Peripheral blood mononuclear cell graphic data identification method, device and equipment

Abstract

The invention relates to a peripheral blood mononuclear cell graphic data recognition method, device and equipment, and belongs to the field of cell graphic recognizing.The peripheral blood mononuclear cell graphic data recognition method comprises the steps that through a sorting system combining immunomagnetic beads and a micro-fluidic chip technology, mononuclear cells are sorted and enriched in a high-flux mode from peripheral blood, and cell images are collected through a full-automatic cell morphology analyzer; the marker is detected and recognized by means of self-developed artificial intelligence image recognition algorithm software, finally recognition data is compared with a normal coronary artery contrast database, and a coronary heart disease risk assessment report is generated by combining ROC curve analysis. The method comprises the following specific steps: after immunomagnetic beads are specifically combined with mononuclear cells, sorting is realized under the action of a micro-fluidic chip and a magnetic field; accurate setting of camera parameters and confirmation of double light source focuses ensure the image acquisition quality; and the AI algorithm realizes marker pixel-level positioning through feature extraction and mode matching.

Description

Technical Field

[0001] The present invention relates to the technical field of cell pattern recognition, and in particular to a method, device and equipment for recognizing peripheral blood mononuclear cell pattern data. Background Art

[0002] Currently, the recognition of graphical data from peripheral blood mononuclear cells (monocytes) in early screening for coronary artery disease (CAD) faces numerous technical bottlenecks. Traditional sorting methods (such as density gradient centrifugation) suffer from low throughput and significant cell viability impairment, making it difficult to efficiently enrich monocytes from trace amounts of peripheral blood, resulting in insufficient sample quantities for subsequent testing. During image acquisition, manual microscopy is not only time-consuming and laborious, but focus adjustment errors can easily distort cell morphology and fluorescence signals, affecting data accuracy. Traditional machine learning-based marker recognition algorithms, lacking optimized characteristic parameters for CAD-specific markers (such as the NFAM1 protein), often result in false-positive identification or positioning errors, and are unable to effectively correlate with clinical typing data. Furthermore, existing analytical methods often rely on single-indicator judgments and lack a comprehensive evaluation system integrated with clinical validation data, making it difficult to meet the sensitivity and specificity requirements for early screening. Summary of the Invention

[0003] The main purpose of the present invention is to provide a method, device and equipment for peripheral blood mononuclear cell graphic data recognition, improve the efficiency of mononuclear cell sorting and image recognition accuracy, and provide a quantitative diagnostic basis for early screening of coronary heart disease.

[0004] To achieve the above-mentioned object, the present invention provides a method for identifying peripheral blood mononuclear cell pattern data, comprising the following steps:

[0005] Obtaining a peripheral blood sample and using a sorting system combining immunomagnetic beads and microfluidic chip technology to high-throughput sort and enrich monocytes from the peripheral blood sample;

[0006] The images of the sorted monocytes were collected using a fully automatic cell morphology analyzer;

[0007] Utilizes software embedded with self-developed artificial intelligence image recognition algorithms to detect and identify markers in collected monocyte images;

[0008] The identified graphic data is compared and analyzed with a preset normal coronary artery control database to generate the identification results of the monocyte graphic data.

[0009] Furthermore, the step of high-throughput sorting and enriching monocytes from the peripheral blood sample using a sorting system combining immunomagnetic beads and microfluidic chip technology includes:

[0010] The immunomagnetic beads are mixed with the peripheral blood sample to allow the immunomagnetic beads to specifically bind to the monocyte surface markers;

[0011] The mixed sample is introduced into the microfluidic chip, and the fluid control technology of the microfluidic chip is used to guide the sample to flow within the chip;

[0012] By applying an external magnetic field to the immunomagnetic beads in the microfluidic chip, the immunomagnetic beads bound to monocytes are directionally deflected in the chip, thereby separating them from other cells.

[0013] The fluid fraction containing monocytes after magnetic field sorting is collected to achieve high-throughput sorting and enrichment of monocytes from peripheral blood samples.

[0014] Furthermore, the step of collecting images of the sorted monocytes using a fully automatic cell morphology analyzer includes:

[0015] The sorted and enriched mononuclear cell sample is introduced into the sample detection chamber of the fully automatic cell morphology analyzer;

[0016] Set the camera's shooting parameters, including the time interval, number of horizontal and vertical shots, and distance interval. If the distance interval is less than 12 steps, the number of horizontal or vertical shots should not exceed 10. Set the stitching mode to 2×2 or 3×3.

[0017] Confirm the white light focus and fluorescence focus to ensure that the lens is accurately focused on the mononuclear cell sample;

[0018] Select single or continuous shooting mode to capture images of monocytes and obtain graphic data including cell morphology and marker fluorescence signals.

[0019] Furthermore, the step of collecting images of monocytes to obtain graphic data including cell morphology and marker fluorescence signals includes:

[0020] When collecting images of sorted monocytes using a fully automated cell morphology analyzer, the camera properties were first set, including the time interval and the number of horizontal and vertical frames. When the distance interval was less than 12 steps, the number of horizontal or vertical frames was controlled to not exceed 10, and the stitching mode was set to 2×2 or 3×3.

[0021] After completing the camera parameter settings, perform white light focus confirmation and fluorescence focus confirmation operations in sequence to ensure that the lens is accurately focused on the mononuclear cell sample and ensure the clarity of cell morphology and marker fluorescence signals during shooting;

[0022] Select single-shot or continuous-shot mode according to the detection requirements to capture images of monocyte samples. During the shooting process, white light image data reflecting the cell morphological characteristics and fluorescence image data corresponding to the marker fluorescence signal are simultaneously obtained to form a graphic data set containing cell morphology and marker fluorescence signals.

