm to 1000 nm. BRIEF DESCRIPTION OF THE DRAWINGS
[0048] These and other aspects of the invention will be apparent from and elucidated with reference to the embodiments described hereinafter.
[0049] In the following figures
[0050] Figure 1 A schematic diagram showing a first embodiment of a system according to the present invention is shown;
[0051] Figure 2 shows a schematic diagram of a second embodiment of a system according to the present invention;
[0052] Figure 3 shows a schematic diagram of a third embodiment of a system according to the present invention; and
[0053] Figure 4 A schematic diagram showing a fourth embodiment of a system according to the invention is shown. DETAILED DESCRIPTION
[0054] Figure 1 A schematic diagram of a first embodiment of a system 1 according to the invention for determining at least one vital sign of a subject 50 is shown. The subject 50 may be, for example, a patient lying in a bed in a hospital or other medical facility, but may also be, for example, a newborn or premature baby lying in an incubator, or a person at home or in other environments.
[0055] The system 1 comprises an irradiation unit 2, which is configured to irradiate a skin area of a subject 50. The irradiation unit 2 comprises a first radiation source 20 and a second radiation source 21, which are configured to emit electromagnetic radiation modulated in two or more wavelength channels and / or in two or more polarization channels in a different manner. For example, the radiation sources may be configured as LEDs. In an embodiment, a plurality of radiation sources, i.e. more than two radiation sources, may be provided.
[0056] The radiation sources 20, 21 may be controllable. For this purpose, a control unit 3 may be provided which is configured to control the radiation sources 21, 22 of the irradiation unit 2, which may be, for example, a controller or a processor.
[0057] The system 1 further comprises a detection unit 4 configured to detect electromagnetic radiation reflected from a skin area of the subject 50 and to generate a detection signal from the detected electromagnetic radiation. The detection unit 4 may be, for example, a single photodiode or a photodiode array.
[0058] The system 1 further comprises a processing unit 11 configured to demodulate the detection signal corresponding to the modulation applied by the illumination unit 2 to obtain a demodulated detection signal.
[0059] Finally, the system 1 comprises a vital sign determination unit 5 configured to determine a statistical measure from the demodulated detection signal for each wavelength or polarization channel and to extract a vital sign from the statistical measure. The processing unit 11 and / or the vital sign determination unit 5 may be, for example, a processor or a computer or dedicated hardware.
[0060] The system 1 may also optionally include an interface 6 for displaying the determined information and / or for providing an interface for medical personnel to change the settings of system components. Such an interface may include different displays, buttons, touch screens, keyboards or other human-machine interface devices.
[0061] If the radiation source is, for example, an LED, modulation of the emitted electromagnetic radiation is most easily achieved by modulating the current through the LED. An electronic circuit may provide a current modulated around a bias value, for example in a sinusoidal form (although other waveforms are possible and may be attractive). For light sources other than LEDs, it may be difficult to achieve such HF modulation using the light source itself, so an optical modulator may additionally be used to achieve modulation of the emitted electromagnetic radiation.
[0062] like Figure 1 The system 1 shown may be located, for example, in a hospital, a healthcare facility, an elderly care facility, etc. In addition to monitoring patients, the invention may also be applied in other fields, such as neonatal monitoring, general monitoring applications, safety monitoring or so-called lifestyle environments, such as fitness equipment, wearable handheld devices such as smartphones, etc. One-way or two-way communication between two or more components of the system 1 may operate via a wireless or wired communication interface.
[0063] Figure 1 The embodiment shown is a low-cost embodiment, particularly suitable for automated applications. The illumination unit 2 and the detection unit 4 are observing a scene with an object 50 through a common optical element 7 (e.g., a common lens). Since the face of the object 50 is relatively close to the illumination unit 2, it reflects much more light than the surface behind the object. In such an exemplary embodiment, a semi-transparent mirror 8 can be used, but in practice it may not be possible to sufficiently suppress direct light toward the detection unit 4, i.e., separate lenses for the illumination unit 2 and the detection unit 4 can be used instead.
[0064] The signals from individual detection elements of the detection unit 4 are demodulated, preferably in a 2D array of detection elements, to retrieve the different wavelength / polarization channels for each detection unit 4. Next, statistical measures are calculated for each wavelength / polarization channel, such as centripetal trend (e.g., mean or median of all demodulated detector signals or channels from the array), and dispersion (e.g., variance or standard deviation of all demodulated detector signals or channels from the array) measures to deal with the fact that (part of) the reflected light falling on some detection elements may come from non-skin surfaces (it should be noted that in general no region of interest (RoI) detectors are used, but the entire set of detection signals is representative of the skin and some background (which may be considered as contaminated RoI)).
