Method and system for detecting cerebral oxygen metabolism based on near-infrared spectroscopy

Abstract

The present application relates to physiological spectrum data processing technical field, specifically to a kind of cerebral oxygen metabolism detection method and system based on near infrared spectrum.The method determines the main component in light attenuation signal, obtains low-frequency blood oxygen regulation component reflecting user physiological activity or metabolic change after being eliminated.Based on the characteristics between low-frequency blood oxygen regulation component and main signal component, pulse conduction blood oxygen component is obtained using modified Lambert-Beer law.Based on low-frequency blood oxygen regulation component and pulse conduction blood oxygen component, the asymmetry between left and right brain can be analyzed, risk comprehensive index is determined using signal difference between the two components, and then metabolic early warning signal is effectively fed back.The present application obtains the characteristic signal of cerebral blood vessel in blood flow regulation and pulse conduction by effective signal decomposition method, avoids the interference of other factors on information, and obtains accurate early warning result.

Description

Technical Field

[0001] The present invention relates to the technical field of physiological spectrum data processing, and in particular to a method and system for detecting brain oxygen metabolism based on near-infrared spectroscopy. Background Art

[0002] Near-infrared spectroscopy is a non-invasive, continuous method for monitoring brain tissue blood oxygen saturation. Clinical monitoring equipment can use this method to monitor the user's brain oxygen metabolism data and generate warning signals when the data is abnormal, alerting medical staff to the patient's condition. However, during the actual monitoring process, due to the user's own emotions or physical factors, systemic physiological fluctuations will occur. These fluctuations will cause the blood oxygen signals of both cerebral hemispheres to change significantly and synchronously, forming common-mode interference. This common-mode interference will mask the relatively weak local blood oxygen asymmetry signal caused by cerebral vasospasm, which will make the monitoring equipment unable to effectively distinguish the true asymmetry caused by local vascular lesions in the complex physiological background, resulting in the inability to effectively feedback warning signals. Summary of the Invention

[0003] In order to solve the technical problem that the existing technology cannot enable monitoring equipment to effectively distinguish the warning signal of brain blood oxygen asymmetry, the purpose of the present invention is to provide a brain oxygen metabolism detection method and system based on near-infrared spectroscopy. The technical solution adopted is as follows:

[0004] The present invention proposes a method for detecting brain oxygen metabolism based on near-infrared spectroscopy, the method comprising:

[0005] Using a near-infrared spectroscopy device to obtain light attenuation signals at the left and right foreheads of the user;

[0006] Obtaining a main signal component based on the signal performance of the light attenuation signal in each cardiac cycle; removing the main signal component of the light attenuation signal to obtain a low-frequency blood oxygen regulation component; and using the low-frequency blood oxygen regulation component and the main signal component as a calculation basis for a modified Lambert-Beer law to obtain a pulse-conducted blood oxygen component;

[0007] The DTW algorithm was used to obtain the first morphological difference and time difference of the low-frequency blood oxygen regulation component between the left and right foreheads; the sample entropy difference of the low-frequency blood oxygen regulation component between the left and right foreheads was calculated; and the DTW algorithm was used to obtain the second morphological difference of the pulse-conducted blood oxygen component between the left and right foreheads;

[0008] The first morphological difference, time difference and sample entropy difference are counted to obtain a low-frequency regulation asymmetry feature; a risk comprehensive index is obtained based on the low-frequency regulation asymmetry feature and the second morphological difference; and a metabolic warning signal is fed back based on the risk comprehensive index.

[0009] Furthermore, the cardiac cycle acquisition method includes:

[0010] The user's arterial blood pressure waveform is obtained; and the cardiac cycle is divided according to a change trend of the systolic pressure in the arterial blood pressure waveform.

[0011] Furthermore, a peak detection algorithm is used to identify the systolic pressure peak moment, and the systolic pressure peak moment is used as a segmentation point to segment the signal timing range into multiple initial cardiac cycles; the average length of all initial cardiac cycles is used as the length of the cardiac cycle.

