Automatic blood coagulation detection system

Abstract

The invention relates to the technical field of automatic medical detection, in particular to an automatic blood coagulation detection system which comprises a sample state regulation and control module, a light source balanced imaging module, a dynamic image extraction module, a reaction process quantification module and a risk trend identification module. According to the invention, through interval mapping of the sample concentration degree and the specific gravity value, dynamic parameters are screened in combination with temperature and pH feedback, precise regulation and control of sedimentation and component states are realized, the imaging stability is improved, a light source illumination angle and power are dynamically matched according to intensity distribution and boundary overlap ratio, and the brightness consistency is enhanced; in image acquisition, brightness and boundary offset are synchronously analyzed, a consistent region is extracted, image continuity and quantization precision are improved, a reaction trend is superposed based on a section sequence, time sequence judgment accuracy is enhanced, stable mapping is formed by combining track and sample feature classification, the anomaly recognition capability is improved, and a closed-loop detection process is constructed. And the detection stability and individualized adaptability are enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of automated medical testing, and particularly to an automated coagulation testing system. Background Art

[0002] The technical field of automated medical testing includes technologies related to the automated processing, analysis, and testing of medical samples. The core content of this technical field is to complete the efficient analysis and quantitative testing of biological samples such as blood and urine through the coordinated action of mechanical equipment, sensors, and control systems, thereby improving the testing efficiency and data consistency. An automated medical testing system involves multiple consecutive links such as sample collection, sample separation, reagent addition, reaction control, and result reading, and realizes process standardization through program control, and is widely used in laboratories, biomedical research, and public health monitoring.

[0003] Among them, an automated coagulation testing system refers to a system specifically used for the automated testing of blood coagulation function and the informatization management of the entire testing process. It mainly focuses on links such as sample information entry, reagent preparation, testing step control, data collection, result recording, and testing record traceability during the coagulation testing process. It uses barcode recognition technology to complete the automatic matching of samples and testing tasks, uses temperature control technology to regulate the reaction environment, obtains the time parameters of the coagulation reaction through optical detection methods, executes the testing steps in sequence according to the preset process control logic, and stores each testing parameter in a structured format through a data recording module after the testing is completed to achieve the systematic management of testing data.

[0004] Although the prior art has achieved standardized process operations, there are still significant limitations when facing high-precision and multi-variant samples. In the sample processing link, it is unable to effectively distinguish the changes in sample concentration and specific gravity, resulting in inconsistent states of samples in the pre-reaction processing stage and affecting the subsequent testing accuracy. The imaging light source configuration generally uses fixed irradiation parameters, making it difficult to cope with the dynamic changes in the boundary distribution and light intensity response of different samples during the reaction process, resulting in a decrease in the imaging signal-to-noise ratio. There is a lack of a judgment mechanism for the synchronous fluctuations of brightness and boundaries in the fluorescence image acquisition, so that abnormal frames or non-uniform regions interfere with the coherence of the detection results. The analysis of the reaction process mostly relies on static time-point data, lacking the ability to continuously track the overall trend of the reaction and making it difficult to reflect the true dynamic rhythm of the coagulation process. In terms of sample recognition, existing methods rely on rule templates for classification, ignoring the non-linear relationship between individual characteristics and detection trajectories, resulting in insufficient ability to identify differences in samples of the same batch. For example, some low-concentration samples are misjudged as abnormal states due to severe imaging boundary drift, indicating that the existing process has problems of response lag and insufficient accuracy in sample adaptation and reaction dynamic recognition. Summary of the Invention

[0005] The object of the present invention is to solve the disadvantages existing in the prior art, and to propose an automated blood coagulation detection system.

[0006] To achieve the above object, the present invention adopts the following technical solutions: An automated blood coagulation detection system includes: The sample state regulation module obtains the initial concentration value and specific gravity distribution value of the whole blood sample, performs interval mapping according to the data fed back by the temperature control unit, identifies the dynamic regulation parameters for controlling sedimentation and component equilibrium, and generates a sample state equilibrium parameter set; The light source equilibrium imaging module calls the sample state equilibrium parameter set, collects the intensity distribution of the light source component and the reaction cup boundary positioning, compares the coincidence degree of the imaging area distribution, selects a suitable irradiation angle and light source power configuration, and generates a uniform light projection configuration group; The dynamic image extraction module, based on the uniform light projection configuration group, collects the fluorescence imaging brightness change and the boundary position offset trajectory, screens the consistent area within the imaging cycle, and generates a continuous image feature trajectory set; The reaction process quantification module calls the continuous image feature trajectory set, extracts the fluorescence brightness interval in the image frame sequence, matches the reaction cycle change paragraph, judges the reaction process trend through section superposition, and obtains the blood coagulation process time sequence trajectory trend.

[0007] As a further solution of the present invention, the sample state equilibrium parameter set includes a concentration mapping coefficient, a specific gravity adjustment factor, a temperature and pH regulation weight, a sedimentation equilibrium threshold, and a component distribution balance value. The uniform light projection configuration group includes an irradiation angle parameter, a light source power configuration, a boundary matching coefficient, and an imaging area equilibrium degree. The continuous image feature trajectory set includes a brightness change curve, a boundary offset trajectory, a synchronous offset index, and a consistency area label. The blood coagulation process time sequence trajectory trend includes a fluorescence intensity section sequence, a stage evolution mode, a change trend path, and a time sequence matching result.

[0008] As a further solution of the present invention, the sample state regulation module includes: The sample state acquisition sub-module obtains the initial concentration value and specific gravity distribution value of the whole blood sample, extracts the parameter difference deviated from the specific gravity reference interval and the concentration standard interval according to the temperature value and pH value range fed back by the reaction cup loading cavity temperature control unit, and generates a sample initial offset parameter group; The adjustment parameter extraction sub-module, based on the sample initial offset parameter group, combines the adjustment limit values under the current temperature control conditions, screens the temperature adjustment factor and pH correction factor intervals that can be used to control the offset difference, extracts the adjustment combination that matches the current offset level, and generates a blood coagulation adjustment matching parameter; The detection process mapping sub-module, based on the coagulation regulation matching parameter, combines the current concentration value and the specific gravity distribution value of the sample, identifies the function mapping group for sedimentation and equilibrium regulation, determines the constraint boundary of the loading and reaction rhythm, and generates the sample state equilibrium parameter set.

