Non-contact blood oxygen monitoring device and method improved using notch filter

Abstract

Disclosed in the present invention are a non-contact blood oxygen monitoring device and method improved using a notch filter. The device is used for real-time, non-contact blood oxygen saturation monitoring of a human skin tissue. The device is composed of an optical imaging module, an image analysis module, a blood oxygen calibration module, and a UI interface module. The optical imaging module obtains a video of consecutive image frames of the human skin tissue; the image analysis module receives and analyzes the video of the consecutive image frames of the human skin tissue in real time, and extracts pulse wave relative amplitude characteristics; the blood oxygen calibration module calibrates blood oxygen saturation on the basis of the pulse wave relative amplitude characteristics extracted by the image analysis module and a calibration model to obtain a blood oxygen value monitoring result; and the video of the consecutive image frames of the human skin tissue and the blood oxygen value monitoring result are displayed on a UI interface.

Description

Non-contact blood oxygen monitoring device and method using notch filter improvement Technical Field

[0001] The present invention relates to the field of blood oxygen monitoring, and in particular to a non-contact blood oxygen monitoring device and method improved by using a notch filter. Background Art

[0002] In order to achieve health monitoring and disease prevention for newborns and the elderly population, there is an urgent need to continuously monitor physiological information such as pulse and blood oxygen. Currently, the commonly used and relatively mature wearable physiological continuous monitoring products are mostly two types: (1) sports bracelets, which monitor heart rate signals through a single green band, but the physiological indicators they can monitor are too single. (2) finger oximeters, which monitor heart rate and blood oxygen saturation through red and infrared dual bands, but due to the weak infrared band PPG (Photoplethysmography) signal, the stability of heart rate monitoring is low. In addition, wearable devices are also prone to damage and infection of the skin of newborns or burn patients. Therefore, non-contact monitoring methods need to be further developed.

[0003] In existing technologies, camera-based (camera-PPG) video sensing technology has been used for non-contact vital sign monitoring and health monitoring. This typically involves extracting physiological signals from a video stream consisting of consecutive frames using image / signal processing algorithms. While methods for monitoring heart rate and respiratory rate are relatively mature, research and application of non-contact blood oxygenation monitoring, particularly for neonates, is relatively limited.

[0004] The cameras (camera-PPG) currently used in this type of research are mainly divided into two categories:

[0005] (i) Narrow-band filter camera. This method adds a narrow-band filter in front of the camera so that only light of specific wavelengths can pass through (usually dual-band, one for the red band and the other for near-infrared / green / blue bands). This method ensures the purity of the pulse wave signal, but because the narrow-band filter only allows a small amount of light to pass through, the signal-to-noise ratio of the extracted pulse wave signal in low light conditions is very low, and it is even impossible to distinguish the pulse wave signal from the ambient white noise. Therefore, scenes using narrow-band filter cameras usually require strong fill light. However, newborns (especially those in intensive care) are very sensitive to strong light, and strong fill light can cause discomfort to the baby and even worsen their condition.

[0006] (ii) Broadband RGB camera. This method uses a broadband RGB camera to collect pulse wave signals across a wide band of red, green, and blue channels and extracts blood oxygen saturation estimates from the red and green channels. Broadband RGB cameras have low light source brightness requirements, allowing pulse wave signals with a relatively high signal-to-noise ratio to be obtained even in low light conditions. However, the red and green channels of ordinary RGB cameras overlap, meaning both the red and green channels respond to the orange light band between red and green. This makes it impossible to obtain consistent measurement values ​​even if the actual blood oxygen saturation is consistent when different light sources have different light intensities in the orange light band, reducing the robustness of the estimation model to different indoor lighting conditions.

[0007] Therefore, the existing technology still needs to be improved and developed.

[0008] Summary of the Invention

[0009] In view of the above-mentioned deficiencies in the prior art, an object of the present invention is to provide a non-contact blood oxygen monitoring device improved by using a notch filter to solve the problem of low robustness of the existing non-contact blood oxygen monitoring device caused by the influence of light.

[0010] The technical solutions of the present invention are as follows:

[0011] A non-contact blood oxygen monitoring device improved by using a notch filter is used for non-contact real-time blood oxygen saturation monitoring. The device is characterized in that it consists of an optical imaging module, an image analysis module, a blood oxygen calibration module and a UI interface module;

[0012] A notch filter for obtaining a usable wavelength band is provided in the optical imaging module, and the optical imaging module is used to obtain a video containing continuous image frames of human skin tissue and a signal containing physiological information of human skin tissue;

[0013] The image analysis module is used to receive and analyze in real time the signal containing the physiological information of human skin tissue acquired from the optical imaging module, and extract the relative amplitude feature of the pulse wave therefrom;

[0014] The blood oxygen calibration module is used to calibrate the blood oxygen saturation according to the pulse wave relative amplitude characteristics extracted by the image analysis module and the calibration model to obtain a blood oxygen value monitoring result;

[0015] The UI interface module is used to display the video of the continuous image frames of the human skin tissue and the blood oxygen value monitoring results on the UI interface.