[0012] Furthermore, the method for obtaining the main signal components includes:

[0013] A single cardiac cycle optical signal segment in the optical attenuation signal of each cardiac cycle is obtained; and a plurality of single cardiac cycle optical signal segments are fused by an averaging algorithm to obtain the main signal component.

[0014] Furthermore, the method for obtaining the low-frequency blood oxygen regulation component includes:

[0015] Each single cardiac cycle optical signal segment is subtracted from the main signal component, and the obtained difference signal segments are spliced ​​into the low-frequency blood oxygen regulation component.

[0016] Furthermore, the DTW algorithm uses the Sakoe-Chiba band to constrain the search range during the process of searching for the optimal alignment path in the cost matrix.

[0017] Furthermore, the method for obtaining the low-frequency regulation asymmetric feature includes:

[0018] The normalized first morphological difference, time difference, and sample entropy difference are weightedly summed using preset weights to obtain the low-frequency regulation asymmetric feature.

[0019] Furthermore, the method for obtaining the comprehensive risk index includes:

[0020] The second morphological difference is normalized and then multiplied by the low-frequency regulation asymmetry feature to obtain the comprehensive risk index.

[0021] Furthermore, the risk comprehensive index sequence at each moment is statistically calculated; after the risk comprehensive index sequence is denoised, the denoised risk comprehensive index at the real time moment is compared with a preset threshold, and the metabolic warning signal is fed back according to the comparison result.

[0022] The present invention also proposes a brain oxygen metabolism detection system based on near-infrared spectroscopy, comprising a memory, a processor, and a computer program stored in the memory and runnable on the processor. When the processor executes the computer program, it implements any one of the steps of the brain oxygen metabolism detection method based on near-infrared spectroscopy.

[0023] The present invention has the following beneficial effects:

[0024] To avoid the impact of common-mode interference on the original spectral signal, the present invention does not directly process the original spectral signal. Instead, it decomposes the original light attenuation signal into a low-frequency blood oxygen regulation component and a pulse-conducted blood oxygen component through signal decomposition. Because different blood oxygen proteins have different light absorption characteristics, the present invention first determines the main component in the light attenuation signal, that is, the main component reflecting the changes in arterial blood caused by the pulse, and then eliminates it to obtain the low-frequency blood oxygen regulation component that reflects the user's physiological activity or metabolic changes. Based on the characteristics between the low-frequency blood oxygen regulation component and the main signal component, the modified Lambert-Beer law can be used to further analyze the relative concentration changes of oxyhemoglobin and deoxyhemoglobin, and then obtain the pulse-conducted blood oxygen component. Based on the low-frequency blood oxygen regulation component and the pulse-conducted blood oxygen component, the asymmetry between the left and right brain can be analyzed, and the signal difference between the two components can be used to determine the comprehensive risk index, thereby effectively feeding back metabolic warning signals. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0026] Figure 1 A flowchart of a method for detecting brain oxygen metabolism based on near-infrared spectroscopy is provided in accordance with one embodiment of the present invention. DETAILED DESCRIPTION

[0027] To further illustrate the technical means and effectiveness of the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, describes in detail the specific implementation, structure, features, and effectiveness of a near-infrared spectroscopy-based brain oxygen metabolism detection method and system proposed by the present invention. In the following description, different references to "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.

[0028] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0029] It should be noted that the embodiments of the present invention are directed to the early warning system of the existing clinical monitoring equipment, and the method or system proposed in the embodiments of the present invention can be directly applied to the early warning system of the clinical monitoring equipment.

[0030] The following describes in detail a method and system for detecting brain oxygen metabolism based on near-infrared spectroscopy provided by the present invention with reference to the accompanying drawings.

[0031] See also Figure 1 , which shows a flow chart of a method for detecting brain oxygen metabolism based on near-infrared spectroscopy provided by one embodiment of the present invention, the method comprising:

[0032] Step S1: using a near-infrared spectroscopy device to obtain light attenuation signals at the left and right forehead positions of the user respectively.