[0009] As a further aspect of the present invention, the light source equilibrium imaging module includes: The detection response sub-module calls the sample state equilibrium parameter set, collects the light source power response value and the angle mapping value under multi-directional irradiation, compares the light intensity response with the irradiation angle matching degree, and obtains the light overlap rate index; The boundary positioning sub-module calls the light overlap rate index, extracts the image boundary gradient intensity, jointly screens the angle set that meets the boundary clarity requirement, and obtains the illumination angle selection set; The uniform light configuration sub-module combines the light source power groups according to the illumination angle selection set, identifies the light intensity and the boundary pixel density of each group, calculates the light balance error rate, selects the configuration with the smallest error rate, and generates the uniform light projection configuration group.

[0010] As a further aspect of the present invention, the dynamic image extraction module includes: The brightness trajectory extraction sub-module, based on the uniform light projection configuration group, collects the brightness value of the unit area of each frame of the image, records the time series and identifies the brightness difference between consecutive frames, and obtains the brightness change time trajectory; The boundary offset determination sub-module extracts the boundary coordinates according to the corresponding frame of the brightness change time trajectory, identifies the pixel position change vector of each frame, matches the brightness and the boundary difference, calculates the synchronous offset degree value, determines whether it is lower than the boundary change reference, and obtains the synchronous offset frame set; The feature trajectory construction sub-module calls the frame sequence in the synchronous offset frame set, extracts the brightness and boundary joint vector, sorts and constructs the trajectory, and determines the consistency according to the Euclidean distance between adjacent frames, and generates the continuous image feature trajectory set.

[0011] As a further aspect of the present invention, the reaction process quantification module includes: The brightness interval extraction sub-module calls the time-sequential frame segment in the continuous image feature trajectory set, extracts the fluorescence pixel brightness value of the reaction area according to the set time window, counts the maximum and minimum brightness of each segment, delimits the brightness change interval, and obtains the brightness distribution set of the reaction period; The section matching judgment sub-module, according to the brightness interval range of each section in the brightness distribution set of the reaction period, combines the change feature section set set in the reaction stage, identifies the fitting difference between the brightness amplitude deviation and the preset amplitude mean value, and calculates the reaction stage section offset trend value; The trend trajectory generation sub-module identifies sections with continuous same-type change directions according to the reaction stage section offset trend value, accumulates the trends of adjacent same-direction segments, classifies and labels the trend directions, and obtains the coagulation process time-series trajectory trend.

[0012] As a further solution of the present invention, the system further includes a risk trend identification module: Based on the coagulation process time-series trajectory trend, the risk trend identification module classifies the trajectory trends of samples in the same round, collects the sample feature categories recorded in the sample label information group, judges the stability degree of the associated interval between it and the trajectory classification, and generates an automated coagulation detection status mapping group; The automated coagulation detection status mapping group includes trajectory classification results, sample feature correlation degrees, classification stable intervals, and detection status labels.

[0013] As a further solution of the present invention, the risk trend identification module includes: Based on the coagulation process time-series trajectory trend, the trajectory interval classification sub-module identifies the trajectory trend vector according to the start and end points of the trajectory change and the amplitude difference, delimits the trajectory change interval according to the positions of adjacent inflection points in the time series, marks the corresponding interval numbers of the sample trajectories, and generates a trajectory interval marking table; The category interval association sub-module calls the trajectory interval marking table, collects the feature category items in the sample label information, analyzes the category frequency through the sample distribution quantity between the interval numbers and the feature categories, and screens out the association combinations with frequencies higher than the category interval matching threshold to obtain the category interval correspondence group; Based on the category interval correspondence group, the process status configuration sub-module sets the process label mapping rules between the trajectory intervals and the feature categories, configures the detection process status fields for the interval samples that meet the mapping rules, automatically marks the sample process management fields, and generates an automated coagulation detection status mapping group.

[0014] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In the present invention, through interval mapping of the concentration value and specific gravity distribution value of a whole blood sample, and combined with temperature and pH feedback of the reaction environment for dynamic parameter screening, pre-identification and precise regulation of the sample sedimentation and component equilibrium state are achieved, avoiding imaging deviation caused by sample state fluctuations. The irradiation angle and power configuration of the imaging light source no longer rely on a fixed template, but are matched based on the coincidence degree between the actual light source intensity distribution and the reaction cup boundary, realizing two-way optimization of irradiation uniformity and imaging area coverage rate, and improving the brightness consistency of fluorescence images. A synchronous comparison mechanism of boundary offset and brightness change trajectory is introduced during the acquisition of the image frame sequence to ensure accurate extraction of the consistent area within the imaging cycle, enhancing the continuity and quantifiability of the image feature trajectory. The reaction process analysis is based on the matching of brightness sections and preset change intervals, and a trend trajectory is constructed through section sequence superposition, enabling the judgment of the reaction process to shift from single-point data to overall trend recognition, and improving the ability to accurately depict the reaction timing. In the sample difference evaluation, by combining the trajectory trend and sample feature classification information, a corresponding relationship of a stable interval is formed, expanding the ability to identify potential abnormal states in samples of the same round, strengthening the mapping consistency between the automated detection state and the actual characteristics of the samples. With the cooperation of multi-dimensional participation items, a closed-loop detection chain from sample regulation to image extraction, then to reaction quantification and result identification is constructed, overall improving the stability, continuity and individual adaptation ability of coagulation detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 is the system flow chart of the present invention; Figure 2 is the flow chart of the sample state regulation module in the present invention; Figure 3 is the flow chart of the light source balanced imaging module in the present invention; Figure 4 is the flow chart of the dynamic image extraction module in the present invention; Figure 5 is the flow chart of the reaction process quantification module in the present invention; Figure 6 is the flow chart of the risk trend identification module in the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0016] In order to make the objectives, technical solutions and advantages of the present invention clearer and more understandable, 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 used to limit the present invention.

