Multispectral-based rapid blood hct detection system

Abstract

This invention relates to the field of blood hemoglobin coagulation (HCT) detection technology, specifically to a multispectral-based rapid blood HCT detection system, comprising: acquiring absorbance data of blood samples at various wavelengths; based on the absorbance data, obtaining authenticity indicators, bubble interference indicators, and coagulation interference indicators for the absorbance data corresponding to each wavelength, thereby determining the data quality score of the absorbance data corresponding to each wavelength; constructing and solving a matrix equation based on a preset absorbance model including hemoglobin concentration and scattering background value, a scattering background value solving model, and a diagonal weight matrix composed of the data quality score to obtain the hemoglobin concentration; converting the hemoglobin concentration into an HCT value; and determining whether the HCT value is true and valid based on the data quality score. Using this invention, the detected HCT value is more accurate and clinically reliable.

Description

Technical Field

[0001] This invention relates to the field of blood HCT detection technology, and more specifically to a rapid blood HCT detection system based on multispectral imaging. Background Technology

[0002] Packed Cell Volume (PCV), also known as Hematocrit (HCT), refers to the volume ratio of red blood cells that settle in a given amount of anticoagulated whole blood after centrifugation. It is an important indicator that indirectly reflects the number, size, and volume of red blood cells. Combined with red blood cell count and hemoglobin content, HCT can be used to calculate the mean erythrocyte count, which is of significant reference value for the morphological classification of anemia.

[0003] Currently, optical-based HCT detection technology is widely used, but its measurement accuracy is severely challenged by various interference factors during actual testing. First, during optical detection, the instantaneous passage of air bubbles in the flow path can cause abrupt spikes in absorbance, while the slow coagulation of blood samples leads to a continuous drift in the absorbance baseline. These non-stationary interferences severely compromise the authenticity and stability of the measurement data. Second, in practical applications, multi-wavelength sensing systems may generate abnormal or low-quality signals at certain wavelengths due to individual differences or instantaneous fluctuations in components. If all wavelength data are assigned equal weight without distinction during data processing, the system cannot effectively identify and suppress the negative impact of these low-reliability data.

[0004] The aforementioned problems result in a systematic bias in the final calculation results that is difficult to correct, which severely restricts the accuracy and clinical reliability of HCT measurement results. Summary of the Invention

[0005] To address the measurement bias issues caused by factors such as air bubbles, coagulation, and scattering in existing blood hematocrit (HCT) detection technologies, the present invention aims to provide a rapid blood HCT detection system based on multispectral imaging. The specific technical solution adopted is as follows:

[0006] The acquisition module is used to acquire absorbance data of blood samples at various wavelengths;

[0007] The interference analysis module is used to obtain the authenticity index, bubble interference index and coagulation interference index of the absorbance data corresponding to each wavelength based on the absorbance data.

[0008] The scoring module is used to determine the data quality score of the absorbance data corresponding to each wavelength based on the authenticity index, the bubble interference index and the coagulation interference index.

[0009] The concentration analysis module is used to construct and solve matrix equations based on a preset absorbance model containing hemoglobin concentration and scattering background value, a scattering background value solution model, and a diagonal weight matrix composed of the data quality score, in order to obtain the hemoglobin concentration.

[0010] The conversion module is used to convert the hemoglobin concentration into an HCT value using a preset HCT conversion algorithm;

[0011] The judgment module is used to determine whether the HCT value is true and valid based on the data quality score.

[0012] Furthermore, after collecting the absorbance data of the blood sample at various wavelengths, the method further includes preprocessing the absorbance data. The preprocessing process includes:

[0013] The absorbance data was noise-reduced using a moving average smoothing method.

[0014] When missing values ​​are identified in the absorbance data, linear interpolation is used to fill in the missing values.

[0015] Furthermore, the acquisition of absorbance data for blood samples at various wavelengths includes:

[0016] Obtain fresh blood samples that have been anticoagulated and mixed.

[0017] The absorbance data of the blood sample are collected over time within the characteristic absorption band of hemoglobin, wherein the temporal resolution of the acquisition is matched with the dynamic sedimentation rate of the blood, and the total acquisition time covers the complete process of the blood sample from a uniformly mixed state to significant sedimentation.

[0018] Furthermore, the process of obtaining the authenticity index includes:

[0019] Based on the absorbance change rate at each sampling time point of the absorbance data at each wavelength, construct a sequence of absorbance change rate corresponding to each wavelength;

[0020] The average value of the correlation coefficients between the absorbance change rate sequence corresponding to any wavelength and the absorbance change rate sequences corresponding to all other wavelengths is used as an indicator of the authenticity of the absorbance data corresponding to any wavelength.