[0023] Furthermore, the steps of detecting and identifying markers in the collected monocyte images using the embedded independently developed artificial intelligence image recognition algorithm software include:

[0024] Importing the collected monocyte image data into the artificial intelligence image recognition algorithm software, performing noise reduction and contrast enhancement preprocessing on the image to eliminate noise interference generated during the shooting process;

[0025] Using the custom feature extraction model built into the algorithm software, cell morphology features and fluorescence signal features are extracted from the pre-processed image to generate a feature vector set;

[0026] Based on the feature vector set, the markers in the monocyte image are identified and located using a pattern rule library preset in the algorithm software, wherein the rule library includes threshold values ​​of characteristic parameters of fluorescence signals for coronary heart disease-related markers;

[0027] The confidence analysis is performed on the identified marker signal. When the signal strength and distribution characteristics meet the preset threshold, it is determined to be a valid marker detection result, and an identification report with the marker position coordinates and quantitative parameters is generated.

[0028] Furthermore, based on the feature vector set, the steps of identifying and locating markers in the monocyte image using a pattern rule library preset in the algorithm software include:

[0029] Inputting the feature vector set into a pattern recognition rule base preset in the algorithm software, wherein the rule base is constructed based on the fluorescence signal characteristic parameters of the NFAM1 protein marker in the peripheral blood mononuclear cells of patients with coronary heart disease, and includes quantitative standards such as the fluorescence intensity threshold and the signal distribution area threshold;

[0030] Matching and analyzing the cell morphological features and fluorescence signal features in the feature vector set is performed using the rule base, and when the characteristic parameters of the fluorescence signal reach a preset threshold, marker recognition is triggered;

[0031] Based on the matching results, the specific position coordinates of the marker in the monocyte image are determined, and the fluorescence signal area corresponding to the feature vector is converted into image pixel-level positioning results through a coordinate mapping algorithm to achieve accurate positioning of the marker.

[0032] Furthermore, the step of comparing and analyzing the identified graphic data with a preset normal coronary artery control database to generate a recognition result of the monocyte graphic data includes:

[0033] Extracting the quantitative parameters of the identified markers to form a data set to be analyzed;

[0034] Calling a preset normal coronary artery control database, which is constructed based on mononuclear cell image data of 79 normal coronary artery control samples and contains normal reference range thresholds for characteristic parameters of each marker;

[0035] Compare the data set to be analyzed with the normal reference range threshold in the database. When the characteristic parameters of the marker exceed the normal threshold range, it is determined to be an abnormal signal;

[0036] ROC curve analysis was used to verify abnormal signals and calculate the AUC value, sensitivity, and specificity statistical indicators. The indicators were used to set the discrimination criteria based on clinical validation data.

[0037] Based on the comparison results and statistical analysis, a monocyte graphic data identification report containing a coronary heart disease risk assessment is generated, wherein the report associates the correspondence between the abnormality degree of the markers and the clinical classification of coronary heart disease.

[0038] Furthermore, the ROC curve analysis method is used to verify the abnormal signal and calculate the statistical indicators of AUC value, sensitivity and specificity. The steps of setting the discrimination criteria based on clinical validation data include:

[0039] Organizing marker characteristic parameters and clinical grouping data corresponding to abnormal signals, wherein the clinical grouping includes peripheral blood sample data of normal coronary artery control, stable coronary heart disease, and acute coronary syndrome;

[0040] The ROC curve was constructed using the marker characteristic parameters as the test variables and the clinical grouping results as the state variables. The AUC value was calculated by the area under the curve, and the discrimination threshold was set based on the clinical validation data.

[0041] The optimal critical value was determined based on the ROC curve, and the sensitivity and specificity indices were calculated and used as quantitative criteria for determining the correlation between abnormal signals and coronary heart disease;

[0042] The calculated statistical indicators are compared with the preset standards in the clinical verification data. When the indicators reach or exceed the preset threshold, the abnormal signal is determined to have auxiliary diagnostic value for coronary heart disease.

[0043] The present invention provides a peripheral blood mononuclear cell pattern data recognition device, comprising:

[0044] An acquisition unit, configured to obtain a peripheral blood sample and to perform high-throughput sorting and enrichment of monocytes from the peripheral blood sample using a sorting system combining immunomagnetic beads with microfluidic chip technology;

[0045] An imaging unit, used for collecting images of the sorted monocytes using a fully automatic cell morphology analyzer;

[0046] A detection unit, which is used to detect and identify markers in collected monocyte images using embedded independently developed artificial intelligence image recognition algorithm software;

[0047] The analysis unit is used to compare and analyze the identified graphic data with a preset normal coronary artery control database to generate an identification result of the monocyte graphic data.

[0048] The present invention also provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above-mentioned peripheral blood mononuclear cell graphic data recognition method when executing the computer program.

[0049] The method, device, and apparatus for identifying peripheral blood mononuclear cell pattern data provided by the present invention have the following beneficial effects:

[0050] (1) By combining immunomagnetic beads with microfluidic chips, high-throughput automated sorting of peripheral blood mononuclear cells is achieved.

[0051] (2) With the help of a fully automatic cell morphology analyzer and AI image recognition algorithm, the positioning accuracy of marker detection is improved to the pixel level, reducing manual interpretation errors.

[0052] (3) Construct a comparative analysis system that includes clinical validation data and use the statistical method of ROC curve to achieve quantitative assessment of CHD risk. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 1 is a flow chart of a method for identifying peripheral blood mononuclear cell pattern data in one embodiment of the present invention;

[0054] Figure 2 This is a structural block diagram of a peripheral blood mononuclear cell pattern data recognition device in one embodiment of the present invention;

[0055] Figure 3 It is a schematic block diagram of the structure of a computer device according to an embodiment of the present invention.

