Blood detection method

Abstract

The embodiment of the invention provides a blood detection method. The blood detection method comprises the following steps: performing multi-labeled fluorescent staining on a to-be-detected blood sample; collecting a multi-labeled fluorescence signal through a fluorescence microscope; processing the fluorescence signal based on a signal separation algorithm to reduce background noise; and analyzing the processed fluorescence signal to obtain a detection result. According to the scheme of the embodiment of the invention, the problem of background noise caused by signal cross interference in multi-labeled fluorescence detection can be solved.

Description

Technical Field

[0001] The present application relates to biomedical engineering, and in particular to a blood testing method. Background Art

[0002] A blood test method diagnoses diseases or assesses health status by analyzing specific biomarkers in the blood. This method usually has high sensitivity and accuracy. However, when using multiplexed fluorescence detection technology, the cross-interference of signals between different fluorescent labels often causes background noise problems, which not only reduces the reliability of the test results, but also may affect the final diagnostic accuracy. Summary of the invention

[0003] In view of this, an embodiment of the present disclosure provides a blood testing method, which at least partially solves the problems existing in the prior art.

[0004] A blood testing method comprising:

[0005] Performing multiple labeling fluorescent staining on the blood sample to be tested;

[0006] The fluorescence signals of multiple labels were collected by fluorescence microscopy;

[0007] Processing the fluorescent signal based on a signal separation algorithm to reduce background noise;

[0008] Analyze the processed fluorescence signal to obtain the detection result.

[0009] According to one embodiment, processing the fluorescence signal based on a signal separation algorithm to reduce background noise comprises:

[0010] De-noising the fluorescence signal based on the preprocessing step to reduce environmental interference;

[0011] The main spectrum components of each fluorescent marker are extracted using spectrum analysis method;

[0012] The principal component analysis method was applied to further separate the cross-interference signals;

[0013] The low-intensity background signal is removed by the filter, and the signal with high signal-to-noise ratio is retained.

[0014] According to one embodiment, the denoising of the fluorescence signal based on the preprocessing step includes:

[0015] Smoothing the raw fluorescence signal;

[0016] Detect and remove spike noise based on sliding window method;

[0017] Use wavelet transform technology to perform multi-scale denoising on the signal;

[0018] If the noise intensity exceeds the threshold A, deep denoising is performed;

[0019] Wherein, A represents the set noise intensity threshold, which is used to determine whether deep denoising is required.

[0020] According to one embodiment, the detecting and removing of spike noise based on a sliding window method comprises:

[0021] Set the initial sliding window size B;

[0022] Calculate the mean C and standard deviation D of the signal within the sliding window;

[0023] Compare the current point with the average value C in the window. If the deviation exceeds K×D, it is marked as spike noise.

[0024] Move the sliding window to continue processing subsequent data;

[0025] Among them, B represents the size of the sliding window, C represents the average value of the signal in the sliding window, D represents the standard deviation of the signal in the sliding window, and K represents the deviation coefficient.

[0026] According to one embodiment, comparing the current point with the average value C in the window and marking it as spike noise if the deviation exceeds K×D further comprises:

[0027] If the signal strength E at the current point exceeds the window average C plus K×D or is lower than the window average C minus K×D, it is marked as spike noise;

[0028] Use interpolation methods to repair points marked as spike noise;

[0029] Update the mean value C and standard deviation D within the sliding window;

[0030] Repeat the steps until all data points have been processed;

[0031] Among them, E represents the signal strength of the current point, C represents the signal average value in the sliding window, D represents the signal standard deviation in the sliding window, and K represents the deviation coefficient.

[0032] According to one embodiment, updating the mean value C and the standard deviation D within the sliding window includes:

[0033] Whenever the sliding window moves forward, the data point at the beginning of the window is removed and a new data point F is added;

[0034] Recalculate the mean C and standard deviation D within the sliding window;

[0035] Check whether the new data point F is spike noise, if so, continue repairing;

[0036] Record the processing status of all data points within the window;

[0037] Among them, F represents the data point of the newly added sliding window, C represents the signal average value in the sliding window, and D represents the signal standard deviation in the sliding window.

[0038] According to one embodiment, extracting the main spectrum components of each fluorescent marker using a spectrum analysis method includes:

[0039] Perform Fourier transform on the fluorescence signal to obtain the frequency domain diagram;

[0040] According to the fluorescence frequency characteristic G of the known marker, the peak position H of the spectrum is located;

[0041] The height I of each peak is calculated to quantify the signal intensity of the marker;

[0042] If the peak height I exceeds the set threshold J, the spectrum component is confirmed to be the main component;

[0043] Wherein, G represents the fluorescence frequency characteristic of a known marker, H represents the position of the spectrum peak, I represents the height of the spectrum peak, and J represents the threshold of the signal intensity.