[0016] A further configuration of the present invention is

[0017] The optical imaging module includes: a light source emitting unit;

[0018] The light source emitting unit is used to provide a continuous and stable spectrum;

[0019] The light source emitting unit includes: a full-spectrum LED light source controlled by rare earth phosphor and a first polarizer;

[0020] The full-spectrum LED light source controlled by the rare earth phosphor is used to generate a continuous spectrum;

[0021] The first polarizer is placed between the front side of the full-spectrum LED light source controlled by the rare earth phosphor and the human skin tissue, and is used to convert the continuous spectrum photons generated by the full-spectrum LED light source controlled by the rare earth phosphor into photons with a single vibration direction.

[0022] A further configuration of the present invention is

[0023] The optical imaging module further includes: a camera sensor unit, the camera sensor unit being configured to monitor photons in the wavelength bands of 450-583 nm and 605-700 nm;

[0024] The camera sensor unit includes: a second polarizer, a camera and an optical sensor;

[0025] The second polarizer is arranged on the front side of the camera, and is placed between the camera and the human skin tissue, and the second polarizer is used to ensure that the camera obtains a diffuse reflection signal containing physiological information;

[0026] The camera is a broadband RGB camera;

[0027] The notch filter is provided at the front end of the camera to block photons in the 583nm-605nm band and enhance the independence of photons between the 450-583nm and 605-700nm bands;

[0028] The camera is used to focus photons between the 450-583nm and 605-700nm bands onto the optical sensor;

[0029] The optical sensor is used to convert the photons between the 450-583nm and 605-700nm bands into the signal containing the physiological information of human skin tissue, and then convert the signal containing the physiological information of human skin tissue into continuous image frames of the human skin tissue.

[0030] According to a further configuration of the present invention, the continuous spectrum photons emitted by the full-spectrum LED light source controlled by the rare earth phosphor are irradiated onto the human skin tissue through the first polarizer, and the photons reflected after being absorbed and scattered by the human skin tissue enter the camera through the second polarizer.

[0031] In a further arrangement of the present invention, the image analysis module includes a multispectral pulse wave signal extraction unit;

[0032] The multispectral pulse wave signal extraction unit is used to extract pulse wave signals of multiple bands in the time domain from the video containing continuous image frames of human skin tissue, perform detrending and bandpass filtering on the pulse wave signals of the multiple bands in the time domain, filter out non-pulse wave signals in the pulse wave signals of the multiple bands in the time domain, and obtain RGB three-channel pulse wave signals;

[0033] A green light pulse wave signal and a red light pulse wave signal are extracted from the RGB three-channel pulse wave signal.

[0034] In a further arrangement of the present invention, the image analysis module includes: a pulse wave relative amplitude estimation unit;

[0035] The pulse wave relative amplitude estimation unit is used to extract the relative amplitude characteristics of the green light and red light pulse waves from the green light pulse wave signal and the red light pulse wave signal obtained by the multi-spectral pulse wave signal extraction unit;

[0036] The relative amplitude characteristic of the green and red light pulse waves is the ratio of the peak-to-peak value of the red light pulse wave signal to the peak-to-peak value of the green light pulse wave signal in the same pulsation cycle; or,

[0037] The ratio of the amplitude of the Fourier transform of the red light pulse wave signal corresponding to the frequency point of the heart rate to the amplitude of the Fourier transform of the green light pulse wave signal corresponding to the frequency point of the heart rate.

[0038] According to a further configuration of the present invention, the ratio of the peak-to-peak value of the red light pulse wave signal to the peak-to-peak value of the green light pulse wave signal in the same pulsation period is obtained by:

[0039] respectively monitoring the peaks and troughs of the red light pulse wave signal and the green light pulse wave signal using a peak-to-peak value monitoring algorithm;

[0040] Based on the monitored peaks and troughs of the red light pulse wave signal and the green light pulse wave signal, the difference between the peaks and troughs of the red light pulse wave signal and the difference between the peaks and troughs of the green light pulse wave signal are calculated, and the relative amplitude characteristics of the red light and green light pulse wave signals are obtained by performing a ratio operation on the two calculated differences.

[0041] In a further arrangement of the present invention, the blood oxygen calibration module includes: a regression model, wherein the regression model establishes a functional relationship between the relative amplitude characteristics of the pulse wave and the reference blood oxygen value for continuous dynamic estimation of blood oxygen saturation;

[0042] The regression model is established by polynomial regression method;

[0043] The regression model is provided with calibration model parameters;

[0044] The calibration model parameters are obtained through pre-training;

[0045] The regression model and calibration model parameters are used to calibrate the relative amplitude characteristics of the input red and green light pulse wave signals to obtain the blood oxygen value monitoring result.

[0046] The present invention is further configured such that the image analysis module inputs the relative amplitude characteristics of the red and green pulse wave signals into the blood oxygen calibration module;

[0047] The blood oxygen calibration module outputs the obtained blood oxygen value monitoring result to the UI interface.