[0033] In this embodiment of the present invention, the probes of the near-infrared spectroscopy device are placed symmetrically on the left and right sides of the patient's forehead to obtain light attenuation signals from the left and right sides of the brain. Light attenuation signals are spectral signals that are subject to common-mode interference caused by physiological and psychological factors of the user. Therefore, the information contained in the original light attenuation signals is relatively complex. Directly comparing the data to generate warning information can lead to inaccurate warnings.

[0034] In the embodiment of the present invention, at the real time moment, the data of the previous 60 seconds is selected as the time period to be analyzed at the real time moment, that is, the light attenuation signal within the time period to be analyzed is obtained, and then the warning signal at the real time moment is determined.

[0035] Step S2: Obtain the main signal component based on the signal performance of the light attenuation signal in each cardiac cycle; remove the main signal component of the light attenuation signal to obtain the low-frequency blood oxygen regulation component; use the low-frequency blood oxygen regulation component and the main signal component as the calculation basis for the modified Lambert-Beer law to obtain the pulse conduction blood oxygen component.

[0036] For the problem of mixed information in the original light attenuation signal, the embodiment of the present application considers that the high-frequency pulse wave in the light attenuation signal belongs to the main component of the signal, which reflects the arterial blood change caused by the pulse, and in order to obtain effective monitoring data, more attention should be paid to the low-frequency component in the light attenuation signal, because the low-frequency component is the data reflecting physiological activity and metabolic change. Further considering that the human body has a heart cycle during the monitoring process, the main signal component can be obtained according to the signal performance of the light attenuation signal in each heart cycle. Further removing the main signal component of the light attenuation signal, the low-frequency blood oxygen regulation component can be obtained. The low-frequency blood oxygen regulation component reflects the cerebral blood flow autoregulation process driven by the slow wave of blood pressure, metabolic activity and the like, and the signal embodies the data of the brain blood in the functional regulation dimension.

[0037] The present application further extracts the signal to obtain the pulse conduction blood oxygen component, which reflects the vascular mechanical response characteristics after the heart beat is conducted to the cortex through the upstream blood supply artery. In order to obtain the pulse conduction blood oxygen component, the low-frequency blood oxygen regulation component and the main signal component can be used as the operation basis of the modified Lambert-Beer law to obtain the pulse conduction blood oxygen component through the law. The core principle of the law is that when the near-infrared light passes through the tissue, the degree of attenuation is affected by the concentration of the internal chromophore, which is mainly oxyhemoglobin HbO2 and deoxyhemoglobin Hb. Because the two have different light absorption abilities at specific wavelengths, this provides the basis for distinguishing them. Therefore, the modified Lambert-Beer law can be used to solve based on the main signal component and the low-frequency blood oxygen regulation component, and then the pulse conduction blood oxygen component is obtained. It should be noted that the modified Lambert-Beer law will obtain two signals respectively, one of which represents the relative amount of change of oxyhemoglobin concentration with time, and the other represents the relative amount of change of deoxyhemoglobin concentration with time, and the essence of the pulse conduction blood oxygen component in the embodiment of the present application is to express the relative concentration change of oxyhemoglobin and deoxyhemoglobin, so the two signals obtained are subtracted to obtain the pulse conduction blood oxygen component.

[0038] Preferably, in the embodiment of the present application, considering that the arterial blood pressure waveform has good heart cycle performance, the arterial blood pressure waveform of the user can be obtained; the heart cycle is divided according to the change trend of the systolic pressure in the arterial blood pressure waveform. Specifically, in one embodiment of the present application, the method for determining the heart cycle comprises:

[0039] The systolic pressure peak time is identified by using the peak detection algorithm, and the systolic pressure peak time is used as a segmentation point to divide the signal time sequence range into a plurality of initial heart cycles. Because the lengths of the initial heart cycles obtained by segmentation may not be consistent, which is not conducive to subsequent alignment analysis, the average length of all initial heart cycles is used as the length of the heart cycle.