[0021] Furthermore, the process of obtaining the bubble interference index includes:

[0022] The coefficient of variation is obtained based on the absorbance data at any wavelength;

[0023] The absorbance data after drift-reduction processing is transformed in the frequency domain to obtain the FFT result. The amplitude of the FFT result is squared and then normalized to obtain the power spectral density. Based on the power spectral density, the ratio of high-frequency energy in the high-frequency range to full-frequency energy in the full-band range is calculated.

[0024] The bubble interference index corresponding to the absorbance data at any wavelength is determined based on the coefficient of variation and the ratio.

[0025] Furthermore, the process of obtaining the coefficient of variation includes:

[0026] For any sampling time point under any wavelength, a local time window of a preset width is defined with the sampling time point as the center;

[0027] Calculate the local coefficient of variation of the absorbance data within the local time window;

[0028] The process of obtaining the local variation coefficient is repeated to obtain the local variation coefficient corresponding to each sampling time point under any wavelength;

[0029] The mean of the local coefficients of variation is used as the coefficient of variation.

[0030] Furthermore, the process of obtaining the high-frequency range includes:

[0031] The frequency at which the power spectral density first shows an inflection point from high frequency to low frequency is marked as the starting frequency.

[0032] The preset multiple of the starting frequency is taken as the high frequency, and the frequency range between the high frequency and the maximum frequency of the power spectral density is taken as the high frequency interval.

[0033] Furthermore, the process of obtaining the coagulation interference index includes:

[0034] The absorbance data corresponding to any wavelength is divided into a predetermined number of absorbance data segments;

[0035] A linear fit is performed on each absorbance data segment to obtain the fitting slope corresponding to each absorbance data segment, thereby forming a slope sequence;

[0036] The mean of the absolute values ​​of the differences between adjacent fitted slopes in the slope sequence is calculated as the first nonstationarity index of coagulation interference.

[0037] The run number and expected run number of the absorbance data are obtained according to the run test algorithm, and then the second nonstationarity index of the coagulation interference is determined.

[0038] The coagulation interference index is determined based on the first non-stationarity index and the second non-stationarity index.

[0039] Furthermore, the process of obtaining the second nonstationarity index includes:

[0040] The second nonstationarity index is determined based on the standardized distance between the number of runs and the expected number of runs, wherein the second nonstationarity index is used to characterize the degree to which the number of runs deviates from the expected number of runs.

[0041] Furthermore, the process of obtaining the data quality score includes:

[0042] Based on the bubble interference index and the coagulation interference index, the interference weight value is determined, wherein the interference weight value is a normalized value, and the bubble interference index and the coagulation interference index are negatively correlated with the interference weight value.

[0043] Calculate the normalized value of the authenticity index, and then normalize the product of the normalized value and the interference weight value to obtain the data quality score.

[0044] The present invention has the following beneficial effects:

[0045] Absorbance data of blood samples were collected at various wavelengths. This data forms the basis for subsequent analysis.

[0046] Based on the absorbance data, the authenticity index, bubble interference index, and coagulation interference index of the absorbance data corresponding to each wavelength are obtained. These indices are used to quantify the reliability of the absorbance data at each wavelength.

[0047] Based on the authenticity index, the bubble interference index, and the coagulation interference index, a data quality score is determined for the absorbance data corresponding to each wavelength. The higher the data quality score, the more reliable the corresponding absorbance data.

[0048] Based on a preset absorbance model that includes hemoglobin concentration and scattering background value, a scattering background value solution model, and a diagonal weight matrix composed of the data quality scores, a matrix equation is constructed and solved to obtain the hemoglobin concentration. A data quality weighting mechanism is introduced in the hemoglobin concentration inversion process, using high-quality data as the core weight to effectively suppress interference from low-quality data, making the hemoglobin concentration value more accurate.

[0049] The hemoglobin concentration is converted into an HCT value using a preset HCT conversion algorithm. Obtaining the HCT value based on the hemoglobin concentration is a prior art technique.

[0050] The validity of the HCT value is determined based on the data quality score. The validity of the HCT value is verified to ensure the reliability of the output results. Attached Figure Description

[0051] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0052] Figure 1 A flowchart of the multispectral-based rapid blood HCT detection system provided in the first embodiment of the present invention;

[0053] Figure 2 A flowchart illustrating the process of obtaining the authenticity index provided in the second embodiment of the present invention;

[0054] Figure 3 A flowchart illustrating the process of obtaining the bubble interference index as provided in the third embodiment of the present invention;

[0055] Figure 4 This is a flowchart illustrating the process of obtaining the coefficient of variation according to the fourth embodiment of the present invention.

[0056] Figure 5 A flowchart illustrating the process of obtaining the high-frequency range provided in the fifth embodiment of the present invention;

[0057] Figure 6 A flowchart illustrating the process of obtaining coagulation interference indicators provided in the sixth embodiment of the present invention;

[0058] Figure 7 This is a flowchart illustrating the process of obtaining a data quality score as provided in the seventh embodiment of the present invention. Detailed Implementation

[0059] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of the multispectral-based rapid blood HCT detection system proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0060] Unless otherwise defined, 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 pertains.