[0065] Finally, the vital signs are extracted from the statistical measures of the channels (preferably all channels), for example, the PPG signal (pulse) can be obtained as a linear combination of the statistical measures of all wavelength channels. In general, any previously reported method (PCA, ICA, PBV, CROM, POS, APBV, PBVGOP) can be applied to the respective statistical measures, for example PCA to the mean, PCA to the variance, resulting in two PPG signals, from which the output PPG signal is obtained by selection or combination, for example linear weighting using a quality indicator (e.g. skewness of the spectrum, height of the highest peak in the normalized spectrum, etc.). The background of this selection is that the mean works well in the case where a few channels correspond to non-skin surfaces, while the variance works better in the case where most of the channels correspond to non-skin and there may be only a few channels corresponding to skin.
[0066] Typically, electromagnetic radiation used for pulse, respiration and blood oxygen saturation measurements is in the range of 400nm to 1000nm, specifically in the range of 620nm to 920nm. This specific range is best suited for SpO2 measurements and is attractive for non-intrusive monitoring during sleep (for near-dark situations, further restriction to the range of 760nm to 920nm may be more preferred), however, the visible light portion of the spectrum may allow for higher quality in case visibility of the light is non-intrusive (i.e., NIR is not necessarily the preferred option in all cases). The detection signal can be collected by a photosensor (or photosensor array) that remotely senses the subject's skin, i.e., at least some skin needs to be imaged by the detector, but the detector may also see some background.
[0067] By using an illumination unit 2, illumination of non-skin surfaces can in principle be prevented or minimized. Various methods can be used to achieve this, but an example is to embed the illumination into a feedback loop, where the quality (e.g. SNR) of the extracted PPG signal is optimized by varying the intensity of the "pixels" of the illumination unit.
[0068] In a practical embodiment, the illumination unit 2 may emit radiation in at least two or three different wavelength intervals (RGB, or invisible wavelengths using NIR) in a modulated (frequency or phase) manner to achieve stability to ambient light. The (pseudo) color signal can then be stably demodulated from these modulated wavelength channels (or alternatively, obtained from different photodiodes equipped with different optical filters, in which case the modulation has enhanced stability to ambient light only and can be the same for all wavelengths (i.e. same modulation frequency)), and the PPG signal can be stably extracted from the wavelength channel shifts using the same basic algorithms already proposed for camera-based extraction (e.g. ICA, PCA, CHROM, POS, (A)PBV methods, etc.), which will be explained in detail below.
[0069] In order to obtain good stability to ambient illumination variations, the illumination unit 2 can be modulated at a relatively high frequency (e.g. between 1 kHz and several MHz), wherein it can be expected that the ambient light spectrum is generally quite clean and therefore does not cause interference to the demodulated wavelength channels. These high frequencies are easier to achieve because the detectors are high-speed detectors, such as photodiodes, rather than imaging sensors typically used in remote PPG applications. Although the channels do not need to be separated very far, different wavelengths / polarizations can be modulated at different frequencies (in the case of monochromatic detectors). A separation of about 8 Hz is sufficient, since for maximum human pulse rates the sidebands caused by the pulse signal are limited to about + / - 4 Hz.
[0070] In another embodiment, ambient light stability is improved by using a detection unit (e.g., a photodetector) that is equipped with an optical filter that selectively transmits the wavelengths used in the modulated illumination unit, but blocks all (or most) other ambient light, for example by means of a visible light blocking filter.
[0071] Another embodiment may improve ambient light stability by using a separate photodetector for each wavelength used, wherein each photodetector may be equipped with an illumination unit matched to the optical filter.
[0072] To increase the stability to ambient light or to reduce the sensitivity to specular reflections, a polarizer can be used in front of the detection unit and / or the illumination unit. Obtaining two signals obtained by means of different polarization filters can also allow a cleaner combined (partially independent) signal than the individual signals themselves. This enables the option of having a single wavelength, or provides twice as many detection signals to be combined than without a polarizer.