[0048] The present invention further provides a non-contact blood oxygen monitoring method using an improved notch filter, comprising:

[0049] The optical imaging module acquires a video containing continuous image frames of human skin tissue and a signal of physiological information of the human skin tissue;

[0050] The image analysis module receives and analyzes the signal containing the physiological information of human skin tissue obtained from the optical imaging module in real time, and extracts the relative amplitude feature of the pulse wave therefrom;

[0051] The blood oxygen calibration module calibrates the blood oxygen saturation according to the pulse wave relative amplitude feature extracted by the image analysis module and the calibration model to obtain a blood oxygen value monitoring result;

[0052] The UI interface module displays the video of the continuous image frames of the human skin tissue and the blood oxygen value monitoring results on the UI interface.

[0053] The present invention provides a non-contact blood oxygen monitoring device and method using a notch filter, which is used for non-contact real-time blood oxygen saturation monitoring. The device includes an optical imaging module, an image analysis module, a blood oxygen calibration module, and a user interface module. The optical imaging module is configured to capture a video containing continuous image frames of human skin tissue and a signal containing physiological information of human skin tissue. The image analysis module is configured to receive and analyze the signal containing physiological information of human skin tissue acquired from the optical imaging module in real time, extracting pulse wave relative amplitude features from the signal. The blood oxygen calibration module is configured to calibrate blood oxygen saturation based on the pulse wave relative amplitude features extracted by the image analysis module and a calibration model to obtain blood oxygen monitoring results. The user interface module is configured to display the video containing continuous image frames of human skin tissue and the blood oxygen monitoring results on a UI interface. By adding a notch filter to an RGB camera, the optical imaging module of the present invention can selectively filter out interfering signals, thereby improving the accuracy of blood oxygen saturation estimation. The image analysis module further extracts pulse wave relative amplitude features from the signal acquired by the optical imaging module. The blood oxygen calibration module calibrates the signal acquired by the image analysis module to obtain the final blood oxygen value detection result, which is displayed on the UI display module. This invention uses a non-contact method to estimate blood oxygen saturation, achieving improved contactless blood oxygen monitoring for newborns. An RGB camera sensor with a notch filter is used to capture and analyze images. The image analysis module and blood oxygen calibration module generate a measurement value consistent with the actual blood oxygen saturation, greatly improving the robustness of the estimation model to different indoor lighting conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary personnel in this field, other drawings can be obtained based on the structures shown in these drawings without paying any creative work.

[0055] FIG1 is a connection diagram of modules of a preferred embodiment of the device of the present invention.

[0056] FIG2 is a schematic diagram of an optical imaging module of a preferred embodiment of the device of the present invention.

[0057] FIG3 is a schematic diagram of a notch filter camera assembly according to a preferred embodiment of the present invention.

[0058] FIG4 is a comparison experiment result of a notch camera and an ordinary camera under the superposition of an orange light source and an incandescent lamp in a preferred embodiment of the device of the present invention.

[0059] FIG5 is a comparative experimental result of a notch filter camera and a narrow-band filter camera of a preferred embodiment of the device of the present invention under weak light.

[0060] FIG6 is a spectrum analysis and comparative experimental result of a preferred embodiment of the device of the present invention.

[0061] FIG7 is a diagram showing rPPG amplitude changes at different blood oxygen saturations for the red and green light channels of a preferred embodiment of the device of the present invention.

[0062] FIG8 is a fitting result of the pulse wave signal characteristics (ratio of ratios) and the blood oxygen value of the regression model of the preferred embodiment of the device of the present invention.

[0063] The marks in the accompanying drawings are: optical imaging module 100, light source emitting unit 110, camera sensor unit 120, full-spectrum LED light source controlled by rare earth phosphor 111, first polarizer 112, second polarizer 121, camera 122, notch filter 123, optical sensor 124, image analysis module 200, blood oxygen calibration module 300 and UI interface module 400. DETAILED DESCRIPTION

[0064] The present invention provides a non-contact blood oxygen monitoring device utilizing an improved notch filter. To further clarify the objectives, technical solutions, and advantages of the present invention, the present invention is further described below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are intended only to illustrate the present invention and are not intended to limit its scope.

[0065] In the embodiments and patent claims, unless otherwise specified herein, the words "a," "an," "the," and "the" may include plural forms. If the embodiments of the present invention include descriptions of "first," "second," etc., such descriptions are for descriptive purposes only and should not be understood as indicating or implying their relative importance or implicitly specifying the number of the technical features indicated. Therefore, a feature specified as "first" or "second" may explicitly or implicitly include at least one of such features.

[0066] It should be further understood that the term "comprising" as used in the description of the present invention refers to the presence of the stated features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof. It should be understood that when we refer to an element as being "connected" or "coupled" to another element, it can be directly connected or coupled to the other element, or there can be intermediate elements. In addition, "connected" or "coupled" as used herein can include wireless connections or wireless couplings. The term "and / or" as used herein includes all or any units and all combinations of one or more associated listed items.

[0067] It will be understood by those skilled in the art that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by those skilled in the art in the art to which the present invention belongs. It should also be understood that terms such as those defined in common dictionaries should be understood to have meanings consistent with their meanings in the context of the prior art and will not be interpreted in an idealized or overly formal sense unless specifically defined as herein.