[0061] The specific solution of the multispectral-based rapid blood HCT detection system provided by the present invention will be described in detail below with reference to the accompanying drawings.

[0062] Please see Figure 1 The diagram illustrates a flowchart of a multispectral-based rapid blood HCT detection system provided in the first embodiment of the present invention, the system comprising:

[0063] The acquisition module 101 is used to acquire absorbance data of blood samples at various wavelengths.

[0064] To ensure the accuracy and reliability of hemoglobin concentration inversion results, a standardized data acquisition and processing workflow needs to be established. This workflow begins with standardized sample preparation: fresh blood samples treated with anticoagulation are selected, thoroughly mixed using a vortex mixer, and then slowly injected into the detection cell to avoid air bubble formation. During the fiber optic spectrometer data acquisition stage, the wavelength range should cover the characteristic absorption band of hemoglobin. Based on the timescale of dynamic processes such as erythrocyte sedimentation, the sampling interval is set to Δt = 1 second, and data is continuously acquired for 120 seconds to obtain high temporal resolution absorbance data at 120 time points. After sample injection, the sample should be allowed to stand for 3 seconds before starting acquisition to eliminate the influence of fluid disturbance. The entire acquisition process should be conducted in a constant temperature environment (25±1℃). The obtained raw absorbance data will be stored as a data matrix where rows represent time points and columns represent wavelengths. Before subsequent core calculations, the matrix needs to be preprocessed: first, the moving average method is used to smooth the data, then the data integrity is checked, and any missing values ​​are filled using linear interpolation, thus laying a high-quality data foundation for subsequent accurate inversion.

[0065] The interference analysis module 102 is used to obtain the authenticity index, bubble interference index and coagulation interference index of the absorbance data corresponding to each wavelength based on the absorbance data.

[0066] The authenticity index reflects the consistency of the trend of a certain wavelength data with the data of other wavelengths, while the bubble interference index and the coagulation interference index quantify the severity of the two specific interferences of bubbles and coagulation at that wavelength, respectively.

[0067] The process of obtaining the authenticity index will be described in detail in the second embodiment, and will not be repeated here.

[0068] The process of obtaining the bubble interference index will be described in detail in the third embodiment, and will not be repeated here.

[0069] The process of obtaining the coagulation interference index will be described in detail in the sixth embodiment, and will not be repeated here.

[0070] The scoring module 103 is used to determine the data quality score of the absorbance data corresponding to each wavelength based on the authenticity index, the bubble interference index and the coagulation interference index.

[0071] The data quality score is jointly determined by the authenticity index, the bubble interference index, and the coagulation interference index. The calculation model of the data quality score introduces dynamic weights to balance data reliability (reflected by the authenticity index) and interference sensitivity (reflected by the bubble interference index and the coagulation interference index), wherein the bubble interference index and the coagulation interference index serve as penalty items.

[0072] The process of obtaining the data quality score will be described in detail in the seventh embodiment, and will not be repeated here.

[0073] The higher the data quality score, the less the absorbance data corresponding to the current wavelength is affected by bubbles or coagulation, and the greater its reference weight in the subsequent calculation of the actual HCT content.

[0074] The concentration analysis module 104 is used to construct and solve matrix equations based on a preset absorbance model containing hemoglobin concentration and scattering background value, a scattering background value solution model, and a diagonal weight matrix composed of the data quality score, in order to obtain the hemoglobin concentration.

[0075] Traditional methods use an averaging algorithm for all wavelength data, which cannot eliminate the influence of abnormal data, such as bubbles and blood clotting, on the calculation results.

[0076] In spectral detection, the detector can only receive light rays propagating along the original optical path. Scattering effects cause some photons to deviate from the optical path and fail to reach the detector, resulting in a decrease in the measured transmittance. Consequently, the apparent molar absorbance is significantly higher than the actual absorbance of hemoglobin. Therefore, a scattering background value S is introduced to represent the interference caused by scattering.

[0077] According to the Lambert-Beer law and light scattering theory, absorbance is composed of hemoglobin absorption and scattering background; for each wavelength, the absorbance model is as follows:

[0078] ;

[0079] Among them, the Represents the current wavelength Corresponding absorbance data, Represents hemoglobin at the current wavelength The molar absorptivity is given below, where c represents the hemoglobin concentration. The optical path length represents the detection cell. Represents the current wavelength The background scattering value below.

[0080] The molar absorptivity of hemoglobin at the current wavelength can be obtained from publicly available clinical laboratory databases; the optical path length of the detection cell can be obtained from the equipment design parameters.