[0044] According to one embodiment, locating the spectrum peak position H according to the fluorescence frequency characteristic G of the known marker comprises:

[0045] Set the search range K and search around the preset marked frequency point;

[0046] Use the peak detection algorithm to find the spectrum peak H;

[0047] Verify whether the peak position H is within the theoretical spectrum range L of the marker;

[0048] If position H is within L, it is confirmed as the spectrum peak of the marker;

[0049] Wherein, K represents the search range, H represents the position of the spectrum peak, and L represents the theoretical spectrum range of the marker.

[0050] According to one embodiment, the verifying whether the peak position H is within the theoretical spectrum range L of the marker further comprises:

[0051] Calculate the deviation N between the peak position H and the theoretical spectrum position M of the marker;

[0052] If the deviation N is less than the set threshold O, it is confirmed as a valid spectrum peak;

[0053] Record the information of spectrum peak H;

[0054] Continue searching for the next possible peak;

[0055] Wherein, H represents the position of the spectrum peak, M represents the theoretical spectrum position of the marker, N represents the deviation value, and O represents the threshold value.

[0056] According to a blood testing method as claimed in claim 8, the information based on recording spectrum peak H further comprises:

[0057] Record the intensity P of the spectrum peak H into the data table;

[0058] Check whether there are multiple spectrum peaks Q that meet the conditions;

[0059] If there are multiple spectrum peaks Q that meet the criteria, further verification is performed;

[0060] Determine the final effective spectrum peak H and analyze it;

[0061] Wherein, H represents the position of the spectrum peak, P represents the intensity of the spectrum peak, and Q represents multiple spectrum peaks that meet the standard.

[0062] The embodiment of the present disclosure provides a blood detection method, including: performing multi-labeled fluorescent staining on a blood sample to be detected; collecting multi-labeled fluorescent signals through a fluorescent microscope; processing the fluorescent signals based on a signal separation algorithm to reduce background noise; analyzing the processed fluorescent signals to obtain a detection result. The solution of the embodiment of the present disclosure can solve the background noise problem caused by signal cross-interference in multi-labeled fluorescent detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] In the accompanying drawings, unless otherwise specified, the same reference numerals throughout the multiple drawings represent the same or similar parts or elements. These drawings are not necessarily drawn to scale. It should be understood that these drawings only depict some embodiments disclosed in the present application and should not be regarded as limiting the scope of the present application.

[0064] Figure 1 is a flow chart of a blood testing method of the present application;

[0065] Figure 2 is a flow chart of processing fluorescence signals based on signal separation algorithms to reduce background noise;

[0066] Figure 3 It is a flow chart of denoising the fluorescence signal based on the preprocessing steps;

[0067] Figure 4 It is a flow chart for detecting and removing spike noise based on the sliding window method;

[0068] Figure 5 is a flow chart of marking spike noise according to one embodiment of the present application;

[0069] Figure 6 It is a flowchart for updating the mean value C and standard deviation D in the sliding window;

[0070] Figure 7 It is a flow chart for extracting the main spectrum components of each fluorescent marker using spectrum analysis method;

[0071] Figure 8 It is a flow chart for locating the spectrum peak position H based on the fluorescence frequency characteristic G of the known marker;

[0072] Fig. 9 is a flow chart of searching for peak positions according to one embodiment of the present application;

[0073] Fig.10 It is a flow chart for recording the information of the spectrum peak H. DETAILED DESCRIPTION

[0074] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the statement "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0075] Next, refer to Figure 1 , describing a blood testing method of the present invention.

[0076] S101: Perform multi-label fluorescent staining on the blood sample to be tested. In this step, specific fluorescent markers need to be selected, which can specifically bind to different target molecules. Specifically, for example, when detecting specific leukocyte subsets, antibodies such as CD3, CD4 and CD8 can be selected, and these antibodies can be labeled with fluorescent dyes of different colors, such as AlexaFluor488, AlexaFluor555 and AlexaFluor647. This multi-label fluorescent staining enables researchers to observe multiple different target molecules in the same field of view at the same time, greatly improving the efficiency and accuracy of the detection.

[0077] S102: Collecting multi-labeled fluorescence signals by fluorescence microscopy. In this step, a high-resolution fluorescence microscope or flow cytometer is used to capture the fluorescence signals emitted by the stained samples. These signals contain a large amount of information, such as the expression level, location, and distribution of the target molecules. For example, in one embodiment, an inverted fluorescence microscope is used to photograph the stained blood sample to generate a series of high-quality fluorescence images. After preliminary processing, these images can show the location and expression intensity of target molecules with different fluorescence labels in cells.

[0078] S103: Process the fluorescent signal based on a signal separation algorithm to reduce background noise. Signal cross-interference is a common problem in multi-label fluorescence detection. When multiple fluorescent markers coexist in the same sample, the fluorescent signals of different markers may overlap with each other, resulting in an increase in background noise and affecting the final detection result. To this end, the present application uses a signal separation algorithm to process the original fluorescent signal. Specifically, the algorithm analyzes fluorescent signals of different wavelengths through a mathematical model and performs deconvolution processing on them, thereby effectively removing cross-interference and background noise. For example, using blind source separation technology (such as independent component analysis) or a multimodal fusion method based on machine learning, it is possible to effectively separate and enhance different fluorescent signals, making the signal of each fluorescent marker clearer and improving the accuracy and sensitivity of detection.