[0073] The control unit 3 for controlling the irradiation unit 2 may be configured to control the direction and / or width and / or intensity of the (generally relatively narrow) beam 30 from the irradiation unit 2, such that, for example, as much skin as possible is irradiated and as little non-skin surface as possible is irradiated, using the same optimization criteria. Preferably, the radiation sources 21, 22 may be controlled individually or in groups to achieve this effect.
[0074] exist Figure 1 In the embodiment of the system 1 shown, different wavelengths can be multiplexed or multiple LEDs (different wavelengths) can be modulated at different frequencies. Very high modulation frequencies (kHz-MHz) are possible and good separation using, for example, a product detector (or heterodyne demodulator) is also possible. For example, the demodulation can be effected by computing the inner product over a sliding analysis window with sine and cosine waves, and computing the square root of the squared inner product, and finally low pass filtering (e.g. 4 Hz cutoff) the value obtained from the sliding window processing. The high modulation frequency prevents problems with motion and can make the system very stable to ambient illumination, i.e. the frequency selectivity makes the system "blind" to ambient illumination variations typically below 1 kHz (50 / 100 Hz for classical light sources, hundreds of Hz for LED lamps), and below 100 kHz. In this case, the relative proximity of the face (compared to the background) can avoid the need for pixelated light, or alternatively, the highly absorbing background directly behind the subject can avoid disruptive contributions to non-skin surfaces.
[0075] Figure 2 A schematic diagram of another embodiment of a system 100 according to the present invention is shown, which mainly shows an irradiation unit and a detection unit, while omitting other components of the system, such as a control unit and a vital sign detection unit.
[0076] In an embodiment of the system 100, it is possible to have the illumination unit 2 project a narrow beam 30, e.g. wide enough to allow the necessary freedom of movement when acting as a car driver, towards, e.g., an object 50, and have the detection unit 4 view the object 50 with a similarly narrow viewing angle 40. Thus, unlike system 1, in this embodiment the light source optical paths and the sensor are separated. Since the face is illuminated most, being closest to the illumination unit 2, the background has little effect on the signal quality (or a highly absorbing background (e.g. a headrest)). Thus, the skin / face contributes most (significantly) to the light on the detector array, but the use of one or more statistical measures allows vital sign extraction even when there is contamination of the detection light from non-skin surfaces.
[0077] Both optical paths 30 and 40 are narrow, and assuming that their intersection contains mainly skin, the PPG signal will have good quality even if the background is not black. In some applications, using a dark background is also a viable option. If the background plays an important role in the reflected light from the scene, it is still possible to obtain a high-quality PPG signal by combining signals formed by one or more statistical measures (e.g., centripetal tendency and dispersion).
[0078] Using a photodiode as a detection unit, its detection signal can be demodulated synchronously in the kHz range. For example, a 3-wavelength system modulated in frequency multiplexing (e.g., using 760nm, 800nm and 900nm) can have its center frequency around 10kHz, with wavelength channels at least 8Hz apart, such as 10, 11 or 12kHz. Due to the possible pulse frequency range (0.5-4Hz, which will provide a sideband of + / -4Hz to the modulation frequency), this separation is preferred. Therefore, they should be 8Hz apart to prevent interference.
[0079] However, there are other interference considerations. Nonlinearities in the radiation source (e.g., LEDs) cause harmonics of the modulation frequency, which should also not cause interference in the pulse bands of other wavelengths (e.g., a 10kHz modulation frequency causes harmonics at 20, 30, 40, etc. kHz). If any demodulator mixes any of these harmonics into the pulse frequency band, interference will result (e.g., when using 10kHz for wavelength 1, other modulation frequencies should also stay clear of 40kHz + / - 32Hz to avoid interference).
[0080] Therefore, there may be various restrictions on the selection of modulation frequency. As becomes clear from the above example, there is not only a lower boundary, but also many frequency bands that should not be used.
[0081] The received light may also be modulated with a different polarization direction (cross and parallel to the polarization of the light of the illumination beam 20). For this purpose, polarizers 9 and 10 may be used. The use of polarized light for illumination and a crossed polarizer 10 in front of or as part of the detection unit 4 (in particular in front of a sensor sensing the received radiation) may help suppress specularly reflected light. Also, the use of polarizers may (partially) replace the use of additional wavelengths to improve motion stability and the SNR of the PPG signal.