[0081] Based on prior knowledge, the scattering background value at the current wavelength indicates that the scattering background of blood changes linearly with wavelength within the detection band. Therefore, based on a linear model fitting, let... ,in, The background intercept represents the scattering parameter to be solved. This represents the scattering background wavelength coefficient, which is also the scattering parameter to be solved.

[0082] The diagonal weight matrix W, composed of the data quality scores, has dimensions N×N, where N is the number of effective wavelengths.

[0083] Construct the following matrix equations for batch solving:

[0084] The observation vector Y (latitude N×1) consists of the average absorbance of each wavelength during the steady-state phase.

[0085] Design Matrix (Latitude N×3): Each row corresponds to a parameter combination for one wavelength, with the following structure: The first column is... (i.e., absorption effect term), the second column is 1 (i.e., scattering intercept term), the third column is... (i.e., the scattering wavelength coefficient).

[0086] Parameter vector (Dimension 3×1) contains 3 parameters to be solved. Here, T represents the transpose of the matrix.

[0087] Diagonal weight matrix W (latitude N×N): It is a diagonal matrix whose diagonal elements are composed of data quality scores for each wavelength.

[0088] The final matrix equation is . For design matrix (latitude) ).

[0089] The aforementioned diagonal weight matrix W is used to introduce data quality scores corresponding to different wavelengths, and the weighted least squares method is used to solve for the parameter vector. The estimated value, i.e. Thus, the hemoglobin concentration value can be directly obtained from the parameter estimates. .

[0090] The conversion module 105 is used to convert the hemoglobin concentration into an HCT value using a preset HCT conversion algorithm.

[0091] The HCT conversion algorithm is existing technology and will not be described in detail here.

[0092] The judgment module 106 is used to determine whether the HCT value is true and valid based on the data quality score.

[0093] In some embodiments of the present invention, based on the actual blood HCT detection process, several wavelengths corresponding to key indicators are usually located in the characteristic absorption peak, isoabsorption point or the region most sensitive to HCT changes of hemoglobin, and these wavelengths are designated as key wavelengths (set according to actual detection requirements).

[0094] Three thresholds were set: a single wavelength pass threshold, an overall quality threshold, and a critical wavelength pass rate threshold. Based on experience, these three thresholds were set to 0.6, 0.4, and 0.7, respectively.

[0095] Among the key wavelengths, the total number of key wavelengths whose quality scores exceed the single wavelength pass threshold is obtained; the pass rate of key wavelengths is then calculated as: number of pass key wavelengths / total number of key wavelengths.

[0096] Calculate the mean of the overall data quality score for all effective wavelengths;

[0097] At this point, if the mean of the comprehensive data quality score of all effective wavelengths is greater than or equal to the overall quality threshold, and the pass rate of the key wavelength is greater than or equal to the pass rate threshold of the key wavelength, then the quality of the data collected this time is deemed to be qualified, and the hemoglobin concentration and the final HCT value calculated based on the current weight matrix are reliable and are output.

[0098] Conversely, if neither of the above conditions is met, or only one condition is met, the data collected is deemed unreliable, and medical staff are advised to conduct the test again.

[0099] In other embodiments of the present invention, it is determined whether the mean of the data quality score reaches a preset threshold: if so, the HCT value is valid; otherwise, the blood sample is retested.

[0100] The average data quality score is obtained by averaging the data quality scores of the absorbance data corresponding to all wavelengths. The average data quality score is used to reflect the overall data quality. The higher the value, the higher the reliability of the HCT value.

[0101] The threshold can be set by the user according to the actual situation, and is preferably 0.5.

[0102] In multispectral time-series data acquisition, not all temporal variations in all wavelength channels accurately reflect the optical properties of blood. Individual sensors may generate abnormal signals due to transient failures, sudden changes in ambient light, or electronic noise, leading to distortion of some wavelength data.

[0103] Theoretical analysis shows that during dynamic measurement, changes in the physical state of the blood sample (such as erythrocyte sedimentation and mixing) trigger a coordinated response across all wavelengths. For example, when blood mixes, erythrocyte scattering and hemoglobin absorption at each wavelength tend to stabilize synchronously. Therefore, the rate of change curves for each wavelength should exhibit a high degree of consistency over time. Figure 2 The flowchart illustrates the process of obtaining the authenticity index according to the second embodiment of the present invention. The process of obtaining the authenticity index includes:

[0104] S201. Construct a sequence of absorbance change rates for each wavelength based on the absorbance change rate at each sampling time point of the absorbance data at each wavelength.

[0105] Based on the smoothed data obtained through preprocessing, the first derivative of the absorbance data at each sampling time point for each wavelength is calculated, forming a sequence of absorbance change rates for each wavelength. This derivative reflects the rate of change of absorbance data over time and can describe its trend and rate of change. Regardless of the baseline absorbance, as long as the blood sample undergoes the same physical process, the trend (increasing, decreasing, or remaining stable) of absorbance change at each wavelength and its dynamic rhythm should remain synchronized.