[0056] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0057] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0058] Reference Figure 1 The following is a flow chart of a method for identifying peripheral blood mononuclear cell pattern data proposed by the present invention, which includes the following steps:

[0059] S1, obtaining a peripheral blood sample and using a sorting system combining immunomagnetic beads and microfluidic chip technology to high-throughput sort and enrich monocytes from the peripheral blood sample;

[0060] S2, image acquisition of sorted monocytes using a fully automatic cell morphology analyzer;

[0061] S3, uses embedded self-developed artificial intelligence image recognition algorithm software to detect and identify markers in the collected monocyte images;

[0062] S4, comparing and analyzing the identified graphic data with a preset normal coronary artery control database to generate a recognition result of the monocyte graphic data.

[0063] For S1, high-throughput sorting and enrichment of peripheral blood mononuclear cells (monocytes) achieves specific isolation of monocytes by combining immunomagnetic beads with microfluidic chip technology. Peripheral blood samples (including normal coronary artery controls, samples from patients with stable coronary artery disease, and samples from acute coronary syndrome) are collected, mixed with immunomagnetic beads, and antibodies on the surface of the beads specifically bind to monocyte surface markers (such as CD14). The mixed sample is then introduced into a microfluidic chip, where the microchannel structure within the chip controls the flow of fluid. Simultaneously, an external magnetic field is applied, causing the monocytes bound to the magnetic beads to undergo a directional deflection under the action of the magnetic field, separating them from other cells. Finally, the fluid fraction containing monocytes is collected.

[0064] For S2, automated monocyte image acquisition involves introducing the sorted monocyte sample into the detection chamber of a fully automated cell morphology analyzer and initiating the camera parameter setup program: setting the distance interval to 10 steps (when the distance interval is less than 12 steps, the number of horizontal / vertical images is controlled to less than 10), and selecting a 2×2 or 3×3 stitching mode to cover the entire field of view of cell distribution. Subsequently, dual confirmation of white light and fluorescence focus is performed, and automatic Z-axis adjustment is used to precisely focus the lens on the cellular level to ensure clarity of cell morphology and marker fluorescence signals. Depending on the sample size, single or continuous shooting mode is selected to simultaneously capture white light images (reflecting morphological characteristics such as cell outlines and nuclear-cytoplasmic ratio) and fluorescence images (marker fluorescence signal distribution), forming a graphical data set containing multi-dimensional features.

[0065] For S3, AI-based marker detection and identification involves importing collected image data into proprietary AI image recognition algorithm software. Initially, preprocessing with noise reduction and contrast enhancement is performed to eliminate optical noise and background interference. Using a built-in feature extraction model, cell outlines are automatically identified, morphological features such as the nuclear-cytoplasmic ratio are calculated, and parameters such as the intensity and distribution area of ​​the fluorescence signal are extracted to generate a set of feature vectors. These feature vectors are then input into a pre-set pattern recognition rule library (based on the fluorescence signal characteristics of the coronary artery disease-related marker NFAM1 protein, which includes quantitative criteria for fluorescence intensity thresholds and signal distribution area percentages). Algorithmic matching analysis triggers the marker recognition mechanism, ultimately locating the pixel-level coordinates of the marker within the cell and generating an identification report containing signal intensity and position parameters.

[0066] For S4, graphical data comparison and analysis, along with result generation, extracts marker quantitative parameters (such as fluorescence intensity and cell morphology) from the identification report. A database constructed from 79 normal coronary artery control samples stores the normal reference range thresholds for each marker characteristic parameter. The data to be analyzed is compared with the thresholds, and signals outside the range are classified as abnormal. Receiver-operating characteristic (ROC) curve analysis is used to validate abnormal signals. A curve is constructed using marker parameters as test variables and clinical groups (normal controls / stable coronary artery disease / acute coronary syndrome) as status variables. Area under the curve (AUC), sensitivity, and specificity are calculated, and discrimination criteria are set based on clinical validation data (e.g., AUC > 0.8, sensitivity > 70%, and specificity > 85%). The resulting identification report correlates the degree of marker abnormality with the clinical classification of coronary artery disease.

[0067] In one embodiment, the step of high-throughput sorting and enriching monocytes from the peripheral blood sample using a sorting system combining immunomagnetic beads and microfluidic chip technology includes:

[0068] The immunomagnetic beads are mixed with the peripheral blood sample to allow the immunomagnetic beads to specifically bind to the monocyte surface markers;

[0069] The mixed sample is introduced into the microfluidic chip, and the fluid control technology of the microfluidic chip is used to guide the sample to flow within the chip;

[0070] By applying an external magnetic field to the immunomagnetic beads in the microfluidic chip, the immunomagnetic beads bound to monocytes are directionally deflected in the chip, thereby separating them from other cells.

[0071] The fluid fraction containing monocytes after magnetic field sorting is collected to achieve high-throughput sorting and enrichment of monocytes from peripheral blood samples.

[0072] In specific implementation, the immunomagnetic beads are mixed with the peripheral blood sample. The specific antibodies coated on the surface of the magnetic beads (such as antibodies against monocyte surface markers) will undergo an antigen-antibody binding reaction with the monocytes, causing the immunomagnetic beads to specifically attach to the surface of the monocytes. Subsequently, the mixed sample is introduced into a microfluidic chip. The chip uses microchannel structure design and fluid drive technology (such as peristaltic pump or pneumatic drive) to achieve precise control of the sample flow path, guiding the sample to flow evenly along a preset trajectory within the chip, ensuring that monocytes and other cells maintain an orderly distribution before the magnetic field acts. When an external magnetic field (such as a gradient magnetic field) is applied to the microfluidic chip, the immunomagnetic beads bound to the monocytes are subjected to magnetic force due to their magnetic response characteristics, resulting in a directional deviation within the chip, while other cells (such as lymphocytes) that are not bound to the magnetic beads flow along the original fluid path, thereby achieving physical separation of monocytes from other cells. Finally, a collection device is set at the specific outlet position of the microfluidic chip to collect the fluid portion containing monocytes after magnetic field sorting to obtain a high-purity monocyte sample. This sorting system can achieve high-throughput enrichment of monocytes from trace amounts of peripheral blood, meeting the sample volume and activity requirements of subsequent testing.