[0068] In addition, the technical solutions between the various embodiments can be combined with each other, but they must be based on the fact that ordinary technicians in this field can implement them. When the combination of technical solutions is mutually contradictory or cannot be implemented, it should be deemed that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0069] Please refer to FIG. 1 to FIG. 7 simultaneously. The present invention provides a preferred embodiment.

[0070] As shown in FIG1 , the present invention provides a non-contact blood oxygen monitoring device improved by a notch filter, which is used for non-contact real-time blood oxygen saturation monitoring. The device includes: an optical imaging module 100 , an image analysis module 200 , a blood oxygen calibration module 300 and a UI interface module 400 .

[0071] Specifically, the present invention sets a notch filter for obtaining an available wavelength band in the optical imaging module, and the optical imaging module is used to obtain a video containing continuous image frames of human skin tissue and a signal containing physiological information of human skin tissue; and the image analysis module 200 receives and analyzes the signal containing physiological information of human skin tissue obtained from the optical imaging module 100 in real time, and extracts the relative amplitude feature of the pulse wave from it; the blood oxygen calibration module 300 calibrates the blood oxygen saturation according to the relative amplitude feature of the pulse wave extracted by the image analysis module 200 and the calibration model to obtain the monitored blood oxygen value result; finally, the UI interface module 400 displays the video containing continuous image frames of human skin and the blood oxygen value monitoring result on the UI interface. The image analysis module 200 and the blood oxygen calibration module 300 are actually software function modules that can be run on a processor in the device of the present invention, and can be used to implement corresponding functions respectively, that is, the image analysis module 200 is used to extract the relative amplitude characteristics of the pulse wave, and the blood oxygen calibration module 300 is used to obtain the actual blood oxygen value based on the relative amplitude characteristics obtained above using a regression model.

[0072] Please refer to FIG2 , in a further implementation of an embodiment, the optical imaging module 100 includes: a light source emitting unit 110 and a camera sensor unit 120;

[0073] Specifically, in this embodiment, the light source emitting unit 110 is used to provide a continuous and stable spectrum, providing a sufficient light source for monitoring pulse wave signals in different bands, thereby enhancing the ability to monitor arterial blood oxygen content.

[0074] Please continue to refer to FIG. 2 . In a further implementation of an embodiment, the light source emitting unit 110 includes: a full-spectrum LED light source 111 controlled by rare earth phosphors and a first polarizer 112 .

[0075] The full-spectrum LED light source 111 controlled by the rare earth phosphor can generate a continuous spectrum. In this embodiment, the continuous spectrum includes visible light and a near-infrared band of 450nm-490nm.

[0076] The first polarizer 112 is placed between the front side of the full-spectrum LED light source 111 controlled by the rare earth phosphor and the human skin tissue. The light emitted by the full-spectrum LED light source 111 controlled by the rare earth phosphor is 360 degrees. The first polarizer 112 only allows light at vertical angles to pass through, and can convert the continuous spectrum photons generated by the full-spectrum LED light source 111 controlled by the rare earth phosphor into photons with a single vibration direction.

[0077] Specifically, the full-spectrum LED light source 111 regulated by rare earth phosphors is a light-emitting diode that can excite rare earth phosphors by injecting current to emit photons. In the full-spectrum LED light source 111 regulated by rare earth phosphors, the role of rare earth phosphors is to convert short-wavelength excitation light into long-wavelength emission light. When current passes through the LED, the semiconductor material in the LED chip will produce short-wavelength light, which will be absorbed by the rare earth phosphors. Subsequently, the rare earth phosphors will re-radiate emission light with a continuous spectrum. These emission lights will cover the range of visible light and near-infrared bands (450-950nm), becoming the light source for illumination and imaging in the pulse wave imaging system. Therefore, photons are generated by full-spectrum LEDs, and then these photons will pass through the first polarizer 112 to achieve a single vibration direction, thereby being used in the imaging process of the band-polarized light pulse wave imaging system.

[0078] Referring to FIG. 2 , in a further embodiment, the optical imaging module 100 further includes a camera sensor unit 120 capable of monitoring photons in the 450-583 nm and 605-700 nm wavelength bands. This embodiment utilizes an RGB camera sensor, which is a sensor capable of capturing light signals in the three basic colors of red, green, and blue. RGB camera sensors are typically used to capture images and record videos. Using an RGB camera sensor allows a device to accurately capture light signals of different colors and convert them into digital images or video files.

[0079] Specifically, the camera sensor unit 120 includes a second polarizer 121 , a camera 122 and an optical sensor 124 .

[0080] The second polarizer 121 is arranged on the front side of the camera 122. At the same time, the second polarizer 121 is placed between the camera 122 and the human skin tissue. The second polarizer 121 and the first polarizer 112 only allow light at a perpendicular angle to pass through. The angle between the second polarizer 121 and the first polarizer 112 is perpendicular, so that the light reflected by the mirror cannot reach the camera 122, which is used to ensure that the camera 122 obtains a diffuse reflection signal containing physiological information.

[0081] The camera 122 may be a broadband RGB camera. The use of the broadband RGB camera can reduce the brightness requirement of the light source for contactless blood oxygen monitoring, so that blood oxygen monitoring can be performed on newborns under a light intensity that they can accept.