[0017] In the description of the present invention, it should be understood that the orientation or positional relationship indicated by the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc. is based on the orientation or positional relationship shown in the drawings. It is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation to the present invention. In addition, in the description of the present invention, the meaning of "a plurality of" is two or more, unless otherwise specifically defined.

[0018] Please refer to Figure 1 , an automated coagulation detection system includes: The sample state regulation module obtains the initial concentration value and specific gravity distribution value of the whole blood sample, screens the adjustment value according to the temperature and pH value range fed back by the temperature control unit of the reaction cup loading chamber, and performs interval mapping in combination with the concentration and specific gravity distribution value to identify the dynamic regulation parameters for controlling sedimentation and component balance, and generates a set of sample state balance parameters; The light source balanced imaging module calls the set of sample state balance parameters, collects the intensity distribution of the detection light source component and the reaction cup boundary positioning, compares the coincidence degree of their distributions in the imaging area, selects the corresponding irradiation angle and light source power group configuration, and generates a set of uniform light projection configurations; The dynamic image extraction module, based on the set of uniform light projection configurations, collects the brightness change time trajectory and the boundary position offset trajectory of the fluorescence imaging frame sequence, judges the synchronous offset amplitude between the trajectories, screens the consistent area within the imaging cycle, and generates a set of continuous image feature trajectories; The reaction process quantification module calls the set of continuous image feature trajectories, extracts the fluorescence brightness interval of each time period in the image frame sequence, and matches it with the set change section interval in the reaction cycle. By superimposing the section sequences, it judges the trend of the reaction process and obtains the time sequence trajectory trend of the coagulation process; The risk trend identification module, based on the time sequence trajectory trend of the coagulation process, classifies the trajectory trends of the samples in the same round, collects the sample feature categories recorded in the sample label information group, judges the stability degree of the association interval between it and the trajectory classification, and generates a set of automated coagulation detection state mappings.

[0019] The sample state equilibrium parameter set includes the concentration mapping coefficient, the specific gravity adjustment factor, the temperature and pH regulation weights, the sedimentation equilibrium threshold, and the component distribution balance value. The uniform light projection configuration group includes the irradiation angle parameter, the light source power configuration, the boundary matching coefficient, and the imaging area equilibrium degree. The continuous image feature trajectory set includes the brightness change curve, the boundary offset trajectory, the synchronous offset index, and the consistency area label. The blood coagulation process timing trajectory trend includes the fluorescence intensity section sequence, the stage evolution mode, the change trend path, and the timing matching result. The automated blood coagulation detection state mapping group includes the trajectory classification result, the sample feature correlation degree, the classification stable interval, and the detection state label.

[0020] Please refer to Figure 2 , the sample state regulation module includes: The sample state acquisition sub-module obtains the initial concentration value and specific gravity distribution value of the whole blood sample, extracts the parameter difference deviated from the specific gravity reference interval and the concentration standard interval according to the temperature value and pH value range fed back by the temperature control unit of the reaction cup loading cavity, and generates the sample initial offset parameter group; First, the collector extracts a certain amount of whole blood sample from the sample and uses precise measuring tools to determine the initial concentration value and specific gravity distribution value of the sample. The initial concentration value represents the solute concentration contained in the blood sample and is measured using a densitometer or an optical sensor, while the specific gravity distribution value is obtained by detecting the precipitation distribution of different solutes in the sample. These two data reflect the current physical characteristics of the sample and help to determine whether the sample meets the preset reaction standards. The temperature control unit monitors and feeds back the current ambient temperature and pH value range in the reaction cup loading cavity. The temperature value affects the activity and reaction rate of the reaction, while the pH value is related to the solubility of the solutes in the sample and the blood coagulation process. During this process, the fed-back temperature and pH data will be used to compare with the specific gravity reference interval and the concentration standard interval. If the specific gravity or concentration deviates from the preset standard interval, it is necessary to calculate the deviation value to determine whether the reaction conditions need to be adjusted. The deviated parameter difference can help to identify potential processing requirements during the sample processing. For example, when the concentration of the whole blood sample exceeds the upper limit of the standard interval, it is necessary to determine whether to adjust the temperature or pH value according to the deviation value in order to restore the sample to the appropriate reaction state. Through a series of measurements, comparisons, and calculations, the sample initial offset parameter group is generated to provide a basis for subsequent adjustment steps.

[0021] The adjustment parameter extraction sub-module, based on the sample initial offset parameter group and combined with the adjustment limit values under the current temperature control conditions, screens the temperature adjustment factor and pH correction factor intervals that can be used to control the offset difference, extracts the adjustment combination matching the current offset level, and generates the blood coagulation adjustment matching parameter; Evaluate each piece of data in the initial offset parameter group of the sample. The offset parameter group includes the concentration offset of the sample, the specific gravity offset, and the differences between the temperature and pH values fed back by the temperature control unit and the reference values. By calculating the offset values and comparing them with the preset ranges of the temperature adjustment factor and the pH correction factor, the ranges of the temperature adjustment factor and the pH correction factor are determined through historical data analysis and experiments. For example, assume the current temperature is 30°C while the standard temperature is 37°C. Then the temperature control unit needs to calculate the appropriate temperature adjustment range according to the preset limit value. Assume the adjustment range is ±3°C, the range of the pH correction factor is set from 6.5 to 7.5, and the current pH value is 6.8. Based on this data and combined with the offset of the sample, the adjustment factor will extract the optimal adjustment combination to ensure that the blood sample can be processed in the best reaction conditions in the subsequent process, and appropriate adjustment parameters can be screened out to make the sample better meet the experimental requirements, generating the coagulation regulation matching parameters for the sample processing in the next stage.