[0073] In one embodiment, the step of collecting images of the sorted monocytes using a fully automatic cell morphology analyzer includes:

[0074] The sorted and enriched mononuclear cell sample is introduced into the sample detection chamber of the fully automatic cell morphology analyzer;

[0075] Set the camera's shooting parameters, including the time interval, number of horizontal and vertical shots, and distance interval. If the distance interval is less than 12 steps, the number of horizontal or vertical shots should not exceed 10. Set the stitching mode to 2×2 or 3×3.

[0076] Confirm the white light focus and fluorescence focus to ensure that the lens is accurately focused on the mononuclear cell sample;

[0077] Select single or continuous shooting mode to capture images of monocytes and obtain graphic data including cell morphology and marker fluorescence signals.

[0078] During the specific implementation, the sorted and enriched monocyte samples are introduced into the sample detection chamber of the fully automatic cell morphology analyzer, which has a constant temperature and humidity environment control function to maintain the cell activity state. Then enter the camera shooting parameter setting stage. The operator needs to set the time interval to control the frequency of continuous shooting according to the sample concentration and detection requirements, and accurately adjust the number of horizontal and vertical shots and the distance interval. When the distance interval is less than 12 steps, the number of horizontal or vertical shots is strictly controlled within 10 to avoid data redundancy caused by excessive shooting density; the stitching image mode is set to 2×2 or 3×3, and full field of view coverage is achieved through multi-area image stitching to ensure that no cell samples are missed. After the parameter setting is completed, the white light focus and fluorescence focus confirmation operations are carried out in turn. Through the built-in autofocus system of the instrument, the lens is accurately focused on the monocyte sample under white light and fluorescence light sources respectively, eliminating the problem of blurred cell morphology or fluorescence signal distortion caused by focal length deviation. Finally, depending on the sample size and detection accuracy requirements, single-shot or continuous shooting mode is selected to start image acquisition. Two sets of data are acquired synchronously during the shooting process: one set is a white light image, which is used to record the morphological characteristics of cell contours and nuclear-cytoplasmic ratio; the other set is a fluorescence image, which is used to capture the fluorescence signal distribution of markers (such as NFAM1 protein), ultimately forming a multidimensional graphical data set containing cell morphology and marker fluorescence signals.

[0079] In one embodiment, the step of capturing images of monocytes to obtain graphic data including cell morphology and marker fluorescence signals includes:

[0080] When collecting images of sorted monocytes using a fully automated cell morphology analyzer, the camera properties were first set, including the time interval and the number of horizontal and vertical frames. When the distance interval was less than 12 steps, the number of horizontal or vertical frames was controlled to not exceed 10, and the stitching mode was set to 2×2 or 3×3.

[0081] After completing the camera parameter settings, perform white light focus confirmation and fluorescence focus confirmation operations in sequence to ensure that the lens is accurately focused on the mononuclear cell sample and ensure the clarity of cell morphology and marker fluorescence signals during shooting;

[0082] Select single-shot or continuous-shot mode according to the detection requirements to capture images of monocyte samples. During the shooting process, white light image data reflecting the cell morphological characteristics and fluorescence image data corresponding to the marker fluorescence signal are simultaneously obtained to form a graphic data set containing cell morphology and marker fluorescence signals.

[0083] During implementation, after starting the image acquisition program, the camera properties must first be systematically configured: The time interval is set to control the shooting frequency based on the sample detection requirements, and the number of horizontal and vertical shots and the distance interval are simultaneously adjusted. When the distance interval is less than 12 steps, the number of horizontal or vertical shots is strictly controlled to less than 10 to avoid data redundancy and increased processing load due to excessive shooting density. At the same time, the stitching mode is set to 2×2 or 3×3, and multi-region image stitching technology is used to achieve full coverage of the detection field of view to ensure that no cell samples are missed. After completing the parameter configuration, the white light focus and fluorescence focus are double-checked in sequence: using the instrument's built-in autofocus system, the lens is adjusted to the clearest state of monocyte outlines under white light, and then the focus is calibrated by switching to the fluorescence light source to calibrate the fluorescent signal of the marker. Fine adjustment of the Z-axis displacement ensures that the imaging under both light sources is at the optimal focus, fundamentally avoiding cell morphological distortion or blurred fluorescence signals caused by focal length deviation. Finally, based on the sample size and detection accuracy requirements, select single-shot or continuous shooting mode to start image acquisition: During the shooting process, the instrument synchronously activates the white light imaging channel and the fluorescence imaging channel. The former records the morphological characteristics of cell contours and nuclear-cytoplasmic ratio in real time, and the latter captures the fluorescence signal distribution and intensity parameters of markers (such as NFAM1 protein), ultimately forming a multidimensional graphic data set containing cell morphology and marker signals. In this process, the dynamic adjustment mechanism of camera parameters and the dual-light source focus calibration process directly determine the reliability of the graphic data. The clear correspondence between the distance interval and the number of shots in the document, as well as the standardized setting of the stitching image mode, are the core technical points for achieving high-throughput and high-precision image acquisition.