[0106] Based on the above analysis, the rate of change of absorbance at different wavelengths should be synergistic. Specifically, when the change is caused by the dynamic process of the blood sample itself (such as erythrocyte sedimentation), the trend of the rate of change at each wavelength should be consistent; however, if the change originates from local interference (such as a single-wavelength sensor malfunction), the rate of change at that wavelength will deviate from that of other wavelengths.

[0107] Based on this, the Pearson correlation coefficient can be used to correlate any two different wavelengths. and The correlation between the corresponding absorbance change rate sequences is calculated. The coefficient ranges from -1 to 1. The closer the value is to 1, the more consistent the linear relationship between the absorbance change rates of the two wavelengths over time, and the higher the reliability of the data. Conversely, if the value is close to 0 or negative, it indicates that the data is significantly affected by irrelevant noise and the reliability is low.

[0108] S202. Calculate the average value of the correlation coefficients between the absorbance change rate sequence corresponding to any wavelength and the absorbance change rate sequences corresponding to all other wavelengths as an indicator of the authenticity of the absorbance data corresponding to any wavelength.

[0109] Calculate any wavelength The mean of the correlation with all other wavelengths is denoted as This mean value can be used as an evaluation index of the authenticity of the absorbance data corresponding to any wavelength, i.e., the authenticity index. The larger the value, the higher the authenticity.

[0110] According to the principle of optical measurement, when a bubble passes through the light path, it will scatter or block the light momentarily, causing a brief spike in absorbance data in the time domain. Figure 3 The flowchart below shows the process for obtaining the bubble interference index according to the third embodiment of the present invention. The process for obtaining the bubble interference index includes:

[0111] S301. Obtain the coefficient of variation based on the absorbance data at any wavelength.

[0112] The process of obtaining the coefficient of variation will be described in detail in the fourth embodiment, and will not be repeated here.

[0113] The coefficient of variation can be used The coefficient of variation is used to represent this. The larger the value of the coefficient of variation, the greater the likelihood of the presence of air bubbles.

[0114] S302. Perform frequency domain transformation on the absorbance data after drift trend removal processing to obtain FFT results, square the amplitude of the FFT results and normalize them to obtain power spectral density, and calculate the ratio of high-frequency energy in the high-frequency range to full-frequency energy in the full-band range based on the power spectral density.

[0115] In signal processing, rapid abrupt changes in the time domain (such as pulse signals) manifest as a wide range of high-frequency components in the frequency domain. The sharper and more frequent the pulse, the more high-frequency energy the signal contains.

[0116] The drift trend in the signal containing the absorbance data was removed using the linear least squares method, and then the processed signal was transformed to the frequency domain using a Fast Fourier Transform (FFT). Next, based on the classical periodogram method, the squared amplitude of the FFT result was normalized to obtain the power spectral density. ,in, This represents the actual frequency. The larger this value, the broader the energy distribution in the frequency domain caused by the signal abrupt change induced by the bubble.

[0117] In the high-frequency range, the power spectral density typically enters a "flat region" caused by system electronic noise, i.e., a background noise plateau.

[0118] The high-frequency energy is defined as the integral value of the power spectral density in the high-frequency range, while the full-band energy is the integral value of the power spectral density in the full-band range.

[0119] The ratio can be expressed as: , wherein The high-frequency energy, the This represents the full-frequency energy. It is understood that if the full-frequency energy is 0, the formula calculation is meaningless. Therefore, in the specific analysis process of this embodiment of the invention, the full-frequency energy is limited to be greater than 0.

[0120] The ratio is used to quantify the relative importance of bubble interference in the signal containing the absorbance data; the larger the value, the greater the importance.

[0121] S303. Determine the bubble interference index corresponding to the absorbance data of any wavelength based on the coefficient of variation and the ratio.

[0122] The bubble interference index can be expressed by the formula:

[0123] ;

[0124] Wherein, B represents the bubble interference index. The larger the value, the more severe the bubble interference and the lower the signal confidence. norm represents normalization processing.

[0125] In one embodiment of the present invention, the normalization process can be specifically, for example, maximum and minimum value normalization. Furthermore, the normalization in subsequent steps can all adopt maximum and minimum value normalization. In other embodiments of the present invention, other normalization methods can be selected according to the specific range of the numerical values, which will not be elaborated further.

[0126] It should be noted that, for ease of calculation, all indicator data involved in the calculation in this embodiment of the invention have undergone data preprocessing to eliminate the influence of dimensions. The specific methods for eliminating the influence of dimensions are well known to those skilled in the art and are not limited here.