[0022] The detection process mapping sub-module, according to the coagulation regulation matching parameters, combines the current concentration value and specific gravity distribution value of the sample to identify the function mapping group for sedimentation and equilibrium regulation, determines the constraint boundaries of the loading and reaction rhythms, and generates the sample state equilibrium parameter set. Conduct an analysis of the sample state. According to the current concentration and specific gravity distribution values of the sample, identify the function mapping group for the sedimentation characteristics and equilibrium regulation of the sample. The sedimentation characteristics refer to the distribution of solutes or particles in the blood sample under the action of gravity, and the equilibrium regulation function is used to adjust the uniformity of the solute distribution during the reaction process. According to the characteristics, further judge whether the loading and reaction rhythms of the sample meet the specified boundary conditions. For example, if the concentration value exceeds a certain range, resulting in uneven solute distribution, it will affect the accuracy of the subsequent coagulation reaction. By analyzing the sample loading conditions, judge whether it is necessary to adjust the rhythm to ensure the consistency and stability of the sample processing during the experiment. Combining the current state of the sample and the regulation matching parameters, a sample state equilibrium parameter set will be generated to provide data support for further experimental steps and ensure that the reaction can proceed according to the set rhythm.

[0023] Please refer to Figure 3 , the light source equilibrium imaging module includes: The detection response sub-module calls the sample state equilibrium parameter set, collects the light source power response values and angle mapping values under multi-directional illumination, compares the light intensity response and the illumination angle matching degree, and obtains the illumination overlap rate index. According to the sample state equilibrium parameter set, including the sample light absorption characteristics, sample surface reflectivity and liquid surface disturbance degree, etc., the response coefficient of the reaction cup under different wavelengths of light is extracted through the preset equilibrium parameter table, and then combined with the angle configuration of each light source listed in the illumination matrix index, the illumination sensor is used to synchronously record the light source output power value and angle mapping value under each configuration. For example, the sample is placed in the detection position, and the light source module is used to irradiate at multiple incident angles such as 10°, 20°, 30°, etc. The corresponding power recorded for each irradiation is 30mW, 45mW, and 50mW, forming a configuration matrix, and then the collected power response value and angle mapping value are paired, and the edge change trend of the illumination area under different powers is numerically compared after the angle is fixed to form a light intensity change curve, and the angle response is compared with each other. The edge overlap statistics of the corresponding illuminated area contours are performed, and the overlap ratio is obtained by dividing the overlap area by the total illuminated area, and the overlap rate range is set. For example, the overlap rate less than 0.6 is a low overlap range, between 0.6 and 0.85 is a medium overlap range, and greater than 0.85 is a high overlap range. If the overlap area is 4cm² at a 30° illumination angle and the total area is 4.5cm², the overlap rate is 4 / 4.5≈0.89, which is classified as a high overlap range. At the same time, the data is judged by the difference with the set illumination reference value. The reference value is obtained through a large number of sample pre-experiments. The corresponding intensity value when the mean illumination intensity and the actual pixel recognition rate are maximized is set to 120lux. If the illuminance is less than this value, it will not be adopted. Finally, the configuration with a high overlap rate and a light intensity response higher than the benchmark is screened, and the output is the illumination overlap rate index.

[0024] The boundary positioning submodule calls the illumination overlap rate index, extracts the image boundary gradient intensity, jointly screens the angle set that meets the boundary clarity requirements, and obtains the illumination angle selection set; The pixel gradient of the cuvette edge image under each group of illumination was calculated, and the edge detection operator was used to extract the area with the most dramatic grayscale change in the image. The relative position and corresponding gradient amplitude of the area with the maximum pixel gradient were recorded. The changes of the boundary image under different illumination conditions were compared, and the pixel gradient map was superimposed frame by frame and normalized to obtain a pixel dense distribution map. The average gradient value of the pixel concentrated area at each angle was used as a reference for screening. For example, the average edge pixel gradient value was 35 at an angle of 20° and 42 at an angle of 30°. The average gradient value of each angle was divided into intervals. If the boundary clarity benchmark was set to 40, those greater than 40 were classified into the clear interval. Finally, the angle combination with high clarity and overlap rate index in the medium and high intervals was combined into an illumination angle selection set, which was recorded as the illumination angle selection set.

[0025] The light averaging submodule combines the light source power groups according to the lighting angle selection set, identifies the light intensity and boundary pixel density of each group, and uses the formula: ; Calculate the illumination balance error rate, select the configuration group with the smallest error rate, and generate the uniform light projection configuration group; Among them, represents the group light source power value, represents the group boundary pixel density, is the group target light intensity value, is the illumination balance error rate, represents the total number of light source configuration groups, represents the index of the light source configuration group; The illumination balance error rate refers to the numerical deviation degree between the actual illumination effect and the set target illumination intensity under the combination of multiple light source power configurations and the corresponding image boundary pixel densities. By calculating the difference between the product result of the light intensity and the pixel density under each group of configurations and the target illumination intensity, taking the absolute value, squaring and summing, and finally performing a square root operation, the overall error level of the overall lighting configuration in terms of illumination uniformity is reflected. The smaller this index, the closer the illumination uniformity of the current light source combination is to the set standard when meeting the imaging requirements, and it has higher imaging stability and regional illuminance consistency; According to several irradiation angle configurations selected from the illumination angle selection set, extract the corresponding light source power groups. Each group of power values is measured by an illuminance sensor with the light intensity per unit area (unit: mW / cm²). It is set that within the irradiation area of each group of light sources at a fixed distance from the sample surface (for example, 5 cm), the effective irradiation area is 1 cm². Record the illuminances of each group of light sources obtained as 42 mW / cm², 48 mW / cm², and 45 mW / cm², and set them as , , ; Extract the pixel density of the area in the reaction cup edge image, calculate the number of edge pixels per unit area (set as 0.5 cm²), and use the gray-scale change edge recognition method to scan the image contour area point by point to count the number of edge pixels. Set the number of edge pixels in the corresponding areas under 3 groups of configurations as 120 points, 135 points, and 128 points respectively, and perform normalization processing to match the power dimension. The normalization standard is the maximum pixel density of 255 points per unit. Calculate the density ratio coefficients as follows: ; ; ; Illumination target value Based on the pixel density - illuminance mapping relationship corresponding to the best light illumination balance during the imaging of the reaction cuvette, extract the point value of the maximum recognition efficiency of illuminance from a large number of sample experimental data, and set it as the target light intensity of 125 mW / cm² (unify the unit, consistent with the power), that is, for each group ; Calculate item by item: , the error term , the square is 11078.6; , the error term , the square is 9917.8; , the error term , the square is 10489.6; ; The final error rate is 177.4, corresponding to the overall error level in the three - group combination. Subsequently, select the single - group configuration with the lowest error to independently compare the value. For example, the group with the smallest error within a single group is the second group, and output the configuration of this group as the uniform light projection configuration group; : The light source power (mW / cm²) under the irradiation configuration of the th group; : The pixel density (unit normalized value, between 0 - 1) under the configuration of the th group; : The target light intensity (mW / cm²), and the value is obtained from the best - boundary imaging point of the experimental sample; By introducing the calculation of the product error of the normalized pixel density and the target light value, the numerical measurement and optimal group selection between the light intensity and the image recognition accuracy of the multi - angle irradiation configuration are realized, and the dimension of each participating value is unified during the operation process. The result has a direct guiding role. The result shows that the error of the current configuration group is 177.4. Compared with the set light error tolerance (such as setting the threshold to 200), it meets the balance requirement of the projection configuration, indicating that this group can be selected to generate the final uniform light projection configuration group.