[0082] To suppress specular reflections, it is sufficient to use a single polarizer for all radiation sources and orthogonal (crossed) polarizers in front of the detection unit. To improve motion stability, there are several options:
[0083] a) using a single polarizer in front of all radiation sources and using crossed polarizers and parallel polarizers in front of two detection elements (forming a detection unit);
[0084] b) alternating (switching) polarization in front of a single detection element (forming a detection unit);
[0085] c) using a single polariser in front of the detection unit and switching the polarisation in front of the radiation source; or
[0086] d) Use multiple polarizers on a parallel radiation source.
[0087] Generally, polarization filters can be used as polarizers. However, for this purpose, not only transmission filters can be used, but also mirrors and / or reflectors (eg polarizers) can be used to achieve the same effect.
[0088] Figure 3 A schematic diagram of a third embodiment of a system 200 according to the invention is shown. In this embodiment, a plurality of illumination units 2a, 2b are used, each with a relatively narrow beam 30a, 30b, and optionally each with a polarizer 9a, 9b. Only objects at the intersection of the narrow beam 30a, 30b and the narrow sensor viewing angle 40 contribute to the detection signal.
[0089] Figure 4 A schematic diagram of a fourth embodiment of a system 300 according to the invention is shown. This embodiment is particularly suitable for automated applications, since background illumination can be completely avoided. The face of the subject is illuminated from the side, preferably using polarized light. The detection unit 4 is equipped with a crossed polarizer 10, so that only reflections that have been depolarized by the semi-transparent object are seen in its view. In a car, it should be relatively easy to avoid any other (significantly reflective) objects in the field of view, so that only scattered light reflected from the skin can be "seen" by the detection unit 4.
[0090] Another alternative embodiment employs a very narrow beam (narrow enough to cause an illumination area smaller than the target body part (e.g. the face)) that can be directed by the controller. In this embodiment, the detection unit can "see" a larger environment, but despite this it can give a high quality PPG signal if the spot only reaches the subject's skin. Possibly, the controller changes the direction of the light beam in such a way as to optimize the PPG signal quality.
[0091] In general, the PPG signal is generated as a function of the changes in the blood volume in the skin. Thus, the changes give a characteristic pulsation "signature" when observed in the different spectral components of the reflected / transmitted light. This "signature" is essentially generated as a contrast (difference) between the absorption spectrum of blood and the absorption spectrum of bloodless skin tissue. If the detector, such as a camera or sensor, has a discrete number of color channels, each channel sensing a specific part of the spectrum, it is possible to arrange the relative pulsations in these channels in a "signature vector", also called a "normalized blood volume vector", PBV. This has been shown in G. de Haan and A. van Leest "Improved motion robustness of remote-PPG by using the blood volume pulse signature" (2014, Physiol. Meas. 35 1913), which is incorporated herein by reference, that is, if this signature vector is known, then a motion-stable pulse signal extraction based on the color channels and the signature vector is possible. For the quality of the pulse signal, it is crucial that, although the signature vector is accurate, there are other known methods to mix noise into the output pulse signal to achieve the described correlation of the pulse vector with the normalized color channel indicated by the signature vector.
[0092] Details of the PBV method and the use of a standardized blood volume vector (referred to as a "predetermined index element having a set direction indicating reference physiological information") have also been described in US 2013 / 271591 A1, the details of which are also incorporated herein by reference.
[0093] When the blood composition changes, the characteristic wavelength dependence of the PPG signal also changes. In particular, the oxygen saturation of arterial blood has a strong effect on light absorption in the wavelength range between 620nm and 780nm. This changing signature for different SpO2 values results in a relative PPG pulsation that depends on the arterial blood oxygen saturation. This dependence can be used to implement a motion-stable remote SpO2 monitoring system, which has been named the adaptive PBV method (APBV) and is described in detail in M. van Gastel, S. Stuijk and G. de Haan, "New principle for measuring arterial blood oxygenation, enabling motion-robust remote monitoring" (November 2016, Nature Scientific Reports). The description of the details of the APBV method in this document is also incorporated herein by reference.
[0094] The PBV method gives the cleanest pulse signal when the PBV vector reflecting the relative pulsation in the different wavelength channels is accurate. Since this vector depends on the actual SpO2 value, the SpO2 value is generated as a value corresponding to the PBV vector that gives the pulse signal with the highest SNR by testing the PBV method with different PBV vectors corresponding to the range of SpO2 values.
[0095] In the following, some basic considerations about the PBV method are briefly explained using RGB wavelength channels as an example.