[0040] Preferably, in an embodiment of the present invention, the method for obtaining the main signal components includes:

[0041] Obtain a single cardiac cycle optical signal segment from the optical attenuation signal for each cardiac cycle; fuse the multiple single cardiac cycle optical signal segments using an averaging algorithm to obtain the primary signal component. It should be noted that in this embodiment of the present invention, to reduce computational complexity, 30 single cardiac cycle optical signal segments are selected for fusion, starting from the real time.

[0042] It should be noted that if the data on the signal is insufficient to fill the timing range of the cardiac cycle during the division of the single cardiac cycle optical signal segment, interpolation is used to supplement it. The specific technical means are well known to those skilled in the art and will not be described in detail here.

[0043] It should be noted that in some embodiments of the present invention, the initial cardiac cycle length is averaged to obtain a fixed-length cardiac cycle length. The lengths of the single cardiac cycle optical signal segments obtained by segmenting the cardiac cycle are also the same, and can be directly aligned and averaged. In other embodiments of the present invention, in order to further preserve the original cycle information, the initial cardiac cycle length may not be averaged, and the division result of the initial cardiac cycle may be directly used as the division result of the cardiac cycle, thereby corresponding to single cardiac cycle optical signal segments that may have different lengths. During the alignment process, a signal alignment method such as a dynamic time warping algorithm may be used to determine the correspondence between signal points, and then average and fuse them. The specific algorithm is a technical means well known to those skilled in the art and will not be described in detail here.

[0044] Preferably, in an embodiment of the present invention, based on the main signal component obtained above, each single cardiac cycle optical signal segment can be directly subtracted from the main signal component, and the obtained difference signal segments can be spliced ​​into the low-frequency blood oxygen regulation component.

[0045] It should be noted that in some embodiments of the present invention, the time range of the main signal component obtained is a single cardiac cycle. Therefore, when performing the modified Lambert-Beer law operation, the difference signal segment and the main signal component can be selected as the basis for operation to obtain the mechanical response characteristics reflected in each cardiac cycle, and then form the pulse-conducted blood oxygen component by splicing.

[0046] Step S3: Use the DTW algorithm to obtain the first morphological difference and time difference of the low-frequency blood oxygen regulation component between the left forehead and the right forehead; calculate the sample entropy difference of the low-frequency blood oxygen regulation component between the left forehead and the right forehead; use the DTW algorithm to obtain the second morphological difference of the pulse-conducted blood oxygen component between the left forehead and the right forehead.

[0047] The low-frequency blood oxygen regulation component and the pulse-conducted blood oxygen component obtained in the above steps can be compared between the left and right hemispheres of the brain to analyze their asymmetry.

[0048] When unilateral cerebral vasospasm occurs, the autoregulatory function of blood flow in the affected hemisphere becomes unbalanced, resulting in deviations in the morphology, timing, and dynamic pattern of the blood oxygen slow wave. Therefore, this embodiment of the present invention utilizes the dynamic time warping (DTW) algorithm to obtain the first morphological difference and temporal difference of the low-frequency blood oxygen regulation component between the left and right foreheads. The sample entropy difference of the low-frequency blood oxygen regulation component between the left and right foreheads is then calculated. Specifically, the first morphological difference analyzes the morphological deviation of the blood oxygen slow wave, while the temporal difference analyzes the temporal deviation. Sample entropy can characterize the dynamic changes within the data, and therefore, the sample entropy difference is used to characterize the deviation of the dynamic pattern.

[0049] Furthermore, unilateral cerebral vasospasm can alter the elasticity and blood flow resistance of the affected blood vessels, leading to distortion of the transmitted cardiac pulse waveform. Therefore, a DTW algorithm is used to obtain the second-order morphology difference in the blood oxygen component of the pulse transmission between the left and right foreheads. This second-order morphology difference can characterize detailed changes in the signal between the left and right hemispheres, such as rising edge slope and peak position.