[0084] In one embodiment, the steps of detecting and identifying markers in the collected monocyte images using software embedded with a self-developed artificial intelligence image recognition algorithm include:

[0085] Importing the collected monocyte image data into the artificial intelligence image recognition algorithm software, performing noise reduction and contrast enhancement preprocessing on the image to eliminate noise interference generated during the shooting process;

[0086] Using the custom feature extraction model built into the algorithm software, cell morphology features and fluorescence signal features are extracted from the pre-processed image to generate a feature vector set;

[0087] Based on the feature vector set, the markers in the monocyte image are identified and located using a pattern rule library preset in the algorithm software, wherein the rule library includes threshold values ​​of characteristic parameters of fluorescence signals for coronary heart disease-related markers;

[0088] The confidence analysis is performed on the identified marker signal. When the signal strength and distribution characteristics meet the preset threshold, it is determined to be a valid marker detection result, and an identification report with the marker position coordinates and quantitative parameters is generated.

[0089] In practice, the collected monocyte image data is imported into artificial intelligence image recognition algorithm software. The preprocessing module is then activated to perform image noise reduction (e.g., Gaussian filtering to eliminate random noise) and contrast enhancement (histogram equalization to highlight cell boundaries). This effectively eliminates noise caused by light source fluctuations or equipment interference during the capture process, thereby improving image quality. After preprocessing, the algorithm software's built-in custom feature extraction model is invoked. Based on a convolutional neural network architecture, this model performs deep learning extraction of cell morphological features (e.g., cell outline curvature, nuclear-cytoplasmic ratio, particle distribution) and fluorescence signal features (e.g., fluorescence intensity gradient, signal spot area, and fluorescence distribution symmetry) in the image. The extracted multidimensional feature parameters are converted into a standardized set of feature vectors, providing the data foundation for subsequent recognition.

[0090] Based on the generated feature vector set, the algorithm software further invokes a pre-defined pattern rule library for marker identification and localization. This rule library, constructed based on clinical validation data for the NFAM1 protein marker in peripheral blood mononuclear cells from patients with coronary artery disease, includes quantitative criteria for fluorescence intensity thresholds (e.g., ≥2 times the baseline value) and signal distribution area thresholds (e.g., ≥15% of the cell area). Through matching analysis between the feature vectors and the rule library, the marker recognition mechanism is triggered when fluorescence signal characteristic parameters (e.g., fluorescence intensity and signal distribution uniformity) reach the preset thresholds. A coordinate mapping algorithm is then used to convert the fluorescence signal area corresponding to the feature vector into pixel-level localization results, enabling precise localization of the marker within the cell. Finally, the identified marker signal undergoes confidence analysis by calculating the signal intensity agreement and distribution feature matching rate. When both the signal intensity and distribution features meet the preset thresholds, the result is considered valid and an identification report is automatically generated, including the marker's location coordinates, fluorescence intensity, and morphological parameters.

[0091] In one embodiment, based on the feature vector set, the steps of identifying and locating markers in the monocyte image using a pattern rule library preset in the algorithm software include:

[0092] Inputting the feature vector set into a pattern recognition rule base preset in the algorithm software, wherein the rule base is constructed based on the fluorescence signal characteristic parameters of the NFAM1 protein marker in the peripheral blood mononuclear cells of patients with coronary heart disease, and includes quantitative standards such as the fluorescence intensity threshold and the signal distribution area threshold;

[0093] Matching and analyzing the cell morphological features and fluorescence signal features in the feature vector set is performed using the rule base, and when the characteristic parameters of the fluorescence signal reach a preset threshold, marker recognition is triggered;

[0094] Based on the matching results, the specific position coordinates of the marker in the monocyte image are determined, and the fluorescence signal area corresponding to the feature vector is converted into image pixel-level positioning results through a coordinate mapping algorithm to achieve accurate positioning of the marker.

[0095] During specific implementation, the set of cell morphology and fluorescence signal feature vectors generated by the feature extraction model is input into the preset pattern recognition rule base in the algorithm software. This rule base is not a general image recognition database, but is specifically constructed based on the clinical detection data of NFAM1 protein in peripheral blood mononuclear cells of patients with coronary heart disease. For example, the NFAM1 fluorescence signal characteristic parameters of patients with stable coronary heart disease, acute coronary syndrome and normal coronary artery are included. After statistical analysis, an exclusive rule system with quantitative standards including fluorescence intensity threshold (such as ≥1.5 times the normal mean) and signal distribution area threshold (such as more than 10% of the total cell area) is formed.

[0096] Once the feature vectors are entered into the rule base, the algorithm simultaneously performs a multi-dimensional matching analysis of cell morphological features (such as the nuclear-to-cytoplasmic ratio and cell outline integrity) and fluorescence signal characteristics (such as peak intensity and spatial distribution uniformity of the fluorescence spot). For example, for the NFAM1 protein, if the fluorescence signal characteristic parameters meet the preset threshold combination of "fluorescence intensity exceeding 2 times the normal reference value and the signal distribution area continuously covering more than 15% of the cytoplasm," the marker recognition mechanism is triggered, thus avoiding false positive results caused by misjudgment of a single parameter. After feature matching, the system initiates a coordinate mapping algorithm based on the matching results. This algorithm establishes a spatial correspondence between feature vectors and image pixels, converting the fluorescence signal area represented by the feature vector into precise pixel-level coordinates. For example, if the feature vector indicates a fluorescence intensity gradient that matches the distribution characteristics of the NFAM1 protein, the algorithm maps the feature vector to the corresponding coordinate interval in the image matrix, localizing the marker at the submicroscopic level within the cell with a localization accuracy of up to a single pixel (approximately 1μm²).

[0097] This process combines the clinical data support of the rule base with the algorithm's multidimensional matching mechanism to form a closed-loop technology. First, the rule base's threshold criteria are directly derived from NFAM1 expression profile analysis of 332 clinical samples (including 79 normal controls, 70 cases of stable coronary artery disease, and 183 cases of acute coronary syndrome), ensuring that the recognition results are relevant to the clinical classification. Second, the coordinate mapping algorithm combines cell morphology with fluorescence signal characteristics to avoid positional deviations caused by cell morphological distortion in traditional fluorescence localization, reducing the error rate by over 35% compared to localization methods based solely on fluorescence signals. This embodiment converts the clinical characteristics of CHD-specific markers into algorithmic rules, achieving a precise conversion from image data to clinical diagnostic information, providing a quantitative basis for subsequent CHD risk assessment.