[0096] 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 arteries are elastic, their diameter changes in sync with the pressure changes. These diameter changes occur not only in the large arteries, but even in the smaller blood vessels of the skin on the arterial side of the capillary bed, the arterioles, where changes in blood volume cause changes in light absorption.
[0097] The unit-length normalized blood volume pulse vector (also called signature vector) is defined as PBV, assuming that the relative PPG intensities in the red, green, and blue camera signals are
[0098]
[0099] Here, σ indicates the standard deviation.
[0100] To quantify the expected value, the responses H of the red, green, and blue channels are red (w), H green (w) and H blue (w) is measured as a function of wavelength w with a global shutter color CCD camera, the subject's skin reflectance ρs(w), and the absolute PPG amplitude curves PPG(w) are used. Based on these curves, for example, in the above-mentioned paper by de Haan and van Leest Figure 2 As shown in , the blood volume pulse vector PBV is calculated as:
[0101]
[0102] Here, a white halogen illumination spectrum I(w) is used, resulting in a normalized PBV = [0.27, 0.80, 0.54].
[0103] The blood volume pulse predicted by the used model corresponds quite reasonably to the experimentally measured normalized blood volume pulse vector, PBV = [0.33, 0.78, 0.53], created after averaging the measurements of multiple subjects under white illumination conditions. Given this result, it can be inferred that the PPG amplitudes observed specifically in the red and to a lesser extent in the blue camera channel can be explained mainly by crosstalk of wavelengths in the interval between 500 and 600 nm. As shown in the model, the exact blood volume pulse vector depends on the camera's color filter, the spectrum of the light and the skin reflectivity. In practice, the vector is very stable despite a given set of wavelength channels (the vector is different in the infrared compared to the RGB-based vector).
[0104] It was also found that the relative reflectivity of the skin in the red, green and blue channels under white illumination is not very dependent on the skin type. This is probably because the absorption spectrum of bloodless skin is dominated by melanin absorption. Although higher melanin concentrations can increase the absolute absorption quite significantly, the relative absorption at different wavelengths remains unchanged. This means that an increase in melanin darkens the skin but hardly changes the standardized color of the skin. Therefore, the standardized blood volume pulse PBV is also very stable under white illumination. In the infrared wavelengths, the influence of melanin is further reduced because its maximum absorption occurs at short wavelengths (ultraviolet) and decreases for longer wavelengths.
[0105] The main reason why the PBV vector is not affected too much by melanin is that melanin is present in the epidermis and therefore acts as an optical filter on both AC and DC. Therefore, it reduces the pulsation, but it also reduces the DC value of the reflection. Therefore, the AC / DC (relative pulsation) does not change at all.
[0106] The stable characteristics of the PBV can be used to distinguish color changes caused by changes in blood volume from changes caused by other reasons, that is, the stable PBV can be used as a "signature" of blood volume changes to distinguish its color changes. The known relative pulsations of the color channel PBVs can be used to distinguish pulse signals from distortion. The pulse signal S obtained using known methods can be written as a linear combination of the various DC-free normalized color channels (one of several possible ways to express "mixing"):
[0107] S=WC n
[0108] Among them, WW T = 1, and the 3 x N matrix C n Each of the three rows of contains N red, green, and blue channel signals R without DC normalization. n , G n and B n A sample of , namely:
[0109]
[0110] Here, the operator μ corresponds to the mean. The key difference between the different methods lies in the calculation of the weighting vector W. In one method, the noise and the PPG signal can be separated into two independent signals, which are constructed as a linear combination of two color channels. One combination approximates the clean PPG signal and the other contains the noise caused by motion. As an optimization criterion, the energy in the impulse signal can be minimized. In another method, a linear combination of three color channels can be used to obtain the impulse signal.
[0111] The PBV method generally uses the blood volume pulse vector to obtain the mixing coefficient, essentially as described in US 2013 / 271591 A1 and in the paper by de Haan and van Leest cited above. n , G n and B n The best results are obtained by using a bandpass filtered version of the PBV. According to this method, the known orientation of the PBV is used to distinguish between impulse signals and distortion. This not only cancels the assumption (of the earlier method) that the impulses are the only periodic components in the video, but also eliminates the assumption of the orientation of the distortion signal. To this end, as before, it is assumed that the impulse signal is constructed as a linear combination of the normalized color signals. Since the relative amplitudes of the impulse signals in the red, green and blue channels are known to be given by the PBV, the weight W that gives the impulse signal S is searched for. PBV , for the pulse signal S, and the color channel R n , G n and B n The correlation is equal to PBV
[0112]
[0113] And therefore, the weights that determine the mixture are determined by the following formula
[0114] in,
[0115] And, the scalar k is determined so that W PBV has unit length. It is concluded that the characteristic wavelength dependence of the PPG signal as reflected in the normalized blood volume pulse PBV can be used to estimate the pulse signal from time sequential RGB pixel data averaged over the skin area. This algorithm is called the PBV method.