[0082] Referring to FIG. 3 , the camera 122 further includes a notch filter 123 disposed at the front end of the camera 122 for blocking photons in the 583nm-605nm band (yellow and orange light) and enhancing the independence between usable wavelengths such as 450-583nm (green light) and 605-700nm (red light). These usable wavelengths, after reflection from human skin tissue, are capable of carrying and reflecting signals containing physiological information about human skin tissue, such as pulse wave signals. This allows for further calculation and processing of additional physiological information, such as blood oxygen concentration, through regression model processing. The camera 122 is configured to focus photons in the 450-583nm and 605-700nm bands onto the optical sensor 124.

[0083] Specifically, please refer to Figure 4. A narrow-band filter camera uses a narrow-band filter in front of the camera to allow only specific wavelengths of light to pass through. These wavelengths are typically dual-band: one for red and one for near-infrared / green / blue. Newborns (especially those in intensive care) are extremely sensitive to strong light. Supplemental light exposure can cause discomfort and even worsen their condition. When using a narrow-band filter camera, the filter allows only a small amount of light to pass through, making it impossible to distinguish the pulse wave signal from ambient white noise. This results in a very low signal-to-noise ratio (SNR) in low-light conditions.

[0084] The low-light comparison results shown in Figure 4 show that the notch filter camera can still accurately measure blood oxygen levels in low-light environments with minimal error. However, the camera using a narrow-band filter measured a large deviation, significantly deviating from the true value. This demonstrates that the notch filter camera has superior performance in low-light environments.

[0085] Specifically, please refer to Figure 5. A broadband RGB camera is used to collect pulse wave signals across a wide band of red, green, and blue channels, and to extract blood oxygen saturation estimates from the red and green channels. Broadband RGB cameras have low light source brightness requirements, enabling pulse wave signals with a relatively high signal-to-noise ratio even in low light conditions. However, conventional RGB cameras have overlapping red and green channels. That is, both the red and green channels respond to the orange light band between red and green. Blood oxygen measurement only requires the red (R) and green (G) bands. Hemoglobin absorption in the orange light band varies significantly with wavelength. This means that when different light sources have different intensities in the orange light band, consistent blood oxygen saturation measurements cannot be obtained, even if the true blood oxygen saturation is the same. This reduces the robustness of the estimation model to varying indoor lighting conditions.

[0086] Figure 5 shows a comparison of a notch filter camera and a standard camera under a combination of orange and incandescent light. The experimental results show that the notch filter camera can still accurately estimate blood oxygen levels in this environment, while a standard RGB camera produces a large error. This demonstrates that the notch filter camera also performs better under colored light interference.

[0087] Specifically, please refer to Figure 6. From the spectral analysis and comparative experiments shown in Figure 6, it can be concluded that the notch filter significantly suppresses the overlapping part of the red light (R) channel and the green light (G) channel and the steep part of the oxygenated hemoglobin absorption spectrum. Therefore, the notch filter camera can measure blood oxygen content more accurately.

[0088] Therefore, broadband RGB cameras equipped with notch filters offer improved performance, accurately measuring blood oxygen levels and operating under varying lighting conditions. This ensures that the perceptible wavelengths of the red (R) and green (G) channels lie within the relatively flat absorption spectrum of oxygenated and deoxygenated hemoglobin, significantly suppressing the effect of light source composition changes on the relative intensity of the pulse wave. This is of great significance for applications in medicine and health monitoring. Furthermore, multiple notch filters, or notch filters of varying specifications—that is, notch filters with varying cutoff bands—can be used with broadband RGB cameras.

[0089] The optical sensor 124 is arranged in the camera 122 and is a chip composed of photosensitive elements. It is used to convert photons between the 450-583nm and 605-700nm bands into signals containing physiological information of human skin tissue, and then convert the signals containing physiological information of human skin tissue into continuous image frames of the human skin tissue.

[0090] Please refer to Figure 2. Furthermore, the continuous spectrum photons emitted by the full-spectrum LED light source 111 controlled by the rare earth phosphor are irradiated on the human skin tissue through the first polarizer 112. The photons reflected after being absorbed and scattered by the human skin tissue pass through the second polarizer 121 and enter the optical sensor 124 in the camera 122. That is, only the polarized photons emitted by the full-spectrum LED light source 111 controlled by the rare earth phosphor can pass through the second polarizer 121 on the front side of the camera 122 after being absorbed and scattered by the skin tissue.

[0091] In a further implementation of an embodiment, the image analysis module 200 includes a multi-spectral pulse wave signal extraction unit and a pulse wave relative amplitude estimation unit.

[0092] The multispectral pulse wave signal extraction unit is configured to extract the arithmetic mean of human skin pixels from a video containing consecutive image frames of human skin tissue. The values ​​calculated from the consecutive frames are arranged in chronological order to form pulse wave signals in multiple time domain bands. The pulse wave signals in the time domain bands are subjected to detrending and band-pass filtering to filter out non-pulse wave signals, i.e., to filter out noise interference, from the pulse wave signals in the time domain bands to obtain an RGB three-channel pulse wave signal. A green light pulse wave signal and a red light pulse wave signal are then extracted from the RGB three-channel pulse wave signal.