[0050] It should be noted that to avoid errors during signal comparison, all component signals must be normalized before performing difference calculations. This embodiment of the present invention uses Z-normalization to perform normalization within each dimension, eliminating dimensionality effects and inherent data differences, allowing comparison calculations to focus solely on the signal trends themselves.

[0051] Preferably, in the embodiment of the present invention, in the process of searching for the optimal alignment path in the cost matrix, the DTW algorithm uses the Sakoe-Chiba constraint band to constrain the search range.

[0052] It should be noted that because blood oxygen changes caused by cerebral vasospasm often exhibit nonlinear temporal delays or distortions, the DTW algorithm can be used to nonlinearly align the two time series to be compared. The DTW algorithm is a well-known technical approach for those skilled in the art. During the algorithm execution, a distance matrix, also known as a cost matrix, is constructed. The element values ​​in the matrix are the squares of the differences between the corresponding two elements. In the process of finding the optimal alignment path from the starting point to the end point of the cost matrix, in order to avoid transitional time distortions that do not conform to physiological logic, the Sakagami-Chiba constraint band can be used to limit the path search range to a fixed width around the main diagonal of the matrix. After obtaining the optimal alignment path, all element values ​​on the optimal alignment path are accumulated to obtain the first morphological difference. This feature quantifies the residual difference in shape between the two blood oxygen slow waves after achieving optimal time alignment. The larger the value, the more significant the morphological difference. The time difference is the average of the absolute values ​​of the time index differences of all matching point pairs on the optimal alignment path. The time difference quantifies the average degree of advance or lag in time between the left and right hemisphere blood oxygen slow waves.

[0053] It should be noted that the method for obtaining the second form difference is the same as that for the first form difference, and will not be described in detail.

[0054] It should be noted that methods for obtaining sample entropy are well-known to those skilled in the art. In this embodiment, the embedding dimension is set to 2, and the similarity tolerance is set to 0.2 times the standard deviation of the sequence to be analyzed. The specific method for calculating sample entropy is well-known to those skilled in the art and will not be detailed here.

[0055] Sample entropy measures the regularity and predictability of time series. A healthy cerebral blood flow regulation system exhibits complex, nonlinear dynamic characteristics, while pathological conditions may cause its regulation pattern to become rigid (decreased entropy) or chaotic (increased entropy). In this embodiment of the present invention, the absolute value of the difference in sample entropy between the left and right hemispheres' low-frequency blood oxygen regulation components is used as the sample entropy difference.

[0056] Step S4: Count the first morphological difference, time difference and sample entropy difference to obtain the low-frequency regulation asymmetry feature; obtain the risk comprehensive index based on the low-frequency regulation asymmetry feature and the second morphological difference; and feedback the metabolic warning signal based on the risk comprehensive index.

[0057] After the above processing, the first morphological difference, the time difference, and the sample entropy difference representing the low-frequency blood oxygen regulation component difference between the left and right hemispheres are obtained; the second morphological difference representing the pulse conduction blood oxygen component difference between the left and right hemispheres is obtained. The four asymmetry characteristics representing different physiological dimensions can be comprehensively counted to generate a single and continuous risk comprehensive index, and then the risk comprehensive index is used to feed back a warning signal.

[0058] It should be noted that in the process of obtaining the risk comprehensive index in the embodiment of the present application, the four features obtained are respectively normalized by using range standardization in their respective dimensions to eliminate the dimension and limit the value range to 0 to 1, so as to facilitate the fusion analysis of the four features.

[0059] Since the first morphological difference, the time difference, and the sample entropy difference represent the slow wave asymmetry characteristics of blood oxygen, the three characteristics can be first counted to obtain the low-frequency regulation asymmetry feature. The feature is intended to quantify the overall imbalance degree of the cerebral blood flow autoregulation function.

[0060] Preferably, in the embodiment of the present application, the method for obtaining the low-frequency regulation asymmetry feature comprises:

[0061] The normalized first morphological difference, the time difference, and the sample entropy difference are weighted and summed by using preset weights to obtain the low-frequency regulation asymmetry feature. It should be noted that the weights can be determined in advance according to clinical data to reflect the relative importance of different features in representing the regulation dysfunction. In the embodiment of the present application, the weight of the first morphological difference is set to 0.3, the weight of the time difference is set to 0.3, and the weight of the sample entropy difference is set to 0.4.