[0116] In other words, the weights indicate how the detection signals should be (linearly) combined in order to extract the impulse signal from the detection signals. The weights are unknown and need to be calculated / selected.
[0117] The signature vector (PBV vector) represents the given (known or expected) relative pulsations in the different wavelength channels (i.e. the detection signal), which are caused by the absorption spectrum of the blood and the penetration of light through the skin (if the photons are absorbed more by the blood, the volume change of the blood leads to a larger signal compared to when the blood is almost transparent). With this knowledge, as well as the observed data (i.e. the detection signal), the weights (e.g. the weight vector) can be determined. The obtained weights are data-dependent, i.e. depend on the detection signal.
[0118] Since the pulse signal has a different AC / DC ratio in each wavelength channel (also called relative signal intensity / pulsation), it can be seen that the spectrum shows pulse peaks in the spectrum with different peaks for different colors. This spectrum is the result of Fourier analysis, but basically means that if a sine wave with the pulse frequency is correlated (multiplied) with the detection signal (RGB in the example, NIR-wavelength for SpO2), exactly the peaks in the spectrum are obtained, which are by definition called signature vectors (PBV vectors): these peaks are the relative intensities of the normalized amplitudes of the pulse signal in the different detection signals.
[0119] As a result, using this prior knowledge (i.e., the signature vector), a clean impulse signal S can be obtained (assuming that the impulse signal is the result of a weighted sum of the detection signals). One option is to normalize the detection signal C n The covariance matrix Q of is inverted. Therefore, using the constant signature vector PBV, the weight W for linearly mixing the detection signals to extract the pulse signal S can be calculated based on the covariance matrix (Q) of the detection signals in the current analysis window, which is data-dependent, i.e., continuously changing over time.
[0120] It is recognized that, for example, in the near infrared spectrum, specifically between 620 and 770 nm, the blood absorption spectrum changes depending on the SpO2 level. For this reason, it is proposed to extract pulse signals with different signature vectors (different PBV vectors), wherein each PBV vector is selected to correspond to the relative pulsation in the detection signal for a specific vital sign value, such as an SpO2 value. Since the extracted pulse signal quality depends on the correctness of the PBV vector, different extracted pulse signals will have different qualities. By selecting the pulse signal of the best quality, the vital sign value (e.g. SpO2 value) that caused such a favorable pulse signal can be obtained from the signature vector.
[0121] In another embodiment, the APBV method is used to extract SpO2 values from two or more different combinations of three wavelength channels, for example, from [λ1, λ2], from [λ1, λ3] and / or from [λ1, λ2, λ3]. In the following, some basic considerations about the APBV method will be briefly explained.
[0122] Instead of extracting features from the PPG waveform, APBV indirectly determines SpO2 based on the signal quality of the pulse signal extracted using the SpO2 "signature". This process can be described mathematically as:
[0123]
[0124] Among them, C n contains the DC-normalized color variation, and the scalar k is chosen such that has unit length. The SpO2 signature compiled in can be derived from physiology and optics. Assuming identical cameras, the PPG amplitude of N cameras can be determined by the following formula:
[0125]
[0126] Here, the PPG amplitude spectrum PPG(λ) can be approximated by linearly mixing the light absorption spectra from the two most common variants of the primary chromophore in arterial blood, hemoglobin; oxygenated (HbO2) and reduced (Hb):
[0127]
[0128] where it is assumed that for 600 < λ < 1000 nm and SaO2∈[0,1], the optical path length difference is negligible. It is recognized that wavelength-dependent scattering effects can invalidate this assumption. When two wavelengths are used, the ratio of the parameter R and the APBV parameter The wavelength selection can be based on three criteria: 1) the desire to measure oxygen saturation in darkness for clinical applications (λ>700nm); 2) a wavelength with reasonable SpO2 contrast; and 3) a wavelength within the spectral sensitivity of the camera. The improved two-fold stability of the SpO2 measurement inspired the idea of using three wavelengths in pulse oximetry instead of the commonly used two. This can be explained by how motion affects the PPG waveform when measured with a camera. Since the intensity changes induced by motion are equal at all wavelengths, if the pulse signature If φ is not equal to this motion signature, which can be described as a vector with equal weights, then suppression of these artifacts is possible for the APBV method.