[0093] In a further implementation of an embodiment, the pulse wave relative amplitude estimation unit is used to extract the pulse wave relative amplitude feature from the green light pulse wave signal and the red light pulse wave signal extracted by the multi-spectral pulse wave signal extraction unit.

[0094] This embodiment uses non-contact pulse wave signals for blood oxygen monitoring. By extracting a large number of pulse wave relative amplitude features (ratios) and calibrating them against actual blood oxygen values, a fitting model for the blood oxygen value and pulse wave relative amplitude features is developed. Extracting the pulse wave relative amplitude features involves extracting changes in human skin pixel values ​​in different wavelength bands from a video of continuous image frames containing human skin tissue, using these changes as rPPG measurements for the corresponding wavelength bands. Furthermore, pulse-related changes are extracted from the rPPG, and the absorption of light in the corresponding wavelength band by the pulsating component is obtained. Combined with the differences in absorption of light in different wavelength bands by oxygenated and deoxygenated hemoglobin, the proportion of oxygenated hemoglobin in the total hemoglobin at that time is estimated, resulting in an estimated blood oxygen saturation value.

[0095] In a further implementation of one embodiment, the pulse wave relative amplitude feature of this embodiment may be the ratio of the peak-to-peak values ​​of the red and green light pulse wave signals in the same pulsation cycle, or the ratio of the amplitude of the Fourier transform of the red light pulse wave signal corresponding to the frequency point of the heart rate to the amplitude of the Fourier transform of the green light pulse wave signal corresponding to the frequency point of the heart rate.

[0096] The peak-to-peak value ratio of the pulse wave signals of different bands in the same pulsation cycle is obtained by monitoring the peaks and troughs of the green pulse wave signal and the red pulse wave signal in the RGB three-channel pulse wave signal through a peak-to-peak value monitoring algorithm, calculating the difference between the peaks and troughs of the green pulse wave signal and the red pulse wave signal based on the peaks and troughs of the green pulse wave signal and the red pulse wave signal as the estimated value of the pulse wave amplitude, and obtaining the relative amplitudes of the red and green pulse wave signals by performing a ratio operation on the two calculated differences.

[0097] Further, referring to FIG. 7 , in a further implementation of an embodiment, the relative amplitudes of the pulse wave signals of the red (R) channel and the green (G) channel of a broadband RBG camera equipped with a notch filter change significantly when the blood oxygen level decreases, indicating that the blood oxygen level has an impact on the signal amplitude.

[0098] Please refer to (a) in Figure 7. When the blood oxygen saturation is 99, the rPPG amplitude of the red light (R) channel is significantly smaller than the rPPG (Remote Photoplethysmography) amplitude of the green light (G) channel, indicating that the signal of the red light (R) channel is weaker at high blood oxygen saturation.

[0099] Please refer to (b) in Figure 7. When the blood oxygen saturation drops to 91, the rPPG amplitude of the red light (R) channel is almost the same as the rPPG amplitude of the green light (G) channel, indicating that at lower blood oxygen saturation, the signal amplitudes of the two channels are similar.

[0100] Therefore, the present invention extracts the pulse wave signals of the red (R) and green (G) channels from the obtained RGB three-channel pulse wave signal, and then extracts the characteristics (ratio of ratios) of the red (R) pulse wave signal and the green (G) pulse wave signal from the pulse wave signals of the red (R) and green (G) channels for the next step of blood oxygen estimation.

[0101] In one embodiment of the present invention, the method for extracting the features of the red (R) pulse wave signal and the green (G) pulse wave signal may be as follows:

[0102] After extracting the pulse wave signals of red light (R) and green light (G), a sliding window length of 10s is selected within the sliding window, wherein the step length can be selected from 1 frame to 1s.

[0103] The red (R) and green (G) pulse wave signals within the sliding window are divided by their average values ​​to obtain the value of ratios.

[0104] The red (R) and green (G) pulse wave signals are band-pass filtered respectively to retain only the signals within the heart rate frequency band, which is generally 0.7 Hz-3 Hz for adults and 1.5 Hz-5 Hz for infants.

[0105] The arithmetic mean of the peak-to-peak values ​​of the red (R) and green (G) pulse wave signals after bandpass filtering in the sliding window or the amplitude of the main frequency after Fourier transformation are calculated respectively, so as to obtain the amplitude of the signal in the calculation sliding window.

[0106] A ratio operation is performed on the results of the red light (R) and green light (G) channels respectively, and the ratio is the ratio value in ratio of ratios.

[0107] The blood oxygen calibration module 300 includes a regression model, which establishes a functional relationship between the relative amplitude characteristics of the pulse wave and the reference blood oxygen value for continuous dynamic estimation of blood oxygen saturation.

[0108] The regression model is established by a polynomial regression method, and the regression model is provided with calibration model parameters, and the calibration model is obtained by pre-training.