[0098] In one embodiment, the step of comparing and analyzing the identified graphic data with a preset normal coronary artery control database to generate a recognition result of the monocyte graphic data includes:

[0099] Extracting the quantitative parameters of the identified markers to form a data set to be analyzed;

[0100] Calling a preset normal coronary artery control database, which is constructed based on mononuclear cell image data of 79 normal coronary artery control samples and contains normal reference range thresholds for characteristic parameters of each marker;

[0101] Compare the data set to be analyzed with the normal reference range threshold in the database. When the characteristic parameters of the marker exceed the normal threshold range, it is determined to be an abnormal signal;

[0102] Abnormal signals were verified using ROC curve analysis, and statistical indicators such as AUC values, sensitivity, and specificity were calculated. The criteria for these indicators were set based on clinical validation data.

[0103] Based on the comparison results and statistical analysis, a monocyte graphic data identification report containing a coronary heart disease risk assessment is generated, wherein the report associates the correspondence between the abnormality degree of the markers and the clinical classification of coronary heart disease.

[0104] In practice, marker quantitative parameters, such as the fluorescence intensity of the NFAM1 protein and cell morphological parameters (nuclear-to-cytoplasmic ratio and cell outline curvature), are extracted from the AI ​​recognition report to form a structured data set for analysis. Subsequently, a pre-set normal coronary artery control database, constructed based on mononuclear cell image data from 79 normal coronary artery control samples confirmed by coronary angiography, is used. Statistical analysis of the marker characteristic parameters of these normal samples is performed to establish normal reference range thresholds for each marker characteristic parameter, such as the normal mean and 95% confidence interval for NFAM1 protein fluorescence intensity.

[0105] The data set to be analyzed is compared with the normal reference range thresholds in the database. When the marker characteristic parameter exceeds the normal threshold (e.g., fluorescence intensity exceeds 2 standard deviations above the normal mean), the system automatically identifies it as an abnormal signal. To verify the clinical relevance of the abnormal signal, receiver operating characteristic (ROC) curve analysis is used for statistical validation. A ROC curve is constructed using the marker characteristic parameter as the test variable and the clinical grouping results (normal coronary artery control, stable coronary artery disease, and acute coronary syndrome) as the state variable. The area under the curve (AUC) is calculated using the area under the curve (AUC). A discriminant criterion is established based on validation data from 332 clinical samples (79 normal controls, 70 stable coronary artery disease, and 183 acute coronary syndrome). For example, an AUC value greater than 0.8 indicates good diagnostic efficacy for the marker. Sensitivity and specificity are also calculated to quantitatively determine the correlation between the abnormal signal and coronary artery disease, ensuring the reliability of the test results.

[0106] Finally, based on the comparison results and statistical analysis, a monocyte graphic data identification report containing a coronary artery disease risk assessment is generated. This report achieves risk stratification by establishing a correspondence between the degree of marker abnormality and the clinical classification of coronary artery disease. For example, when the fluorescence intensity of the NFAM1 protein exceeds the normal threshold by 1.5 times, the report indicates the risk of stable coronary artery disease; when it exceeds 2 times, it is associated with the risk of acute coronary syndrome and a simultaneous risk probability assessment is provided. The report also includes a confidence analysis of the abnormal signal and details on the degree of deviation from the normal reference range, providing clinicians with a diagnostic reference that combines quantitative indicators with clinical relevance. This process, supported by clinical samples from a normal coronary artery control database and statistical validation through ROC curve analysis, provides dual guarantees, ensuring that the identification results have both a standardized basis for judgment and accurate correlation with the clinical classification of coronary artery disease.

[0107] In one embodiment, the abnormal signal is verified by ROC curve analysis, and statistical indicators such as AUC value, sensitivity, and specificity are calculated. The steps of setting the discrimination criteria based on clinical validation data include:

[0108] Organizing marker characteristic parameters and clinical grouping data corresponding to abnormal signals, wherein the clinical grouping includes peripheral blood sample data of normal coronary artery control, stable coronary heart disease, and acute coronary syndrome;

[0109] The ROC curve was constructed using the marker characteristic parameters as the test variables and the clinical grouping results as the state variables. The AUC value was calculated by the area under the curve, and the discrimination threshold was set based on the clinical validation data.

[0110] The optimal critical value was determined based on the ROC curve, and the sensitivity and specificity indices were calculated and used as quantitative criteria for determining the correlation between abnormal signals and coronary heart disease;

[0111] The calculated statistical indicators are compared with the preset standards in the clinical verification data. When the indicators reach or exceed the preset threshold, the abnormal signal is determined to have auxiliary diagnostic value for coronary heart disease.

[0112] In specific implementation, the researchers collated marker characteristic parameters corresponding to abnormal signals (such as NFAM1 protein fluorescence intensity and cell morphology parameters) and clinical grouping data, including peripheral blood samples from 79 normal coronary artery controls, 70 patients with stable coronary artery disease, and 183 patients with acute coronary syndrome (ACS) confirmed by coronary angiography, to form a structured dataset. Subsequently, using the marker characteristic parameters as test variables and the clinical grouping results (normal / stable / acute) as the state variable, they constructed a receiver operating characteristic (ROC) curve using statistical software. The diagnostic efficacy of the marker was evaluated by calculating the area under the curve (AUC). This AUC value was used to set a discrimination threshold based on validation data from 332 clinical samples. For example, an AUC value exceeding 0.8 indicates good diagnostic accuracy.