[0062] Preferably, in embodiments of the present invention, considering that a hemodynamically significant cerebral vasospasm should exhibit abnormalities at both the functional level (blood flow regulation) and the physical level (pulse conduction), it is necessary to effectively fuse the low-frequency modulation asymmetry feature and the second morphological difference. In embodiments of the present invention, the second morphological difference is normalized and then multiplied with the low-frequency modulation asymmetry feature to obtain the comprehensive risk index. By fusing these two features through multiplication, when there is a significant imbalance in cerebral blood flow regulation and a significant change in vascular mechanical properties, the fused comprehensive risk index is significantly high, significantly reflecting the magnitude of the risk. Furthermore, if the feature in any dimension is low, for example, if slow waves are asynchronized due to only a brief signal artifact, but the pulse wave morphology remains symmetrical, the comprehensive risk index obtained through multiplication will also remain low, preventing brief noise fluctuations from affecting the early warning judgment. By quantifying the comprehensive risk index, non-cerebral vasospasm events that exhibit asymmetry only in a single dimension can be effectively filtered out, thereby directly utilizing the unique advantage of separating the signal into two orthogonal physiological domains in step S2.

[0063] Preferably, in an embodiment of the present invention, at real time, a risk comprehensive index sequence at each moment is counted; after the risk comprehensive index sequence is denoised, the denoised risk comprehensive index at the real time moment is compared with a preset threshold, and the metabolic warning signal is fed back according to the comparison result. In an embodiment of the present invention, the denoising method is to select a moving average filter with a window length of 120 seconds for smoothing to eliminate short-term noise interference and focus on the continuous risk change trend. The preset threshold can be set to two, a warning threshold and an alarm threshold, that is, if the risk comprehensive index is greater than the warning threshold, the warning signal is mainly used to remind medical staff to pay attention to the patient's state; if the risk comprehensive index is greater than the alarm threshold, the feedback warning signal is mainly used to remind medical staff to effectively intervene in the patient's state; that is, the alarm threshold is greater than the warning threshold, and if the risk comprehensive index is less than the warning threshold, no warning signal is fed back. It should be noted that the two thresholds can be set by medical staff according to the patient's state or medical needs, and are not limited or elaborated here.

[0064] In summary, the present invention determines the main components in the light attenuation signal, removes them, and obtains a low-frequency blood oxygen regulation component that reflects the user's physiological activities or metabolic changes. Based on the characteristics between the low-frequency blood oxygen regulation component and the main signal component, the pulse conduction blood oxygen component is obtained using the modified Lambert-Beer law. Based on the low-frequency blood oxygen regulation component and the pulse conduction blood oxygen component, the asymmetry between the left and right brain can be analyzed, and the signal difference between the two components can be used to determine the comprehensive risk index, thereby effectively feeding back a metabolic warning signal. The present invention obtains characteristic signals of cerebral blood vessels in terms of blood flow regulation and pulse conduction through an effective signal decomposition method, thereby avoiding interference of other factors on the information and obtaining accurate warning results.

[0065] Based on the same inventive concept, the present invention proposes a brain oxygen metabolism detection system based on near-infrared spectroscopy, including a memory, a processor, and a computer program stored in the memory and runnable on the processor. When the processor executes the computer program, it implements any one of the steps of the brain oxygen metabolism detection method based on near-infrared spectroscopy.

[0066] It should be noted that the order in which the embodiments of the present invention are described above is for illustrative purposes only and does not necessarily represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0067] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on the differences from other embodiments.