[0079] S104: Analyze the processed fluorescent signal to obtain the test result. After the above steps, the fluorescent signal has significantly reduced the background noise and improved the signal-to-noise ratio. At this point, computer software tools can be used to further analyze the image, such as quantitatively measuring the expression level of each target molecule, counting the proportion of positive cells, drawing flow cytometry graphs, etc. For example, in a specific application, the specific expression levels of marker molecules such as CD3, CD4 and CD8 in a certain type of cells can be calculated by professional software, and compared with the normal range values ​​to evaluate the patient's immune status or disease progression. The whole process not only improves the accuracy of the test results, but also greatly simplifies the experimental operation process, providing a strong scientific basis and technical support for clinical diagnosis.

[0080] Next, refer to Figure 2 , specifically describing the steps of processing fluorescence signals based on signal separation algorithms of the present invention to reduce background noise.

[0081] S201: De-noising the fluorescence signal based on the pre-processing step to reduce environmental interference. This step usually includes smoothing the raw fluorescence signal data to remove random disturbances that may be introduced by external light sources or sensor noise. For example, in one embodiment, median filtering or Gaussian filtering techniques can be used to reduce fluctuations in a short period of time, thereby obtaining a more stable fluorescence signal.

[0082] S202: Extract the main spectral components of each fluorescent marker using a spectrum analysis method. Spectral analysis converts the time domain signal into a frequency domain signal through Fourier transform, thereby separating the different frequency characteristics of each fluorescent marker. The key to this step is to identify the frequency peaks that are unique to the fluorescent marker and exclude irrelevant frequency bands. Specifically, the fast Fourier transform (FFT) of the fluorescent signal can be calculated, and the low-amplitude spectral components can be filtered by setting a threshold. Fourier transform is a commonly used signal processing method and will not be described in detail here. In this way, the spectral components of the background noise and the target fluorescent signal can be effectively distinguished.

[0083] S203: Apply principal component analysis (PCA) to further separate cross-interference signals. PCA is a linear transformation technique used to reduce the dimensionality of high-dimensional data and extract the main components, thereby reducing the mutual interference between signals. In this process, the frequency components of the fluorescence signal are taken as input, and the principal component directions are obtained by calculating the covariance matrix and performing eigenvalue decomposition. These principal components represent the most important features of the data. For example, in a specific embodiment, C=X T X calculates the covariance matrix C, where X is the matrix of spectrum data. Next, C is subjected to eigenvalue decomposition, and the vector with the largest eigenvalue is selected as the principal component direction to reconstruct the fluorescence signal.

[0084] S204: Filter out low-intensity background signals through a filter and retain signals with a high signal-to-noise ratio. The goal of this step is to remove low-intensity signals in the frequency domain or time domain and retain only those parts with a higher signal-to-noise ratio. Specifically, an adaptive filter, such as a wavelet transform or a multi-bandpass filter, can be designed to dynamically adjust the filter parameters to ensure the best denoising effect. The design of the filter depends on the specific circumstances of the experimental data, but in general, it can be achieved by setting appropriate filter frequency ranges f_low and f_high. For example, when processing a certain fluorescent marker in blood, the frequency range f_low = 50 Hz and f_high = 300 Hz can be selected to retain most of the effective fluorescent signals and filter out low-frequency noise.

[0085] In summary, the present invention processes the fluorescent signal step by step through multiple steps, ultimately achieving effective suppression of background noise and improving the accuracy of the detection result.

[0086] Next, refer to Figure 3 , describing the process of denoising the fluorescence signal based on the preprocessing step of the present invention.

[0087] S301: Smoothing the original fluorescence signal. This step uses a mathematical filter to reduce high-frequency noise in the signal. Common smoothing methods include moving average, Gaussian filtering, etc. For example, in one embodiment, a moving average filter with a length of N (e.g., N=5 or 7) is used to process the original fluorescence signal to reduce the impact of local fluctuations. Here, N represents the width of the sliding window, which is usually selected between 3 and 9, and the optimal value is 5, which can effectively balance the smoothing effect and the retention of signal details.

[0088] S302: Detect and remove spike noise based on the sliding window method. In this step, a sliding window is set to compare the difference between the data point in the window and its adjacent data points, identify the spike noise and remove it. Specifically, if the difference between a data point in the window and the K data points before and after it is greater than the preset threshold B, the point is considered to be a spike noise. For example, K = 2 or 3, the optimal value is 2, because a K that is too large may misjudge non-noise points, and a K that is too small may not effectively detect noise points. Threshold B can be adjusted according to actual conditions, and is generally set to twice the standard deviation of the signal mean.