[0113] Next, the optimal cutoff value is determined based on the ROC curve. This cutoff value is obtained by balancing the parameters of sensitivity (true positive rate) and specificity (true negative rate). For example, the combination of 70.7% sensitivity and 85.9% specificity verified in the document serves as a quantitative standard for determining the correlation between abnormal signals and CHD. Finally, the calculated AUC value, sensitivity, and specificity statistical indicators are compared with pre-set standards in clinical validation data. When the indicators reach or exceed the preset thresholds (e.g., AUC ≥ 0.8, sensitivity ≥ 70%, specificity ≥ 85%), the abnormal signal is considered to have auxiliary diagnostic value for CHD and can be used to distinguish between normal coronary artery status and different clinical types of CHD. This process, by integrating large-sample clinical data with statistical analysis, transforms the graphical data features of the marker into quantitative indicators with clinical decision-making value, addressing the problems of large subjective interpretation errors and lack of a standardized evaluation system in traditional detection methods.

[0114] Reference Attachment Figure 2 This is a structural block diagram of a peripheral blood mononuclear cell pattern data recognition device proposed by the present invention, which includes:

[0115] An acquisition unit, configured to obtain a peripheral blood sample and to perform high-throughput sorting and enrichment of monocytes from the peripheral blood sample using a sorting system combining immunomagnetic beads with microfluidic chip technology;

[0116] An imaging unit, used for collecting images of the sorted monocytes using a fully automatic cell morphology analyzer;

[0117] A detection unit, which is used to detect and identify markers in collected monocyte images using embedded independently developed artificial intelligence image recognition algorithm software;

[0118] The analysis unit is used to compare and analyze the identified graphic data with a preset normal coronary artery control database to generate an identification result of the monocyte graphic data.

[0119] Reference Figure 3 In an embodiment of the present invention, a computer device is also provided. The computer device may be a server, and its internal structure may be as follows: Figure 3 As shown. The computer device includes a processor, memory, display screen, input device, network interface and database connected via a system bus. The processor of the computer design is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The database of the computer device is used to store the corresponding data in this embodiment. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, the above method is implemented.

[0120] Those skilled in the art will understand that Figure 3 The structure shown in the figure is merely a block diagram of a portion of the structure related to the solution of the present invention and does not constitute a limitation on the computer device to which the solution of the present invention is applied.

[0121] In summary, a sorting system combining immunomagnetic beads and microfluidic chip technology enables high-throughput sorting and enrichment of monocytes from peripheral blood. Cell images are acquired using a fully automated cell morphology analyzer. Biomarkers (such as NFAM1 protein) are detected and identified using proprietary artificial intelligence image recognition software. This data is then compared with a normal coronary artery control database, and a CHD risk assessment report is generated using receiver operating characteristic (ROC) curve analysis. The specific steps include: immunomagnetic beads specifically bind to monocytes, followed by sorting via the microfluidic chip and magnetic field; precise camera parameter setting and dual light source focus verification ensure image quality; and an AI algorithm uses feature extraction and pattern matching to achieve pixel-level localization of markers.

[0122] Those skilled in the art will appreciate that all or part of the processes in the above-described method embodiments can be implemented by instructing the relevant hardware using a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the above-described method embodiments. Any reference to memory, storage, database, or other media provided herein and used in the embodiments may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double-speed SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct RAMbus dynamic RAM (DRDRAM), and RAMbus dynamic RAM.

[0123] It should be noted that, in this document, the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, apparatus, article, or method comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, apparatus, article, or method. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, apparatus, article, or method comprising the element.

[0124] The above description is only a preferred embodiment of the present invention and does not limit the patent scope of the present invention. Any effective structure or effective process transformation made by using the contents of the present invention description and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for identifying peripheral blood mononuclear cell pattern data, characterized in that: The following steps are involved: Obtaining a peripheral blood sample and using a sorting system combining immunomagnetic beads and microfluidic chip technology to high-throughput sort and enrich monocytes from the peripheral blood sample; The images of the sorted monocytes were collected using a fully automatic cell morphology analyzer; Utilizes software embedded with self-developed artificial intelligence image recognition algorithms to detect and identify markers in collected monocyte images; The identified graphic data is compared and analyzed with a preset normal coronary artery control database to generate the identification results of the monocyte graphic data.

2. The peripheral blood mononuclear cell pattern data recognition method according to claim 1, characterized in that: The steps of high-throughput sorting and enriching monocytes from the peripheral blood sample using a sorting system combining immunomagnetic beads and microfluidic chip technology include: The immunomagnetic beads are mixed with the peripheral blood sample to allow the immunomagnetic beads to specifically bind to the monocyte surface markers; The mixed sample is introduced into the microfluidic chip, and the fluid control technology of the microfluidic chip is used to guide the sample to flow within the chip; By applying an external magnetic field to the immunomagnetic beads in the microfluidic chip, the immunomagnetic beads bound to monocytes are directionally deflected in the chip, thereby separating them from other cells. The fluid fraction containing monocytes after magnetic field sorting is collected to achieve high-throughput sorting and enrichment of monocytes from peripheral blood samples.

3. The peripheral blood mononuclear cell pattern data recognition method according to claim 1, characterized in that: The steps of acquiring images of sorted monocytes using a fully automated cell morphology analyzer include: The sorted and enriched mononuclear cell sample is introduced into the sample detection chamber of the fully automatic cell morphology analyzer; Set the camera's shooting parameters, including the time interval, number of horizontal and vertical shots, and distance interval. If the distance interval is less than 12 steps, the number of horizontal or vertical shots should not exceed 10. Set the stitching mode to 2×2 or 3×3. Confirm the white light focus and fluorescence focus to ensure that the lens is accurately focused on the mononuclear cell sample; Select single or continuous shooting mode to capture images of monocytes and obtain graphic data including cell morphology and marker fluorescence signals.