[0127] Figure 4 The flowchart below shows the process for obtaining the coefficient of variation according to the fourth embodiment of the present invention. The process for obtaining the coefficient of variation includes:

[0128] S401. For a sampling time point under any wavelength, define a local time window with a preset width centered on the sampling time point.

[0129] First, for any sampling time point under any wavelength, a local time window is set around that sampling time point (i.e., centered on that sampling time point). The window size can be set independently according to the actual situation, preferably 7 sampling time points.

[0130] S402. Calculate the local coefficient of variation of the absorbance data within the local time window.

[0131] The method for obtaining the local coefficient of variation is existing technology and will not be described in detail here. The larger the value of the local coefficient of variation, the greater the probability that there is a bubble in the local time window centered on the sampling time point.

[0132] S403. Repeat the process of obtaining the local variation coefficient to obtain the local variation coefficient corresponding to each sampling time point under any wavelength.

[0133] In other words, each sampling time point corresponds to a local coefficient of variation.

[0134] S404. The mean of the local coefficients of variation is taken as the coefficient of variation.

[0135] The average of all the local coefficients of variation is taken as the overall coefficient of variation for any given wavelength.

[0136] Figure 5 The flowchart below shows the process for obtaining a high-frequency range according to the fifth embodiment of the present invention. The process for obtaining the high-frequency range includes:

[0137] S501. The frequency at which the power spectral density first shows an inflection point from high frequency to low frequency is marked as the starting frequency.

[0138] Starting from the highest frequency and scanning down to lower frequencies, locate the first significant and sustained inflection point in the power spectral density. Specifically, calculate the derivative of the power spectral density; when the derivative changes from 0 to a continuously positive value, it is considered to have left the noise-dominated flat region. The frequency of this inflection point, i.e., the starting frequency, is denoted as . , used to identify the dominant frequency band of the system's inherent noise.

[0139] S502. A preset multiple of the starting frequency is taken as the high frequency, and the frequency range between the high frequency and the maximum frequency of the power spectral density is taken as the high frequency interval.

[0140] Bubble interference is a transient mutation, and its frequency components are significantly higher than normal physiological changes and system noise; therefore, frequencies that can capture these abnormal mutations should be searched on the system background noise platform.

[0141] high frequency Set to a preset multiple of the starting frequency, i.e. , where k is a safety factor that ensures the frequency range of the analysis is clearly above the noise plateau, and its value range is [2, 3].

[0142] The maximum frequency of the power spectral density is The high-frequency range is [ , ].

[0143] In blood hematologic tract coagulation (HCT) testing, in addition to air bubble interference, the onset of blood coagulation can also introduce significant errors. The slow changes in blood components during coagulation cause a unidirectional, slow drift in absorbance. This drift contradicts normal dynamic sedimentation patterns, exhibiting a continuous trend in the time domain and non-stationary characteristics statistically. Ultimately, it introduces systematic bias, reducing the accuracy of HCT measurement results. Figure 6 The flowchart below illustrates the process of obtaining coagulation interference indicators according to the sixth embodiment of the present invention. The process of obtaining coagulation interference indicators includes:

[0144] S601. Divide the absorbance data corresponding to any wavelength into a preset number of absorbance data segments.

[0145] Divide the time-series absorbance data corresponding to any wavelength into M segments (M can be set independently according to the actual situation, and its value can be set to 5 based on experience).

[0146] S602. Perform linear fitting on each of the absorbance data segments to obtain the fitting slope corresponding to each absorbance data segment and form a slope sequence.

[0147] A linear fit is performed on each of the absorbance data segments to obtain a slope sequence consisting of the fitting slope corresponding to each absorbance data segment.

[0148] S603. Calculate the mean of the absolute values ​​of the differences between adjacent fitted slopes in the slope sequence as the first nonstationarity index of coagulation interference.

[0149] First, the absolute value of the difference between adjacent fitted slopes in the slope sequence is calculated. The average of all these absolute values ​​is then used to obtain a first non-stationarity index of coagulation interference. This first non-stationarity index can be used... To express.

[0150] To accurately identify signal drift caused by coagulation, the changing patterns of adjacent slopes can be analyzed. Under stable or interference-free conditions, adjacent slopes are similar, with their difference fluctuating randomly around zero. However, when persistent drift caused by coagulation exists, such as when absorbance shows an accelerating upward trend, the slope sequence will exhibit cumulative changes in the same direction, resulting in a consistently large positive absolute value for the difference between consecutive slopes. Even if the drift rate changes or the direction reverses, the absolute value of the consecutive differences will still be significantly larger. The larger this absolute value, the more unstable the rate of absorbance change, the more drastic the fluctuations, and the higher the likelihood of significant coagulation interference.

[0151] S604. Obtain the number of runs and the expected number of runs of the absorbance data according to the run test algorithm, and then determine the second nonstationarity index of the coagulation interference.