[0129] It should be noted that even if the pulse quality is very good, it does not always mean that the estimated SpO2 value is reliable enough and can be trusted. This can occur in particular when unexpected blood-species (e.g. COHb) are available, causing a shift in the SpO2 calibration curve, i.e. causing a different signature vector, resulting in an optimal pulse quality when using the PBV method or the APBV method for pulse extraction.
[0130] Crucially in the above, the PBV method assumes that the relative pulsations in the wavelength channels are known, which is true if the desired vital sign information, such as SpO2, is known. However, this is not essentially the case in SpO2 monitoring, since this is the parameter being searched for. If the weights are chosen correctly, the resulting pulses are similar to the individual detection signals C n The correlation is exactly the detection signal C n Now, if the vital sign information (e.g. SpO2) is wrong or unknown, the result will be a pulse signal with a relatively poor SNR (i.e. a poor quality indicator).
[0131] The essence of the APBV method is to run multiple PBV methods in parallel using different PBV vectors. The PBV method gives the cleanest pulse signal when the PBV vector reflecting the relative pulsation in the different wavelength channels is accurate. Since this vector depends on the actual SpO2 value corresponding to the range of SpO2 values, the PBV method with different PBV vectors is tested, and the SpO2 value is given as a value corresponding to the PBV vector that gives the pulse signal with the highest SNR.
[0132] Typically, the electromagnetic radiation used according to the invention is in the range of 400nm to 1000nm for pulse, respiration and blood oxygen saturation measurements, specifically in the range of 620nm to 920nm. This specific range is best suited for SpO2 measurements and is attractive for non-intrusive monitoring during sleep (darkness), but if a pulse or respiration signal is required, the visible part of the spectrum may allow for higher quality (i.e., NIR is not necessarily the preferred choice in all cases).
[0133] The above method can be applied to detection signals that have been collected using non-contact sensors. For example, the present invention can be applied in the field of medical care, such as non-intrusive remote patient monitoring, overall monitoring, safety monitoring, and so-called lifestyle environments, such as fitness equipment, etc. Applications may include monitoring of changes in blood oxygen saturation (pulse oximetry), pulse rate, blood pressure, cardiac output, respiration, blood perfusion, evaluation of automatic functions, and detection of peripheral vascular diseases. The present invention can be used, for example, for rapid and reliable pulse detection of critically ill patients, such as during automatic CPR (cardiopulmonary resuscitation). The system can be used to monitor the vital signs of newborns with very sensitive skin, such as in the NICU, and for patients with damaged skin (such as burns), but can also be more convenient than contact sensors used in ordinary wards, and provide better solutions for motion stability. Other applications are in the field of automation. The present invention specifically solves the problem of multi-wavelength systems with ambient light stability, provides a lower cost solution, and eliminates privacy issues.
[0134] Although the present invention has been illustrated and described in detail in the drawings and the foregoing description, such illustration and description should be considered illustrative or exemplary rather than restrictive; the present invention is not limited to the disclosed embodiments. Other variations of the disclosed embodiments can be understood and effected by those skilled in the art in practicing the claimed invention by studying the drawings, the disclosure, and the appended claims.
[0135] In the claims, the word "comprising" does not exclude other elements or steps, and the indefinite article "a" or "an" does not exclude a plurality. A single element or other unit may fulfill the functions of several items recited in the claims. The mere fact that certain measures are recited in mutually different dependent claims does not mean that a combination of these measures cannot be used to advantage.
[0136] Any reference signs in the claims should not be construed as limiting the scope.
Claims
1. A system for determining at least one vital sign of a subject, the system comprising: an irradiation unit (2) configured to irradiate a skin area of the subject, the irradiation unit comprising a first radiation source (20) and a second radiation source (21), the first radiation source and the second radiation source being configured to emit electromagnetic radiation modulated in different ways in two or more wavelength channels and / or in two or more polarization channels, a detection unit (4) configured to detect electromagnetic radiation reflected from the skin area of the subject and to generate a detection signal, wherein the detection unit (4) comprises a plurality of detection elements, and wherein each detection signal is generated by each detection element or a group of two or more detection elements, a processing unit (11) configured to demodulate the detection signal corresponding to the modulation applied by the illumination unit to obtain a demodulated detection signal, and A vital sign determination unit (5) is configured to determine a statistical measure based on the demodulated detection signal for each wavelength or polarization channel and extract a vital sign from the statistical measure, wherein each statistical measure is determined based on samples of the demodulated detection signal obtained by different detection elements at the same time.