[0026] Please refer to Figure 4 , the dynamic image extraction module includes: The brightness trajectory extraction sub - module, based on the uniform light projection configuration group, collects the brightness value of each unit area of each frame of the image, records the time series and identifies the brightness difference between consecutive frames to obtain the time trajectory of the brightness change; The preset area (such as the central area of ​​the reaction cup) in each frame image is divided into unit sampling blocks. The brightness grayscale value corresponding to each sampling block is obtained through the image sensor. The grayscale range is set to 0 to 255. The average grayscale value in a single frame is obtained as the brightness value of the current frame. The brightness time series is constructed according to the sampling time sequence, and the time interval is set to 0.2 seconds. The total number of sampling frames is 20 frames, forming a brightness sequence within 4 seconds. After constructing the sequence, the change amplitude of the brightness value between any two frames in the sequence is calculated, that is, the absolute value of the brightness value of the previous frame is subtracted from the brightness value of the latter frame to obtain the brightness difference between frames, and then the brightness change intensity distribution of the entire sequence is obtained. Then all brightness differences are counted to construct a brightness change vector. For example, the brightness of the third frame is 180, and the brightness of the fourth frame is 192, then the corresponding brightness difference is |192−180|=12. All frame pair differences are connected in series to form a brightness change time trajectory, and frames with brightness values ​​lower than the set sensor sensitivity threshold (such as grayscale lower than 20) are eliminated to ensure the stability of the change trajectory data source.

[0027] The boundary offset determination submodule extracts boundary coordinates according to the corresponding frame of the brightness change time trajectory, identifies the pixel position change vector of each frame, matches the brightness and boundary difference, and uses the formula: ; Calculate the synchronization offset value, determine whether it is lower than the boundary change benchmark, and obtain the synchronization offset frame set; in, is the synchronization offset value, is the total number of frames, For the Frame boundary pixel coordinates, For the Frame boundary pixel coordinates, For the The brightness value of the frame, For the The frame boundary recognition strength value; The synchronization offset value is a comprehensive indicator to measure the coordination of the position change and brightness change of each frame boundary in the fluorescence image sequence in the time dimension. This value is calculated by calculating the Euclidean distance of the boundary center coordinates between adjacent frames and the normalized difference of the brightness value and boundary recognition strength, and the spatial position change and grayscale response change are integrated and evaluated. The smaller the value, the higher the synchronization degree of the boundary position and brightness response change, indicating that the displacement of the imaging boundary in this frame sequence is stable and the brightness change law is consistent, which can be used to extract continuous stable frame segments in the image sequence. According to the frame index corresponding to the time trajectory of brightness change, pixel coordinate data of the reaction cup boundary position is extracted frame by frame from the original fluorescence image sequence. The process of obtaining the boundary position is to identify the region of abrupt change in image grayscale in edge detection. The grayscale gradient threshold is set to 20, and the coordinates of all pixel points where the first-order grayscale difference at the boundary is greater than the threshold are extracted. The coordinate of the boundary center position of this frame is obtained through the centroid calculation method. For example, the centroid of the first frame is (120, 145), the second frame is (123, 147), and the third frame is (125, 150); Meanwhile, the overall brightness value of the corresponding image region of each frame is collected. It is obtained by averaging the grayscale values of all pixels in the target area (such as the central 50×50 pixels). Suppose the first frame is 132, the second frame is 138, and the third frame is 140; Boundary recognition intensity Is the average absolute difference in grayscale change within the region (±2 pixels range) where the boundary pixels are located. This value is obtained by sampling the grayscale change of each boundary line, statistically taking the absolute value of the first-order grayscale difference between sample points and then averaging. The first frame is 135, the second frame is 136, and the third frame is 139; To achieve the dimensional consistency of the brightness value and the boundary intensity, both are normalized to the [0, 1] interval. The normalization reference range is 0–255 (grayscale value range), such as: 、 、 ; 、 、 ; Then calculate the synchronous offset value according to the formula: Assume that 3 frames of data are processed, that is , and calculate item by item as follows: The second frame: The coordinate change is ; The difference between the brightness and the boundary intensity is ; The total is: 3.6056 + 0.0079 = 3.6135; The third frame: The coordinate change is ; The difference is ; The total is: 3.6056 + 0.0039 = 3.6095; Take the average of the two results to get: ; The set boundary change reference value is 4.0 (the upper limit of the 95% confidence interval in the first 100 segments in the training sample analysis). Since 3.6115 is less than this threshold, it indicates a high degree of synchronization between the brightness and the boundary change in this sequence. Finally, a synchronized offset frame set is output, that is, the second frame and the third frame are synchronized and consistent frame segments. The fusion error amount is constructed by the cumulative average of the square root of the horizontal and vertical coordinate displacement amounts and the normalized brightness difference, taking into account the reflection of the position change and the gray information on the imaging change, providing a quantitative standard for subsequent image screening. The result shows that the current inter-frame change fluctuation is below the boundary reference, which can be used as the basic paragraph for continuous screening.