[0109] The regression model can be established by establishing a mapping function, which can be the following formula: SpO2 = K*RR + b

[0110] Where SpO2 is the blood oxygen value, RR is a characteristic obtained by the camera and calculated based on the pulse wave signal (ratio of ratios), k is the cross-band absorption difference correction factor, and b is the red band reference value. In the physiological model, k reflects the difference in hemoglobin absorption between the two bands used, and b reflects the difference in red band absorption between oxygenated and deoxygenated hemoglobin. Both k and b are empirical values ​​derived from fitting a large amount of data and serve as a medium to link RR to the actual blood oxygen value.

[0111] Please further refer to Figure 8. The different light spots in Figure 8 represent different subjects. The relative pulse wave amplitude characteristics of each subject were fitted with their blood oxygen levels to obtain a correlation value between the relative pulse wave amplitude characteristics and their blood oxygen levels. In this embodiment, the fitting results indicate a linear correlation of 0.76 between the relative pulse wave amplitude characteristics and their blood oxygen levels. Based on this fitting result, the calibration model parameters of this embodiment were set to 0.76. With these calibration model parameters, Figure 8 shows that the blood oxygen levels obtained using the non-contact blood oxygen monitoring device and method of the present invention closely match the true blood oxygen levels. Of course, different calibration model parameters may be obtained in different calibration environments, such as under different individuals and lighting conditions.

[0112] Finally, the preferred embodiment of the non-contact blood oxygen monitoring device of the present invention displays the video of the continuous image frames of the human skin tissue obtained by the optical imaging module 100 and the blood oxygen value monitoring results obtained by the blood oxygen calibration module on the UI interface module 400 for subsequent use.

[0113] The present invention provides a non-contact blood oxygen monitoring device and method using a notch filter, which is used for non-contact real-time blood oxygen saturation monitoring. The device includes an optical imaging module, an image analysis module, a blood oxygen calibration module, and a user interface module. The optical imaging module is configured to capture a video containing continuous image frames of human skin tissue and a signal containing physiological information of human skin tissue. The image analysis module is configured to receive and analyze the signal containing physiological information of human skin tissue acquired from the optical imaging module in real time, extracting pulse wave relative amplitude features from the signal. The blood oxygen calibration module is configured to calibrate blood oxygen saturation based on the pulse wave relative amplitude features extracted by the image analysis module and a calibration model to obtain blood oxygen monitoring results. The UI interface module is configured to display the video containing continuous image frames of human skin tissue and the blood oxygen monitoring results on a UI interface. By adding a notch filter to the RGB camera, the optical imaging module of the present invention can selectively filter out interfering signals, thereby improving the accuracy of blood oxygen saturation estimation. The image analysis module further extracts pulse wave relative amplitude features from the signal acquired by the optical imaging module. The blood oxygen calibration module establishes a regression model using polynomial regression, calibrates the signal acquired by the image analysis module, obtains the final blood oxygen value detection result, and displays it on the UI display module. This invention uses a non-contact method to estimate blood oxygen saturation, achieving improved contactless blood oxygen monitoring for newborns. It uses an RGB camera sensor with a notch filter to capture and analyze images. It also combines optical principles and physiological models. By deeply understanding how light propagates in blood and the principles of hemodynamics, it better understands and estimates blood oxygen saturation data, obtaining a measurement value consistent with the actual blood oxygen saturation. This greatly improves the estimation model's robustness to different indoor lighting conditions.

[0114] It should be understood that the application of the present invention is not limited to the above examples. For those skilled in the art, improvements or changes can be made based on the above description. All these improvements and changes should fall within the scope of protection of the claims attached to the present invention.

Claims

1. A non-contact blood oxygen monitoring device improved by notch filter, used for non-contact real-time blood oxygen value monitoring of blood oxygen saturation, characterized in that: The device consists of an optical imaging module, an image analysis module, a blood oxygen calibration module and a UI interface module; A notch filter for obtaining an available wavelength band is provided in the optical imaging module, and the optical imaging module is used to obtain a video containing continuous image frames of human skin tissue and a signal containing physiological information of human skin tissue; The image analysis module is used to receive and analyze in real time the signal containing the physiological information of human skin tissue acquired from the optical imaging module, and extract the relative amplitude feature of the pulse wave therefrom; The blood oxygen calibration module is used to calibrate the blood oxygen saturation according to the pulse wave relative amplitude characteristics extracted by the image analysis module and the calibration model to obtain a blood oxygen value monitoring result; The UI interface module is used to display the video of the continuous image frames of the human skin tissue and the blood oxygen value monitoring results on the UI interface.

2. The non-contact blood oxygen monitoring device improved by using a notch filter according to claim 1, characterized in that: The optical imaging module comprises: a light source emitting unit; The light source emitting unit is used to provide a continuous and stable spectrum; The light source emitting unit comprises: a full-spectrum LED light source regulated by rare earth phosphor and a first polarizer; The full-spectrum LED light source regulated by the rare earth phosphor is used to generate a continuous spectrum; The first polarizer is placed between the front side of the full-spectrum LED light source controlled by the rare earth phosphor and the human skin tissue, and is used to convert the continuous spectrum photons generated by the full-spectrum LED light source controlled by the rare earth phosphor into photons with a single vibration direction.