[0089] S303: Use wavelet transform technology to perform multi-scale denoising on the signal. This step uses wavelet transform to decompose the signal at different scales, separate the noise and useful signal components, and then remove the noise components through threshold processing to reconstruct the signal. The choice of wavelet basis has a great influence on the denoising effect. Commonly used wavelet bases include Db4, Sym4, etc. For example, the Sym4 wavelet basis is used to perform three-level decomposition of the fluorescence signal. The decomposition threshold C of each level is the standard deviation of the signal multiplied by the coefficient α, α is between 0.7 and 0.9, and the optimal value is 0.8. This setting can retain the subtle features of the signal as much as possible while ensuring the denoising effect.

[0090] S304: If the noise intensity exceeds threshold A, perform deep denoising. This step is to re-evaluate the signal after denoising in the previous steps. If the noise intensity is still higher than threshold A, a more complex or deeper denoising algorithm is further applied. Threshold A is usually set to 1.5 to 2.0 times the standard deviation of the original signal mean, and the optimal value is 1.8. This is because a lower threshold may lead to excessive denoising, while a higher threshold may not completely remove the noise. In one embodiment, if the signal after processing in the aforementioned steps still contains significant noise (for example, the peak-to-peak value exceeds threshold A), an adaptive filtering or iterative denoising algorithm is further applied to ensure the accuracy and reliability of the signal.

[0091] Next, refer to Figure 4 , describing the steps of detecting and removing spike noise based on the sliding window method of the present invention.

[0092] S401: Set the initial sliding window size B. The sliding window size B determines the number of data points referenced when detecting noise, and is generally between 3 and 10. The specific value depends on the actual application scenario. In one embodiment, the sliding window size B can be set to 5, which performs well in balancing computational complexity and detection accuracy.

[0093] S401: Calculate the mean value C and standard deviation D of the signal in each sliding window. The mean value C of the signal is used to reflect the general level of the signal in the current window, while the standard deviation D reflects the degree of fluctuation of the signal. These two values ​​are calculated by the following formulas: C = (x1 + x2 + ... + xB) / B and D = sqrt (Σ (xC) ^ 2 / B), where x1, x2, ..., xB represent the data points in the sliding window. These calculation steps ensure a comprehensive understanding of the characteristics of the signal in the current window.

[0094] S403: Compare the current point with the average value C in the window. If the deviation exceeds K×D, it is marked as spike noise. K is a pre-set deviation coefficient, which usually takes a value between 2 and 3. The specific value can be optimized based on experimental results. The meaning of this formula is that when judging whether a data point is a spike noise, not only the difference between the point and the average value in the window is considered, but also the standard deviation of the signal is combined to more accurately identify abnormal points. For example, when K is set to 2, it means that if the deviation between the current data point and the average value in its window exceeds 2 times the standard deviation, the point is considered to be a spike noise.

[0095] S404: Moving the sliding window to continue processing subsequent data is a cyclic step of the method. The sliding step size of each window is usually 1, which ensures that all data points are evaluated. In a specific blood test application, suppose that some mutation values ​​appear during the test, which may be caused by instrument errors or other external factors. Through the above steps, the system can effectively identify these mutation values ​​and mark them as noise.

[0096] Through the above steps, the present invention provides an effective means for detecting and removing spike noise based on a sliding window method, which is particularly suitable for scenarios such as blood testing that require high-precision data analysis.

[0097] Next, refer to Figure 5 , describing the steps of comparing the current point with the average value C in the window, and marking it as spike noise if the deviation exceeds K×D:

[0098] First, obtain the signal data in the sliding window. Specifically, the sliding window contains a certain number of signal strength values. Assuming the window size is N, C represents the signal average value in the sliding window, which is calculated by dividing the sum of all signal strength values ​​in the window by N; D represents the signal standard deviation in the sliding window, which is used to quantify the degree of signal fluctuation, and is calculated by taking the square root of the average of the squares of the deviations between each signal strength value in the window and C. K is a predetermined constant, which is usually between 1 and 3, and the optimal value depends on the specific application scenario. On this basis, the method of the present application also includes:

[0099] S501: If the signal strength E of the current point exceeds the window average C plus K×D or is lower than the window average C minus K×D, it is marked as spike noise. Specifically, the signal strength E of the current point is compared with the average value C in the window. If E exceeds C plus K×D or is lower than C minus K×D, it is considered that there is spike noise at this point and it is marked. The core of this step is to identify outliers through statistical methods. K×D sets a reasonable threshold to determine whether E exceeds the normal range. Specifically, K×D represents K times the standard deviation, which indicates the maximum allowable deviation. The value between 1 and 3 is selected to avoid misjudging normal data while identifying noise.

[0100] S502: Use interpolation methods to repair points marked as spike noise. Interpolation methods usually select adjacent normal values ​​for linear or polynomial fitting to fill or correct spike noise points. For example, in blood test data, a point has an abnormally high or low value due to external interference. Interpolation can restore the reasonable value of the point, thereby maintaining data continuity and accuracy.

[0101] S503: Update the mean value C and standard deviation D in the sliding window. After processing each point, C and D in the window need to be recalculated to reflect the latest data status. Specifically, when a new data point is added to the window and the oldest data point is removed from the window, C and D should be updated immediately to ensure that subsequent comparisons are accurate and effective.