[0028] The feature trajectory construction sub-module calls the frame sequence in the synchronized offset frame set, extracts the joint vector of brightness and boundary, sorts and constructs the trajectory, and judges the consistency based on the Euclidean distance between adjacent frames to generate a continuous image feature trajectory set. The average image brightness of each frame and the coordinate parameters of the boundary pixel contour are extracted frame by frame. The average image brightness is calculated by statistically calculating the gray values in the unit area, and the boundary contour coordinates are marked by the position aggregation extraction method of the image edge pixel points. For example, in a certain frame of the image, the average brightness of the central detection area is 185, and the central point coordinates of the main line of the boundary pixels are (98, 205). Then the joint feature vector corresponding to this frame can be expressed as a triple of brightness and position. By processing all the image frames in the synchronized frame set in this way, according to the frame sequence order, the joint vector data set is constructed frame by frame. Subsequently, for all the constructed vector pairs, a continuity judgment operation is performed. The judgment method is whether the distance change between the joint vector values of two frames is stable. If the brightness difference between a certain frame and the previous frame is within 10, and at the same time the boundary coordinate offset is within 3 pixel points, then it is considered that there is continuity between these two frames. By scanning the entire frame set with a sliding window, the frame pairs with qualified continuity are merged and counted to generate continuous frame segments, and all continuous frame segments are integrated and classified. During the integration process, if a sudden change in brightness (such as the difference exceeding 30) or a drastic jump in the boundary (such as the boundary center offset between adjacent frames exceeding 8 pixels) occurs in a certain frame, the construction of the continuous segment is interrupted, and this frame is excluded and used as a breakpoint. In actual operation, the brightness change interval threshold is set to 10, and the boundary displacement threshold is set to 3 pixels as the continuous judgment standard. Once the continuous number of frames in a frame segment reaches more than 5 frames, it is regarded as a stable frame segment in the imaging stage, and this frame segment is incorporated into the continuous image structure. Finally, all the frame segments that meet the stable feature criteria are integrated to construct the feature trajectory set of each key frame segment in the time series. By uniformly characterizing the two-dimensional features of brightness and boundary coordinates, they are concatenated in the time dimension to form a complete frame trajectory chain, forming a key data set that can be used for subsequent time-series image analysis, and finally obtaining a continuous image feature trajectory set.

[0029] Please refer to Figure 5 , the reaction process quantification module includes: The brightness interval extraction sub-module calls the time-series frame segments in the continuous image feature trajectory set, extracts the fluorescence pixel brightness values in the reaction area according to the set time window, statistically calculates the maximum and minimum brightness of each segment, delimits the brightness change interval, and obtains the brightness distribution set of the reaction period; According to the fixed-time segmentation principle in the automated coagulation detection process, a reaction period is constructed for every 10 consecutive frames of images. The fluorescence gray values of all pixels in the 50×50 pixel rectangular area at the center of the reaction area are extracted within each period, the brightness value distribution of this area in each frame of the image is recorded, and the maximum and minimum brightness of this area within this period are calculated, which are respectively marked as the upper limit brightness and the lower limit brightness values. In this way, the brightness distribution interval of the reaction period is established. To avoid the deviation caused by the brightness range difference in different image acquisition batches to the subsequent operations, the brightness data is processed by normalization, which is: the original gray value divided by the gray upper limit 255, and converted into a numerical value in the range of [0, 1]. In the example, the maximum brightness from the 1st frame to the 10th frame is 210, and the minimum brightness is 125. Then the normalized interval is [0.4902, 0.8235]. This interval is recorded and given a period number, such as "section 01". Each time segment is processed in this way to construct a brightness interval sequence. The sequence structure is time label, brightness upper limit value, brightness lower limit value, and normalized interval value. The result is saved as a structured array for use in subsequent section fluctuation judgment, and finally the brightness distribution set of the reaction period is obtained.

[0030] The section matching and judgment sub-module, according to the brightness interval range of each section in the brightness distribution set of the reaction period, combines the change feature sections set in the reaction stage, identifies the fitting difference between the brightness amplitude deviation and the preset amplitude mean value, and uses the formula: ; Calculate the section offset trend value of the reaction stage; Among them, represents the section offset trend value of the reaction stage, is the total number of brightness sections, is the maximum brightness of the th section, is the minimum brightness of the th section, The reaction stage section offset trend value refers to the average difference obtained by comparing the amplitude of fluorescence brightness fluctuations extracted for each reaction time period with the brightness change range of the theoretical section in the preset reaction cycle in the automated coagulation detection process. It is used to quantify the deviation degree between the current reaction paragraph and the standard template. This value is calculated by taking the average of the differences between the actual brightness interval amplitudes of multiple sections and the template interval mean. The closer the value is to zero, the stronger the consistency of the brightness change trend in this reaction stage with the preset template. A positive value indicates that the actual brightness fluctuation is higher than expected, and a negative value indicates lower than expected. This trend value can be used as a key basis for subsequent determination of whether the reaction process is abnormal or sectional identification; According to the fluorescence brightness range information recorded for the concentration of brightness distribution in each reaction time period, combined with the reaction cycle change paragraph template set in the automated coagulation detection process, the maximum and minimum values of each brightness interval are extracted one by one, the brightness fluctuation amplitude within the current section is calculated, and an offset amount judgment is made with the mean value of the theoretical fluctuation amplitude in the corresponding stage of the template. The brightness value of each section is statistically analyzed using the gray scale mean within a 50×50 pixel central area, and the gray scale range is from 0 to 255; To eliminate the acquisition differences between devices, the maximum brightness value of each section and the minimum brightness value need to be normalized. The normalization formula is , so as to unify all participating items to the range of 0 to 1. For example, for the first section with a maximum value of 210 and a minimum value of 125, after normalization: , , and the amplitude within the section is ; At the same time, the upper and lower limits of the reference brightness fluctuation range defined in the system template for this section are 190 to 120. After normalization: , , and its mean value is ; Perform the offset trend operation according to the formula: Suppose there are currently 3 sections of data, and their parameters are respectively: The first section: , , , ; The second section: maximum 192, minimum 135 → , , and the template is [180, 130] → , ; The third section: maximum 185, minimum 132 → , , and the template is [170, 120] → , ; The operation process is as follows: The first paragraph: , average value = ; The second paragraph: ; The third paragraph: ; Substitute all results into the formula: ; According to the judgment criteria set by the system, when the absolute value is less than 0.50, it is judged that the offset is stable. This value is the upper limit within the 95% confidence interval derived from statistical derivation of empirical data; therefore, the calculation result of this paragraph, -0.3399, indicates that the brightness section fluctuation conforms to the expectation of the reference paragraph, and it is judged that the offset trend is consistent; By directly comparing the actual brightness fluctuation amplitude of the current reaction period with the average value of the preset theoretical paragraph interval, a standard quantitative judgment mechanism for the dynamic trajectory alignment template is established, avoiding the traditional fuzzy processing mode based on boundary judgment. This result shows that the brightness fluctuation in this period is within the expected change rhythm, and the result is classified as a normal component of the offset trend in the reaction stage section.