3. The non-contact blood oxygen monitoring device improved by using a notch filter according to claim 2, characterized in that: The optical imaging module further includes: a camera sensor unit, the camera sensor unit being used to monitor photons in the wavelength bands of 450-583nm and 605-700nm; The camera sensor unit comprises: a second polarizer, a camera and an optical sensor; The second polarizer is arranged on the front side of the camera, and the second polarizer is placed between the camera and the human skin tissue, and the second polarizer is used to ensure that the camera obtains a diffuse reflection signal containing physiological information; The camera is a broadband RGB camera; The notch filter is arranged at the front end of the camera to block the photons in the 583nm-605nm band and enhance the independence of the photons in the 450-583nm and 605-700nm bands; The camera is used to focus photons between the 450-583nm and 605-700nm bands onto the optical sensor; The optical sensor is used to convert the photons between the 450-583nm and 605-700nm bands into the signal containing the physiological information of human skin tissue, and then convert the signal containing the physiological information of human skin tissue into continuous image frames of the human skin tissue.

4. The non-contact blood oxygen monitoring device improved by using a notch filter according to claim 3, characterized in that: The photons of the continuous spectrum emitted by the full-spectrum LED light source controlled by the rare earth phosphor are irradiated onto the human skin tissue through the first polarizer, and the photons reflected after being absorbed and scattered by the human skin tissue enter the camera through the second polarizer.

5. The non-contact blood oxygen monitoring device improved by using a notch filter according to claim 1, characterized in that: The image analysis module includes a multi-spectral pulse wave signal extraction unit; The multi-spectral pulse wave signal extraction unit is used to extract pulse wave signals of multiple bands in the time domain from the video containing continuous image frames of human skin tissue, perform detrending and bandpass filtering on the pulse wave signals of the multiple bands in the time domain, filter out non-pulse wave signals in the pulse wave signals of the multiple bands in the time domain, and obtain RGB three-channel pulse wave signals; A green light pulse wave signal and a red light pulse wave signal are extracted from the RGB three-channel pulse wave signal.

6. The non-contact blood oxygen monitoring device improved by using a notch filter according to claim 5, characterized in that: The image analysis module includes: a pulse wave relative amplitude estimation unit; The pulse wave relative amplitude estimation unit is used to extract the pulse wave relative amplitude characteristics of green light and red light from the green light pulse wave signal and the red light pulse wave signal obtained by the multi-spectral pulse wave signal extraction unit; The relative amplitude characteristic of the green and red light pulse waves is the ratio of the peak-to-peak value of the red light pulse wave signal to the peak-to-peak value of the green light pulse wave signal in the same pulsation cycle; or, The ratio of the amplitude of the Fourier transform of the red light pulse wave signal corresponding to the frequency point of the heart rate to the amplitude of the Fourier transform of the green light pulse wave signal corresponding to the frequency point of the heart rate.

7. The non-contact blood oxygen monitoring device improved by using a notch filter according to claim 6, characterized in that: The ratio of the peak-to-peak value of the red light pulse wave signal to the peak-to-peak value of the green light pulse wave signal in the same pulsation period is obtained by: The peak and trough of the red light pulse wave signal and the peak and trough of the green light pulse wave signal are respectively monitored by a peak-to-peak value monitoring algorithm; According to the monitored peaks and troughs of the red light pulse wave signal and the peaks and troughs of the green light pulse wave signal, the difference between the peaks and troughs of the red light pulse wave signal and the difference between the peaks and troughs of the green light pulse wave signal are calculated, and the relative amplitude characteristics of the red light and green light pulse wave signals are obtained by performing a ratio operation on the two calculated differences.

8. The non-contact blood oxygen monitoring device improved by using a notch filter according to claim 1, characterized in that: The blood oxygen calibration module includes: a regression model, which establishes a functional relationship between the relative amplitude characteristics of the pulse wave and the reference blood oxygen value for continuous dynamic estimation of blood oxygen saturation; The regression model is established by polynomial regression method; The regression model is provided with calibration model parameters; The calibration model parameters are obtained through pre-training; The regression model and calibration model parameters are used to calibrate the relative amplitude characteristics of the input red and green light pulse wave signals to obtain the blood oxygen value monitoring result.

9. The non-contact blood oxygen monitoring device improved by using a notch filter according to claim 8, characterized in that: The image analysis module inputs the relative amplitude characteristics of the red and green pulse wave signals into the blood oxygen calibration module; The blood oxygen calibration module outputs the obtained blood oxygen value monitoring result to the UI interface.

10. A non-contact blood oxygen monitoring method using a notch filter improved for use in the device according to any one of claims 1 to 9, characterized in that: The method comprises: The optical imaging module acquires a video containing continuous image frames of human skin tissue and the signal containing physiological information of human skin tissue; The image analysis module receives and analyzes the signal containing the physiological information of human skin tissue obtained from the optical imaging module in real time, and extracts the relative amplitude characteristics of the pulse wave therefrom; The blood oxygen calibration module calibrates the blood oxygen saturation according to the relative amplitude characteristics of the pulse wave extracted by the image analysis module and the calibration model to obtain a blood oxygen value monitoring result; The UI interface module displays the video of the continuous image frames of the human skin tissue and the blood oxygen value monitoring results on the UI interface.

Images