[0102] S504: Repeat the above steps until all data points are processed. For example, in a complete blood test process, from the beginning to the end of the data sequence, each data point is checked one by one to see if there is spike noise, and corresponding processing and updating are performed until all data points are analyzed. This ensures the consistency and reliability of the entire data set.

[0103] Next, refer to Figure 6 , describing the steps of updating the mean value C and standard deviation D within the sliding window of the present invention.

[0104] S601: Whenever the sliding window moves forward, the sliding window removes the frontmost data point and adds a new data point F. This operation ensures that the data in the sliding window is always the latest continuous data segment, so as to timely reflect the changes in the detection signal. For example, in the blood test process, each time a new blood data point F is collected, the old data point will be removed from the window.

[0105] S602: The average value C and standard deviation D in the sliding window need to be recalculated. The average value C refers to the arithmetic mean of all data points in the sliding window, while the standard deviation D describes the degree of dispersion of the data points. The purpose of recalculating these two values ​​is to accurately reflect the state of the data in the current window. For example, in one embodiment, if the first N-1 data points in the sliding window are X1, X2, ..., X(N-1), after adding the new data point F, the data points of the new window are X2, X3, ..., XN = F. At this time, the average value C = (X2 + X3 + ... + XN) / N, and the standard deviation D = sqrt{[Σ(Xi-C) 2 ] / N}, where Xi represents the i-th data point in the window, and i ranges from 2 to N. This formula is chosen because it can effectively reflect the overall level of the data and its fluctuations.

[0106] S603: Check whether the newly added data point F is spike noise. Spike noise usually refers to abnormal data points that suddenly increase or decrease in a short period of time, which may affect the accuracy of the mean value C and the standard deviation D. If F is identified as spike noise, it is necessary to continue to repair, that is, to exclude abnormal values ​​to avoid interference with subsequent analysis. For example, in one embodiment, a threshold value, such as 3 standard deviations, can be set, that is, when the value of F exceeds the mean value C plus or minus 3 times the standard deviation D, F is determined to be spike noise and corrected or eliminated.

[0107] S604: Record the processing status of all data points in the window. This includes marking which data points are normal, which are identified as spike noise, and how these data points are processed. These records are very useful for subsequent data analysis and review of abnormal situations. For example, in actual applications, the status of each data point can be saved in a log file for subsequent review and reference.

[0108] Through the above steps, the blood test data can be effectively monitored and analyzed in real time, and the reliability and accuracy of the test results can be improved.

[0109] Next, refer to Figure 7 , describing the steps of extracting the main spectrum components of each fluorescent marker using the spectrum analysis method of the present invention:

[0110] S701: Perform Fourier transform on the fluorescence signal to obtain a frequency domain image;

[0111] S702: locating the spectrum peak position H according to the fluorescence frequency characteristic G of the known marker;

[0112] S703: Calculate the height I of each peak to quantify the signal intensity of the marker;

[0113] S704: If the peak height I exceeds the set threshold J, it is confirmed that the spectrum component is the main component.

[0114] In actual operation, the fluorescence signal needs to be Fourier transformed first to convert the time domain signal into a frequency domain signal. The frequency domain diagram can reveal the frequency composition of the signal, which is convenient for subsequent processing. This step can filter out background noise and nonspecific signals and improve the signal-to-noise ratio. For example, in the blood test process, different types of fluorescent markers may show complex mixed signals in the time domain diagram, and the Fourier transform can decompose these mixed signals into specific frequency domain components, thereby showing the information of each marker more clearly.

[0115] Next, the location of the spectrum peak is determined based on the fluorescence frequency characteristics of the known markers. The fluorescence frequency characteristics of the markers are usually obtained through laboratory testing and recorded as a set of standard frequency values. By comparing the peak position in the frequency domain graph with the known frequency characteristics, the characteristic frequencies of different markers can be accurately identified. For example, in a blood sample, fluorescent markers associated with a specific disease may have specific frequency characteristics. By analyzing the spectrum graph, these frequency characteristics can be accurately located, thereby preliminarily identifying potential markers.

[0116] Afterwards, the height of each peak is calculated to quantify the signal strength of the marker. The height of the peak reflects the strength of the signal at that frequency, so it can be used as an important basis for judging the presence and concentration of the marker. The larger the height value, the higher the concentration of the marker. Specifically, the peak height can be determined by taking the maximum value of the spectrum peak. For example, assuming that the height of one peak in the frequency domain graph is 200 and the height of the other peak is 300, by comparing these two values, it can be preliminarily determined that the concentration of marker B is higher than that of marker A.

[0117] Finally, determine whether it exceeds the set threshold. If the height of the peak exceeds the preset threshold, the spectral component is considered to be the main component, otherwise it is not. The setting of the threshold is usually based on a large amount of experimental data and statistical analysis to ensure high accuracy. For example, the threshold can be set to 150. If the height of a peak in the frequency domain graph is 200, which exceeds 150, the marker is confirmed to be the main component; if the height of another peak is 100, it is not identified as the main component. In this way, noise and weak signals can be effectively filtered out, and the reliability and accuracy of the detection results can be improved.