[0031] The trend trajectory generation sub-module identifies the sections with continuous similar change directions according to the offset trend value of the reaction stage section, accumulates the trends of adjacent same-direction segments, classifies and marks the trend directions, and obtains the coagulation process time sequence trajectory trend; Mark the section trend values as positive or negative to indicate the brightness change direction of each section. If multiple consecutive sections have the same-direction offset trend, they are regarded as a type of trend area, record the start and end section index numbers and the number of continuous sections, classify and mark this change trend area. For example, the offset values from the first section to the third section are negative, marked as "descending section A", and the offset values from the fourth section to the fifth section are positive, marked as "ascending section B". Construct a trend sequence structure body, which includes: section identifier, trend direction, continuous length, change interval value. Arrange all trend sequences in chronological order to form the basic data for characterizing the trend evolution of each reaction stage in the automated coagulation detection process, and finally generate a coagulation reaction trend sequence table.

[0032] Please refer to Figure 6 , the risk trend identification module includes: The trajectory interval classification sub-module, based on the coagulation process time sequence trajectory trend, identifies the trajectory trend vector according to the start and end points and the amplitude difference of the trajectory change, delimits the trajectory change interval according to the positions of adjacent inflection points in the time series, marks the corresponding interval serial numbers of the sample trajectories, and generates a trajectory interval marking table; The determination of the start and end points of a trajectory depends on a comprehensive analysis of time series data. In this process, the time series data of the samples is extracted, which contains the relationship between the measured metric values and time. The difference method is used to calculate the amplitude. When the change of the trajectory in the time series exceeds the set amplitude threshold, it is determined as the starting point of the trajectory change. Taking the subsequent data as the benchmark, the amplitude changes in adjacent time periods are continuously compared until the amplitude difference returns to the initial level or is lower than the set threshold, and then it is determined as the end point of the trajectory. For example, when the amplitude difference is greater than 0.5, it is judged as the starting point, and when the amplitude difference returns to below 0.1, it is judged as the end point. According to the starting point and the end point, multiple intervals of the trajectory change are divided according to the time sequence. For each interval, by calculating the time difference between adjacent inflection point positions, the division is further refined to ensure that the interval delimitation conforms to the actual situation. Using the delimited interval information, a unique serial number is assigned to each interval, and this serial number is marked in the trajectory sample. Finally, a trajectory interval marking table is generated, and each sample data will correspond to an interval serial number. For example, for the sample trajectory points (time T1, value V1), (time T2, value V2), if the amplitude change between T1 and T2 exceeds the set threshold, the interval number is defined as 1. When the amplitude change ends, the start point of interval 2 is delimited at time T3.

[0033] The category interval association sub-module calls the trajectory interval marking table, collects the characteristic category items in the sample label information, analyzes the category frequency through the sample distribution quantity between the interval serial number and the characteristic category, and filters out the association combinations with a frequency higher than the category interval matching threshold to obtain the category interval correspondence group; Call the trajectory interval marking table to obtain the respective interval numbers of the sample data and the corresponding sample label information, collect the characteristic category items in each sample data, where the category items represent different coagulation characteristics, such as coagulation time, thrombin generation rate, etc. Analyze the distribution of the characteristic category items within each interval number, count the occurrence frequency of each characteristic category in different intervals, use a frequency analysis tool to calculate the category frequency, and compare it with the set category interval matching threshold to filter out the category interval combinations with a frequency exceeding the threshold. Frequency analysis includes counting the sample data of each category. When the number of occurrences of a certain category within an interval reaches the set threshold, it is considered that the category forms an effective association with this interval. For example, if the characteristic category A appears 30 times in interval 1 and 5 times in interval 2, and the set frequency threshold is 10 times, then the association between interval 1 and category A is strong, and the association between interval 2 and category A is weak. Finally, the category interval correspondence group obtained will include the strong association between interval 1 and category A.