[0152] Coagulation can cause red blood cell aggregation or precipitation, leading to a slow and continuous drift (such as a trend or step change) in absorbance data. This drift disrupts the stationarity of the data sequence, causing its statistical properties (such as mean and variance) to change over time. Therefore, a runs test algorithm is introduced, which is an existing technology. Specifically, after removing the trend through linear fitting, the absorbance residual sequence is obtained. The residual sequence is then converted into a sign sequence, containing either positive or negative values, and the number of runs is calculated, where a run refers to a sequence of consecutive identical signs. The number of points in the residual sequence with residuals greater than the median and the number of points with residuals less than the median are obtained, i.e., the number of data points with positive and negative signs, respectively.

[0153] The expected number of runs can be obtained using the following formula:

[0154] ;

[0155] Wherein, DZ represents the number of points where the residual is greater than the median, and the... This represents the number of points where the residual is less than the median. The term represents the total length of the residual sequence. This represents the expected number of runs. The total length of the residual sequence is greater than 0.

[0156] The process of obtaining the second nonstationarity index includes:

[0157] The second nonstationarity index is determined based on the standardized distance between the number of runs and the expected number of runs, wherein the second nonstationarity index is used to characterize the degree to which the number of runs deviates from the expected number of runs.

[0158] The second nonstationarity index can be expressed by the formula:

[0159] ;

[0160] Among them, the Represents the number of runs, the Indicates the expected number of runs, the This represents the standard deviation of the number of runs. Adding 0.1 to the denominator prevents calculation errors caused by a standard deviation of 0. The addition of 0.1 has negligible impact on subsequent calculations and overall error. This represents the second nonstationarity index.

[0161] The formula for calculating the standard deviation of the number of runs is:

[0162] ;

[0163] In the formula, This represents the number of points where the residual is greater than the median (i.e., the number of positive signs). The number of points where the residual is less than the median (i.e., the number of negative signs); N is the total length of the sequence, N = n1 + n2; Var(R) represents the variance of the number of runs R. Depending on the specific scenario, N is a natural number greater than 1, i.e. and None of them are 0.

[0164] S605. Determine the coagulation interference index based on the first non-stationarity index and the second non-stationarity index.

[0165] The coagulation interference index can be expressed by the formula: Wherein, C represents the coagulation interference index, and norm represents the maximum and minimum value normalization process. Normalization eliminates the dimensions of the first non-stationary index and the second non-stationary index, which facilitates subsequent calculations.

[0166] Figure 7 The flowchart below shows the data quality score acquisition process provided in the seventh embodiment of the present invention. The data quality score acquisition process includes:

[0167] S701. Determine the interference weight value based on the bubble interference index and the coagulation interference index, wherein the interference weight value is a normalized value, and the bubble interference index and the coagulation interference index are negatively correlated with the interference weight value.

[0168] In this embodiment of the invention, negative correlation indicates an inverse relationship between the independent and dependent variables, where a smaller independent variable results in a larger dependent variable. Therefore, the specific formula for calculating the interference weight value is as follows: .

[0169] S702. Calculate the normalized value of the authenticity index, and use the average of the normalized value and the interference weight value as the data quality score.

[0170] By integrating data from three dimensions—authenticity, bubble interference, and coagulation interference—a data quality score is obtained.

[0171] The present invention has the following beneficial effects:

[0172] Absorbance data of blood samples were collected at various wavelengths. This data forms the basis for subsequent analysis.

[0173] Based on the absorbance data, the authenticity index, bubble interference index, and coagulation interference index of the absorbance data corresponding to each wavelength are obtained. These indices are used to quantify the reliability of the absorbance data at each wavelength.

[0174] Based on the authenticity index, the bubble interference index, and the coagulation interference index, a data quality score is determined for the absorbance data corresponding to each wavelength. The higher the data quality score, the more reliable the corresponding absorbance data.

[0175] Based on a preset absorbance model that includes hemoglobin concentration and scattering background value, a scattering background value solution model, and a diagonal weight matrix composed of the data quality scores, a matrix equation is constructed and solved to obtain the hemoglobin concentration. A data quality weighting mechanism is introduced in the hemoglobin concentration inversion process, using high-quality data as the core weight to effectively suppress interference from low-quality data, making the hemoglobin concentration value more accurate.

[0176] The hemoglobin concentration is converted into an HCT value using a preset HCT conversion algorithm. Obtaining the HCT value based on the hemoglobin concentration is a prior art technique.

[0177] The validity of the HCT value is determined based on the data quality score. The validity of the HCT value is verified to ensure the reliability of the output results.