Claims

1. A method for detecting brain oxygen metabolism based on near-infrared spectroscopy, characterized in that: The method comprises: Using a near-infrared spectroscopy device to obtain light attenuation signals at the left and right foreheads of the user; Obtaining a main signal component based on the signal performance of the light attenuation signal in each cardiac cycle; removing the main signal component of the light attenuation signal to obtain a low-frequency blood oxygen regulation component; using a modified Lambert-Beer law to solve the main signal component and the low-frequency blood oxygen regulation component to obtain a relative amount representing the change in oxygenated hemoglobin concentration over time and a relative amount representing the change in deoxygenated hemoglobin concentration over time; subtracting the relative amount representing the change in oxygenated hemoglobin concentration over time from the relative amount representing the change in deoxygenated hemoglobin concentration over time to obtain a pulse-conducted blood oxygen component; The DTW algorithm was used to obtain the first morphological difference and time difference of the low-frequency blood oxygen regulation component between the left and right foreheads; the sample entropy difference of the low-frequency blood oxygen regulation component between the left and right foreheads was calculated; and the DTW algorithm was used to obtain the second morphological difference of the pulse-conducted blood oxygen component between the left and right foreheads; The first morphological difference, time difference and sample entropy difference are counted to obtain a low-frequency regulation asymmetry feature; a risk comprehensive index is obtained based on the low-frequency regulation asymmetry feature and the second morphological difference; and a metabolic warning signal is fed back based on the risk comprehensive index.

2. The method for detecting brain oxygen metabolism based on near-infrared spectroscopy according to claim 1, characterized in that: The cardiac cycle acquisition method includes: The user's arterial blood pressure waveform is obtained; and the cardiac cycle is divided according to a change trend of the systolic pressure in the arterial blood pressure waveform.

3. The method for detecting brain oxygen metabolism based on near-infrared spectroscopy according to claim 2, characterized in that: A peak detection algorithm is used to identify the systolic pressure peak moment, and the systolic pressure peak moment is used as a segmentation point to segment the signal time series range into multiple initial cardiac cycles; the average length of all initial cardiac cycles is used as the length of the cardiac cycle.

4. The method for detecting brain oxygen metabolism based on near-infrared spectroscopy according to claim 1, characterized in that: The method for obtaining the main signal components includes: A single cardiac cycle optical signal segment in the optical attenuation signal of each cardiac cycle is obtained; and a plurality of single cardiac cycle optical signal segments are fused by an averaging algorithm to obtain the main signal component.

5. The method for detecting brain oxygen metabolism based on near-infrared spectroscopy according to claim 4, characterized in that: The method for obtaining the low-frequency blood oxygen regulation component includes: Each single cardiac cycle optical signal segment is subtracted from the main signal component, and the obtained difference signal segments are spliced ​​into the low-frequency blood oxygen regulation component.

6. The method for detecting brain oxygen metabolism based on near-infrared spectroscopy according to claim 1, characterized in that: The DTW algorithm uses the Sakoe-Chiba band to constrain the search range during the process of searching for the optimal alignment path in the cost matrix.

7. The method for detecting brain oxygen metabolism based on near infrared spectroscopy according to claim 1, characterized in that: The method for obtaining the low-frequency regulation asymmetric feature includes: The normalized first morphological difference, time difference, and sample entropy difference are weightedly summed using preset weights to obtain the low-frequency regulation asymmetric feature.

8. The method for detecting brain oxygen metabolism based on near infrared spectroscopy according to claim 7, characterized in that: The method for obtaining the comprehensive risk index includes: The second morphological difference is normalized and then multiplied by the low-frequency regulation asymmetry feature to obtain the comprehensive risk index.

9. The method for detecting brain oxygen metabolism based on near infrared spectroscopy according to claim 1, characterized in that: The risk comprehensive index sequence at each moment is counted; after the risk comprehensive index sequence is denoised, the denoised risk comprehensive index at the real time moment is compared with a preset threshold value, and the metabolic warning signal is fed back according to the comparison result.

10. A brain oxygen metabolism detection system based on near-infrared spectroscopy, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the brain oxygen metabolism detection method based on near-infrared spectroscopy as described in any one of claims 1 to 9 are implemented.

Images