[0118] The above steps can effectively identify and quantify different types of fluorescent markers in complex blood samples through spectrum analysis and signal processing, providing important reference for disease diagnosis and research. In this process, G represents the fluorescence frequency characteristics of known markers, which are usually measured by standard solutions; H represents the position of the spectrum peak, that is, the coordinates of a specific frequency point on the frequency domain graph; I represents the height of the spectrum peak, reflecting the strength of the signal; J represents the threshold of signal intensity, the optimal value determined based on a large amount of experimental data. The effective use of these parameters makes spectrum analysis methods play an important role in blood testing.

[0119] Next, refer to Figure 8 , describes the present invention, based on the fluorescence frequency characteristics G of the known marker, the location of the spectrum peak position H. This process includes four main steps.

[0120] S801: Set the search range K to search around the preset marker frequency point. This means that on the fluorescence spectrum, a specific frequency interval K is selected, which contains the expected frequency points of the known marker. For example, if the expected frequency of the known marker is around 650 nanometers, the search range K can be set to 640 nanometers to 660 nanometers. In this way, the algorithm will search for spectrum peaks within this range, improving the accuracy and efficiency of detection.

[0121] S802: Use the peak detection algorithm to find the spectrum peak H. This step uses mathematical methods or signal processing technology to find the highest signal strength point within the set search range K, thereby determining the location H of the spectrum peak. Specifically, methods such as Gaussian fitting or Fourier transform can be used to identify and accurately locate the spectrum peak. These algorithms can effectively eliminate the interference of background noise and improve the accuracy of spectrum peak identification. For example, the time domain signal is converted into a frequency domain signal through Fourier transform, and then the peak is searched within the specified frequency range.

[0122] S803: Verify whether the peak position H is within the theoretical spectrum range L of the marker. This is a key step to ensure that the spectrum peak found does belong to the target marker. Known markers usually have a theoretical spectrum range L, such as 645 nanometers to 655 nanometers. During the verification process, the peak position H is compared with this theoretical range L. If H falls within the range L, it is considered that the peak found is the true spectrum peak of the marker. Otherwise, it is necessary to recheck the data or adjust the search range.

[0123] S804: If the position H is within L, it is confirmed as the spectrum peak of the marker. Once the position of the spectrum peak is confirmed, it can be used for subsequent analysis and calculations, such as concentration determination or disease diagnosis. For example, in one embodiment, a known marker is added to a blood sample, and the theoretical spectrum range L of the marker is 645 nanometers to 655 nanometers. Through the above steps, the spectrum peak position H is determined to be 650 nanometers, and it is verified that H is indeed within the range of L, and finally 650 nanometers is confirmed as the spectrum peak position of the marker.

[0124] In this process, the purpose of setting the search range K is to reduce the amount of calculation and improve the detection accuracy. Generally speaking, the search range K should be as narrow as possible, but sufficient to cover all possible peak positions of the marker. In theory, the narrower K is, the faster the calculation speed is, but too narrow may lead to missing the real peak. The selection of the optimal value needs to be combined with the actual application scenario. For example, in an environment with high background noise, the range of K can be appropriately relaxed to increase the robustness of the detection.

[0125] Overall, these steps ensure the accurate localization of the marker fluorescence frequency through rigorous search, verification and confirmation, thereby improving the reliability and accuracy of blood testing.

[0126] Next, refer to Fig. 9 , describing the steps of verifying whether the peak position H is within the theoretical spectrum range L of the marker according to the present invention:

[0127] S901: Calculate the deviation N between the peak position H and the theoretical spectrum position M of the marker;

[0128] S902: If the deviation N is less than the set threshold O, it is confirmed as a valid spectrum peak;

[0129] S903: Record the information of spectrum peak H;

[0130] S904: Continue searching for the next possible peak.

[0131] First, calculate the deviation N between the peak position H and the theoretical spectral position M of the marker. Specifically, the deviation N is calculated by the formula N = |HM|, where H represents the actual detected spectral peak position and M represents the theoretical spectral position of the marker. The unit of deviation N is usually wavelength or frequency unit, such as nanometers or hertz. The purpose of this formula is to quantify the difference between the actual detected spectral peak and the theoretical expectation. Usually, the theoretical spectral position M of the marker is determined by literature data or known chemical properties, while the actual measured spectral peak position H is obtained by instrument measurement. In order to ensure the measurement accuracy, M can range from a few nanometers to hundreds of nanometers, and the maximum acceptable value of deviation N is generally within a few wavelength units.

[0132] For example, in one embodiment, assuming that the theoretical spectrum position M of the marker is 530 nm, and the spectrum peak position H actually detected is 529 nm, then the deviation N = |529 nm - 530 nm | = 1 nm. According to laboratory experience, the threshold O of the deviation N is set to 2 nm, so the deviation N is less than the threshold O.