4. The method for identifying peripheral blood mononuclear cell pattern data according to claim 3, wherein: The steps of collecting images of monocytes to obtain graphic data including cell morphology and marker fluorescence signals include: When collecting images of sorted monocytes using a fully automated cell morphology analyzer, the camera properties were first set, including the time interval and the number of horizontal and vertical frames. When the distance interval was less than 12 steps, the number of horizontal or vertical frames was controlled to not exceed 10, and the stitching mode was set to 2×2 or 3×3. After completing the camera parameter settings, perform white light focus confirmation and fluorescence focus confirmation operations in sequence to ensure that the lens is accurately focused on the mononuclear cell sample and ensure the clarity of cell morphology and marker fluorescence signals during shooting; Select single-shot or continuous-shot mode according to the detection requirements to capture images of monocyte samples. During the shooting process, white light image data reflecting the cell morphological characteristics and fluorescence image data corresponding to the marker fluorescence signal are simultaneously obtained to form a graphic data set containing cell morphology and marker fluorescence signals.

5. The peripheral blood mononuclear cell pattern data recognition method according to claim 1, characterized in that: The steps for detecting and identifying markers in collected monocyte images using software embedded with a self-developed artificial intelligence image recognition algorithm include: Importing the collected monocyte image data into the artificial intelligence image recognition algorithm software, performing noise reduction and contrast enhancement preprocessing on the image to eliminate noise interference generated during the shooting process; Using the custom feature extraction model built into the algorithm software, cell morphology features and fluorescence signal features are extracted from the pre-processed image to generate a feature vector set; Based on the feature vector set, the markers in the monocyte image are identified and located using a pattern rule library preset in the algorithm software, wherein the rule library includes threshold values ​​of characteristic parameters of fluorescence signals for coronary heart disease-related markers; The confidence analysis is performed on the identified marker signal. When the signal strength and distribution characteristics meet the preset threshold, it is determined to be a valid marker detection result, and an identification report with the marker position coordinates and quantitative parameters is generated.

6. The method for identifying peripheral blood mononuclear cell pattern data according to claim 5, wherein: Based on the feature vector set, the steps of identifying and locating markers in the monocyte image using a pattern rule library preset in the algorithm software include: Inputting the feature vector set into a pattern recognition rule base preset in the algorithm software, wherein the rule base is constructed based on the fluorescence signal characteristic parameters of the NFAM1 protein marker in the peripheral blood mononuclear cells of patients with coronary heart disease, and includes quantitative standards such as the fluorescence intensity threshold and the signal distribution area threshold; Matching and analyzing the cell morphological features and fluorescence signal features in the feature vector set is performed using the rule base, and when the characteristic parameters of the fluorescence signal reach a preset threshold, marker recognition is triggered; Based on the matching results, the specific position coordinates of the marker in the monocyte image are determined, and the fluorescence signal area corresponding to the feature vector is converted into image pixel-level positioning results through a coordinate mapping algorithm to achieve accurate positioning of the marker.

7. The method for identifying peripheral blood mononuclear cell pattern data according to claim 1, wherein: The steps of comparing and analyzing the identified graphic data with a preset normal coronary artery control database to generate identification results of the monocyte graphic data include: Extracting the quantitative parameters of the identified markers to form a data set to be analyzed; Calling a preset normal coronary artery control database, which is constructed based on mononuclear cell image data of 79 normal coronary artery control samples and contains normal reference range thresholds for characteristic parameters of each marker; Compare the data set to be analyzed with the normal reference range threshold in the database. When the characteristic parameters of the marker exceed the normal threshold range, it is determined to be an abnormal signal; ROC curve analysis was used to verify abnormal signals and calculate the AUC value, sensitivity, and specificity statistical indicators. The indicators were used to set the discrimination criteria based on clinical validation data. Based on the comparison results and statistical analysis, a monocyte graphic data identification report containing a coronary heart disease risk assessment is generated, wherein the report associates the correspondence between the abnormality degree of the markers and the clinical classification of coronary heart disease.

8. The method for identifying peripheral blood mononuclear cell pattern data according to claim 7, wherein: The abnormal signals were verified using ROC curve analysis to calculate the AUC value, sensitivity, and specificity statistical indicators. The steps of setting the discrimination criteria based on clinical validation data included: Organizing marker characteristic parameters and clinical grouping data corresponding to abnormal signals, wherein the clinical grouping includes peripheral blood sample data of normal coronary artery control, stable coronary heart disease, and acute coronary syndrome; The ROC curve was constructed using the marker characteristic parameters as the test variables and the clinical grouping results as the state variables. The AUC value was calculated by the area under the curve, and the discrimination threshold was set based on the clinical validation data. The optimal critical value was determined based on the ROC curve, and the sensitivity and specificity indices were calculated and used as quantitative criteria for determining the correlation between abnormal signals and coronary heart disease; The calculated statistical indicators are compared with the preset standards in the clinical verification data. When the indicators reach or exceed the preset threshold, the abnormal signal is determined to have auxiliary diagnostic value for coronary heart disease.

9. A peripheral blood mononuclear cell pattern data recognition device, characterized in that: include: An acquisition unit, configured to obtain a peripheral blood sample and to perform high-throughput sorting and enrichment of monocytes from the peripheral blood sample using a sorting system combining immunomagnetic beads with microfluidic chip technology; An imaging unit, used for collecting images of the sorted monocytes using a fully automatic cell morphology analyzer; A detection unit, which is used to detect and identify markers in collected monocyte images using embedded independently developed artificial intelligence image recognition algorithm software; The analysis unit is used to compare and analyze the identified graphic data with a preset normal coronary artery control database to generate an identification result of the monocyte graphic data.

10. A computer device comprising a memory and a processor, wherein a computer program is stored in the memory, wherein: When the processor executes the computer program, the steps of the peripheral blood mononuclear cell pattern data recognition method according to any one of claims 1 to 8 are implemented.

Images