[0178] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0179] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A rapid blood hematocrit (HCT) detection system based on multispectral imaging, characterized in that, The system includes: The acquisition module is used to acquire absorbance data of blood samples at various wavelengths; The interference analysis module is used to obtain the authenticity index, bubble interference index and coagulation interference index of the absorbance data corresponding to each wavelength based on the absorbance data. The scoring module is used to determine the data quality score of the absorbance data corresponding to each wavelength based on the authenticity index, the bubble interference index and the coagulation interference index. The concentration analysis module is used to construct and solve the matrix equation using the weighted least squares method based on a preset absorbance model containing hemoglobin concentration and scattering background value, a scattering background value solution model, and a diagonal weight matrix composed of the data quality score, in order to obtain the hemoglobin concentration. The conversion module is used to convert the hemoglobin concentration into an HCT value using a preset HCT conversion algorithm; The judgment module is used to determine whether the HCT value is true and valid based on the data quality score; The process of obtaining the authenticity index includes: Based on the absorbance change rate at each sampling time point of the absorbance data at each wavelength, construct a sequence of absorbance change rate corresponding to each wavelength; The average value of the correlation coefficients between the absorbance change rate sequence corresponding to any wavelength and the absorbance change rate sequences corresponding to all other wavelengths is used as an indicator of the authenticity of the absorbance data corresponding to any wavelength. The process of obtaining the bubble interference index includes: The coefficient of variation is obtained based on the absorbance data at any wavelength; The absorbance data after drift-reduction processing is transformed in the frequency domain to obtain the FFT result. The amplitude of the FFT result is squared and then normalized to obtain the power spectral density. Based on the power spectral density, the ratio of high-frequency energy in the high-frequency range to full-frequency energy in the full-band range is calculated. The bubble interference index corresponding to the absorbance data at any wavelength is determined based on the coefficient of variation and the ratio. The process of obtaining the coagulation interference index includes: The absorbance data corresponding to any wavelength is divided into a predetermined number of absorbance data segments; A linear fit is performed on each absorbance data segment to obtain the fitting slope corresponding to each absorbance data segment, thereby forming a slope sequence; The mean of the absolute values ​​of the differences between adjacent fitted slopes in the slope sequence is calculated as the first nonstationarity index of coagulation interference. The run number and expected run number of the absorbance data are obtained according to the run test algorithm, and then the second nonstationarity index of the coagulation interference is determined. The coagulation interference index is determined based on the first non-stationarity index and the second non-stationarity index.

2. The rapid blood HCT detection system based on multispectral imaging as described in claim 1, characterized in that, After collecting absorbance data of blood samples at various wavelengths, the method further includes preprocessing the absorbance data. The preprocessing process includes: The absorbance data was noise-reduced using a moving average smoothing method. When missing values ​​are identified in the absorbance data, linear interpolation is used to fill in the missing values.

3. The rapid blood HCT detection system based on multispectral imaging as described in claim 1, characterized in that, The absorbance data of the blood samples collected at various wavelengths include: Obtain fresh blood samples that have been anticoagulated and mixed. The absorbance data of the blood sample are collected over time within the characteristic absorption band of hemoglobin, wherein the temporal resolution of the acquisition is matched with the dynamic sedimentation rate of the blood, and the total acquisition time covers the complete process of the blood sample from a uniformly mixed state to significant sedimentation.

4. The rapid blood HCT detection system based on multispectral imaging as described in claim 1, characterized in that, The process of obtaining the coefficient of variation includes: For any sampling time point under any wavelength, a local time window of a preset width is defined with the sampling time point as the center; Calculate the local coefficient of variation of the absorbance data within the local time window; The process of obtaining the local variation coefficient is repeated to obtain the local variation coefficient corresponding to each sampling time point under any wavelength; The mean of the local coefficients of variation is used as the coefficient of variation.

5. The rapid blood HCT detection system based on multispectral imaging as described in claim 1, characterized in that, The process of obtaining the high-frequency range includes: The frequency at which the power spectral density first shows an inflection point from high frequency to low frequency is marked as the starting frequency. The preset multiple of the starting frequency is taken as the high frequency, and the frequency range between the high frequency and the maximum frequency of the power spectral density is taken as the high frequency interval.

6. The rapid blood HCT detection system based on multispectral imaging as described in claim 1, characterized in that, The process of obtaining the second nonstationarity index includes: The second nonstationarity index is determined based on the standardized distance between the number of runs and the expected number of runs, wherein the second nonstationarity index is used to characterize the degree to which the number of runs deviates from the expected number of runs.

7. The rapid blood HCT detection system based on multispectral imaging as described in claim 1, characterized in that, The process of obtaining the data quality score includes: Based on the bubble interference index and the coagulation interference index, the interference weight value is determined, wherein the interference weight value is a normalized value, and the bubble interference index and the coagulation interference index are negatively correlated with the interference weight value. Calculate the normalized value of the authenticity index, and then normalize the product of the normalized value and the interference weight value to obtain the data quality score.

Images