2. The system according to claim 1, in, The first radiation source (20) and the second radiation source (21) are configured to emit electromagnetic radiation at different wavelengths or wavelength ranges or mixtures of wavelengths.
3. The system according to claim 1 or 2, in, The first radiation source (141) and the second radiation source (142) are configured to emit electromagnetic radiation having different polarizations.
4. The system according to claim 1 or 2, in, The irradiation unit (2) is configured to apply a modulation frequency for modulating the emitted electromagnetic radiation in a range above 1 kHz, in particular in a frequency range from 10 kHz to 100 MHz.
5. The system according to claim 1 or 2, in, The first radiation source (20) is configured to apply a first modulation frequency for modulating the emitted electromagnetic radiation, the first modulation frequency differing by at least 8 Hz from a second modulation frequency applied by the second radiation source (22) for modulating the emitted electromagnetic radiation.
6. The system according to claim 1 or 2, in, The statistical measure indicates at least one of: a standard deviation, a variance, a mean absolute difference, a median absolute difference, and / or an interquartile range.
7. The system according to claim 1 or 2, in, The detection unit (4) comprises a photodiode array, a CCD array or a CMOS array.
8. The system according to claim 3, Also included is a polarization unit (10) configured to apply polarization to the electromagnetic radiation reflected from the skin area of the subject.
9. The system according to claim 2, Also included is an optical filter unit configured to filter the electromagnetic radiation reflected from the skin area of the subject to allow only a modulation frequency used by the illumination unit to pass.
10. The system according to claim 1 or 2, in, The processing unit (11) is configured to apply synchronous demodulation to the detection signal to obtain the demodulated detection signal.
11. The system according to claim 1 or 2, in, The vital sign determination unit (5) is configured to determine the vital sign based on the demodulated detection signal by linearly combining the demodulated detection signal by weighted combination, wherein the weights of the weighted combination are determined by blind signal separation, specifically by principal component analysis or independent component analysis, and by selecting component channels of the combined detection signal according to a predetermined criterion.
12. The system according to claim 1 or 2, in, The vital sign determination unit (5) is configured to determine a vital sign from the demodulated detection signal by linearly combining the demodulated detection signal by a weighted combination, wherein the weighted combination uses weights that produce pulse signals for which the product with the original detection signal is equal to the relative pulsation represented by a corresponding signature vector, which provides the expected relative strength of the detection signal in the original detection signal.
13. The system according to claim 1 or 2, in, At least one of the first radiation source (20) and the second radiation source (21) is configured to emit a controllable narrow radiation beam.
14. The system according to claim 1 or 2, in, The vital sign determination unit (5) is configured to determine a centripetal tendency measure and a dispersion measure as statistical measures from the demodulated detection signal for each wavelength or polarization channel.
15. A method for determining at least one vital sign of a subject, the method comprising: irradiating the skin area of the subject by an irradiation unit, the irradiation unit comprising a first radiation source (20) and a second radiation source (21), the first radiation source and the second radiation source being configured to emit electromagnetic radiation modulated in different ways in two or more wavelength channels and / or in two or more polarization channels, detecting electromagnetic radiation reflected from the skin area of the subject and generating a detection signal by a detection unit (4), wherein the detection unit (4) comprises a plurality of detection elements and wherein each detection signal is generated by each detection element or a group of two or more detection elements, demodulating the detection signal corresponding to the modulation applied by the illumination unit to obtain a demodulated detection signal, determining a statistical measure from the demodulated detection signal for each wavelength or polarization channel, wherein each statistical measure is determined from samples of the demodulated detection signal taken by different detection elements at the same time, and A vital sign signal is extracted from the statistical measure.
Citation Information
Patent Citations
Device and method for extracting information from characteristic signals
US20130271591A1
Interference reduction in monitoring a vital parameter of a patient
WO2011117780A1
photoplethysmography
US20140243622A1
Method and apparatus for optical sensing of tissue variation at increased accuracy
US20170055860A1
Vital sign monitoring
WO2016166651A1