[0133] Next, determine whether the deviation N is less than the set threshold O. If the deviation N is less than the threshold O, the spectrum peak is confirmed to be a valid spectrum peak. In this step, the threshold O is a pre-set value used to distinguish between valid peaks and noise. The selection of threshold O is based on experimental experience and the accuracy of the equipment. For blood testing applications, the common threshold range is 1nm to 5nm, and the preferred value is 2nm. This is because a threshold that is too small may miss the real peak, while a threshold that is too large may misjudge the background noise.

[0134] Next, record the information of the spectrum peak H. The recorded information includes, but is not limited to, the position, intensity, width, etc. of the spectrum peak. Specifically, during the recording process, the value, peak intensity, and other characteristic parameters of H are saved in a database or result report for subsequent analysis and verification. For example, the position of the spectrum peak H is 529 nm, the peak intensity is 250 arbitrary units, and the peak width is 15 nm.

[0135] Finally, the search continues for the next possible peak. After confirming that a spectrum peak is a valid peak, the system will rescan the spectrum graph to find the next possible peak and repeat the above verification steps. This process continues until all peaks in the spectrum graph have been verified.

[0136] Throughout the validation process, each step is designed to ensure the accuracy and reliability of the test results, thereby providing credible data support in the blood testing method.

[0137] Next, refer to Fig.10 , describing the steps of the present invention based on recording the information of the spectrum peak H.

[0138] S1001: Record the intensity P of the spectrum peak H into the data table. This means that during the spectrum analysis process, the detected spectrum peak and its corresponding intensity are accurately recorded. The purpose of this is to ensure that all possible signals are recorded in detail for subsequent analysis. For example, in one embodiment, the spectrum analyzer detects an absorbance value of 1.2 at a certain wavelength, and this value is recorded as the intensity P of the spectrum peak H.

[0139] S1002: Check whether there are multiple spectrum peaks Q that meet the conditions. This means that after recording all spectrum peaks H, these peaks are further screened to confirm whether there are multiple peaks that meet the pre-set standards. For example, a peak with a threshold value of more than 0.8 is set as a spectrum peak that meets the standard. Specifically, during the analysis of a blood sample, there may be more than two wavelengths where the absorbance exceeds 0.8, and these peaks need to be further processed.

[0140] S1003: If there are multiple spectral peaks Q that meet the criteria, further verification is performed. When multiple spectral peaks that meet the criteria are found, a more stringent verification method is required to distinguish the validity of these peaks. For example, different solvents can be added within the same wavelength range, and the test can be performed again to observe the changes in the spectrum. Specifically, if the presence of a certain solvent causes a certain spectral peak to disappear or weaken, while other peaks do not change significantly, then this step can eliminate the interfering peaks and determine the valid spectral peaks.

[0141] S1004: Determine the final valid spectrum peak H and analyze it. After the above verification, it is finally confirmed which spectrum peaks are truly meaningful, and then these peaks are analyzed in detail. For example, it is determined that a certain spectrum peak is caused by the presence of a specific protein, and then the protein content or other important information in the blood sample is inferred.

[0142] In this process, the parameter H represents the position of the spectrum peak, P represents the intensity of the spectrum peak, and Q represents multiple spectrum peaks that meet the standards. The position H of the spectrum peak is usually a measurement value at a specific wavelength, and its range depends on the measurement range of the spectral analysis instrument. The intensity P of the spectrum peak is usually the absorbance or fluorescence intensity value detected at that wavelength, and its range is also determined by the sensitivity of the instrument. The optimal spectrum peak H and intensity P need to be determined in combination with the specific application scenario and background, but generally speaking, high accuracy and high sensitivity are the main goals of optimization.

[0143] For example, in a blood test method for detecting a specific protein, if the standard threshold is set to 0.8, the instrument detects an absorbance value of 1.2 at a wavelength of 650nm, records the peak value and its intensity, and then continues to detect the absorbance values ​​at wavelengths of 700nm and 750nm, which are 1.0 and 0.9, respectively. Since all three values ​​exceed 0.8, they are preliminarily determined to be spectrum peak Q. Through further verification, such as retesting after adding a specific solvent, it is found that only the peaks at 650nm and 700nm remain unchanged, while the peak at 750nm is significantly weakened. Finally, the effective spectrum peak H is determined to be 650nm and 700nm, and further analysis can obtain specific protein information in the blood sample.

[0144] Through the above steps, the present invention not only realizes an efficient multi-target blood detection method, but also effectively solves the background noise problem commonly found in multi-label fluorescence detection, thereby greatly improving the reliability and accuracy of the detection results.

[0145] The specific implementation methods described above further illustrate the purpose, technical solutions and beneficial effects of the embodiments of the present disclosure. It should be understood that the above description is only the specific implementation method of the embodiments of the present disclosure and is not intended to limit the protection scope of the embodiments of the present disclosure. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the embodiments of the present disclosure should be included in the protection scope of the embodiments of the present disclosure.

Claims

1. A blood testing method, characterized in that: include: Performing multiple labeling fluorescent staining on the blood sample to be tested; The fluorescence signals of multiple labels were collected by fluorescence microscopy; Processing the fluorescent signal based on a signal separation algorithm to reduce background noise; The processed fluorescence signal is analyzed to obtain the detection result.

2. A blood testing method according to claim 1, characterized in that: Processing the fluorescence signal based on a signal separation algorithm to reduce background noise comprises: De-noising the fluorescence signal based on the preprocessing step to reduce environmental interference; The main spectrum components of each fluorescent marker are extracted using spectrum analysis method; The principal component analysis method was applied to further separate the cross-interference signals; The low-intensity background signal is removed by the filter, and the signal with high signal-to-noise ratio is retained.

3. A blood testing method according to claim 2, characterized in that: The denoising process of the fluorescence signal based on the preprocessing step comprises: Smoothing the raw fluorescence signal; Detect and remove spike noise based on sliding window method; Use wavelet transform technology to perform multi-scale denoising on the signal; If the noise intensity exceeds the threshold A, deep denoising is performed; Wherein, A represents the set noise intensity threshold, which is used to determine whether deep denoising is required.

4. A blood testing method according to claim 3, characterized in that: The method of detecting and removing spike noise based on a sliding window method includes: Set the initial sliding window size B; Calculate the mean C and standard deviation D of the signal within the sliding window; Compare the current point with the average value C in the window. If the deviation exceeds K×D, it is marked as spike noise. Move the sliding window to continue processing subsequent data; Among them, B represents the size of the sliding window, C represents the average value of the signal in the sliding window, D represents the standard deviation of the signal in the sliding window, and K represents the deviation coefficient.

5. A blood testing method according to claim 4, characterized in that: The comparing the current point with the average value C in the window, if the deviation exceeds K×D, marking it as a spike noise also includes: If the signal strength E at the current point exceeds the window average C plus K×D or is lower than the window average C minus K×D, it is marked as spike noise; Use interpolation methods to repair points marked as spike noise; Update the mean value C and standard deviation D within the sliding window; Repeat the steps until all data points have been processed; Among them, E represents the signal strength of the current point, C represents the signal average value in the sliding window, D represents the signal standard deviation in the sliding window, and K represents the deviation coefficient.

6. A blood testing method according to claim 5, characterized in that: The updating of the mean value C and the standard deviation D in the sliding window includes: Whenever the sliding window moves forward, the data point at the beginning of the window is removed and a new data point F is added; Recalculate the mean C and standard deviation D within the sliding window; Check whether the new data point F is spike noise, if so, continue repairing; Record the processing status of all data points within the window; Among them, F represents the data point of the newly added sliding window, C represents the signal average value in the sliding window, and D represents the signal standard deviation in the sliding window.

7. A blood testing method according to claim 2, characterized in that: The method of extracting the main spectrum components of each fluorescent marker by spectrum analysis includes: Perform Fourier transform on the fluorescence signal to obtain the frequency domain diagram; According to the fluorescence frequency characteristic G of the known marker, the peak position H of the spectrum is located; The height I of each peak is calculated to quantify the signal intensity of the marker; If the peak height I exceeds the set threshold J, the spectrum component is confirmed to be the main component; Wherein, G represents the fluorescence frequency characteristic of a known marker, H represents the position of the spectrum peak, I represents the height of the spectrum peak, and J represents the threshold of the signal intensity.

8. A blood testing method according to claim 7, characterized in that: The locating the spectrum peak position H according to the fluorescence frequency characteristic G of the known marker comprises: Set the search range K and search around the preset marked frequency point; Use the peak detection algorithm to find the spectrum peak H; Verify whether the peak position H is within the theoretical spectrum range L of the marker; If position H is within L, it is confirmed as the spectrum peak of the marker; Wherein, K represents the search range, H represents the position of the spectrum peak, and L represents the theoretical spectrum range of the marker.

9. A blood testing method according to claim 8, characterized in that: The verification of whether the peak position H is within the theoretical spectrum range L of the marker also includes: Calculate the deviation N between the peak position H and the theoretical spectrum position M of the marker; If the deviation N is less than the set threshold O, it is confirmed as a valid spectrum peak; Record the information of spectrum peak H; Continue searching for the next possible peak; Wherein, H represents the position of the spectrum peak, M represents the theoretical spectrum position of the marker, N represents the deviation value, and O represents the threshold value.

10. A blood testing method according to claim 8, characterized in that: The information of recording the spectrum peak H also includes: Record the intensity P of the spectrum peak H into the data table; Check whether there are multiple spectrum peaks Q that meet the conditions; If there are multiple spectrum peaks Q that meet the criteria, further verification is performed; Determine the final effective spectrum peak H and analyze it; Wherein, H represents the position of the spectrum peak, P represents the intensity of the spectrum peak, and Q represents multiple spectrum peaks that meet the standard.

Images