[0034] The process status configuration sub-module sets the mapping rules for the process tags of the trajectory interval and the feature category according to the category interval correspondence group, configures the detection process status field for the interval samples that meet the mapping rules, automatically marks the sample process management field, and generates an automated coagulation detection status mapping group; Set the mapping rules for the trajectory interval and the feature category. The setting of the mapping rules depends on the actual detection standard. For example, it is set that when the frequency of a certain feature category A in a certain interval exceeds a certain threshold, the process status of the samples in that interval is "normal", otherwise it is "abnormal". The setting of the mapping rules depends on historical data or empirical rules, and a reasonable mapping range is obtained through experiments. For each interval sample, apply this mapping rule to detect its status and automatically configure the status field. If an interval meets the mapping rule, it is automatically marked as "normal" or "abnormal" status. For example, if the frequency of category A in interval 1 exceeds the threshold 10, the corresponding process status of this interval is "normal", and if the frequency of category B is lower than the set threshold, it is marked as "abnormal". Through an automated algorithm, the status information of all intervals is summarized to generate an automated coagulation detection status mapping group. This mapping group will be used for subsequent detection and processing, and the automatically marked process status field will also reflect the management situation of the samples; Table 1: Example of frequency analysis of category interval association:

[0035] As shown in Table 1, the table lists the occurrence frequencies of different categories in certain intervals. For example, in interval 1, category A appears 30 times, category B appears 5 times, category C appears 10 times, and the frequency of category A is higher than the set matching threshold of 10 times. Therefore, the association between interval 1 and category A is stronger, while the frequency of category A in interval 2 is lower than the threshold, so the association is weaker.

[0036] The above is only a preferred embodiment of the present invention, and it is not intended to limit the present invention in other forms. Any person skilled in the art may use the disclosed technical content to make changes or modifications into equivalent embodiments with equivalent changes and apply them to other fields. However, as long as it does not depart from the technical solution content of the present invention, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention still fall within the protection scope of the technical solution of the present invention.

Claims

1. An automated coagulation detection system, characterized in that: The system comprises: The sample state control module obtains the initial concentration value and specific gravity distribution value of the whole blood sample, performs interval mapping based on the data fed back by the temperature control unit, identifies the dynamic control parameters for controlling sedimentation and component balance, and generates a sample state balance parameter set; The light source balanced imaging module calls the sample state balanced parameter set, collects the intensity distribution of the light source component and the boundary location of the reaction cup, compares the distribution overlap of the imaging area, selects the appropriate irradiation angle and light source power configuration, and generates a uniform light projection configuration group; The dynamic image extraction module collects the fluorescence imaging brightness change and boundary position offset trajectory based on the uniform light projection configuration group, screens the consistent area within the imaging period, and generates a continuous image feature trajectory set; The reaction process quantification module calls the continuous image feature trajectory set, extracts the fluorescence brightness interval in the image frame sequence, matches the reaction cycle change segment, determines the reaction process trend by segment superposition, and obtains the coagulation process time series trajectory trend.

2. The automated coagulation detection system according to claim 1, characterized in that: The sample state equilibrium parameter set includes concentration mapping coefficient, specific gravity adjustment factor, temperature and pH control weight, sedimentation equilibrium threshold, and component distribution balance value; the uniform light projection configuration group includes illumination angle parameter, light source power configuration, boundary matching coefficient, and imaging area balance; the continuous image feature trajectory set includes brightness change curve, boundary offset trajectory, synchronization offset index, and consistency area label; the coagulation process timing trajectory trend includes fluorescence intensity segment sequence, stage evolution pattern, change trend path, and timing matching result.

3. The automated coagulation detection system according to claim 1, characterized in that: The sample state control module includes: The sample status acquisition submodule obtains the initial concentration value and specific gravity distribution value of the whole blood sample, extracts the parameter difference that deviates from the specific gravity reference interval and the concentration standard interval according to the temperature value and pH value range fed back by the reaction cup loading chamber temperature control unit, and generates the sample initial offset parameter group; The adjustment parameter extraction submodule screens the temperature adjustment factor and pH correction factor interval that can be used to control the offset difference based on the sample initial offset parameter group and the adjustment limit under the current temperature control condition, extracts the adjustment combination that matches the current offset level, and generates the coagulation adjustment matching parameter; The detection process mapping submodule identifies the function mapping group of sedimentation and equilibrium regulation according to the coagulation regulation matching parameters, combines the current concentration value and specific gravity distribution value of the sample, determines the constraint boundaries of loading and reaction rhythm, and generates a sample state equilibrium parameter set.

4. The automated coagulation detection system according to claim 3, characterized in that: The light source balanced imaging module comprises: The detection response submodule calls the sample state balance parameter set, collects the light source power response value and angle mapping value under multi-directional illumination, compares the light intensity response and the illumination angle matching degree, and obtains the illumination overlap rate index; The boundary positioning submodule calls the illumination overlap rate index, extracts the image boundary gradient intensity, jointly screens the angle set that meets the boundary clarity requirement, and obtains the illumination angle selection set; The light balancing configuration submodule combines light source power groups according to the lighting angle selection set, identifies each group of light intensity and boundary pixel density, calculates the illumination balance error rate, selects the group configuration with the minimum error rate, and generates a light balancing configuration group.

5. The automated coagulation detection system according to claim 4, characterized in that: The dynamic image extraction module comprises: The brightness trajectory extraction submodule collects the brightness value of the unit area of ​​each frame image based on the average light projection configuration group, records the time series and identifies the brightness difference between consecutive frames, and obtains the brightness change time trajectory; The boundary offset determination submodule extracts boundary coordinates according to the corresponding frame of the brightness change time trajectory, identifies the pixel position change vector of each frame, matches the brightness and boundary difference, calculates the synchronization offset value, determines whether it is lower than the boundary change benchmark, and obtains the synchronization offset frame set; The feature trajectory construction submodule calls the frame sequence in the synchronous offset frame set, extracts the brightness and boundary joint vector, sorts and constructs the trajectory, judges the consistency according to the Euclidean distance of adjacent frames, and generates a continuous image feature trajectory set.

6. The automated coagulation detection system according to claim 5, characterized in that: The reaction progress quantification module comprises: The brightness interval extraction submodule calls the time sequence frame segments in the continuous image feature trajectory set, extracts the brightness values ​​of the fluorescent pixels in the reaction area according to the set time window, counts the maximum and minimum brightness of each segment, delineates the brightness change interval, and obtains the brightness distribution set of the reaction period; The segment matching judgment submodule identifies the fitting difference between the brightness amplitude deviation and the preset amplitude mean according to each brightness interval range of the brightness distribution concentration in the reaction period and the change characteristic segment set in the reaction stage, and calculates the segment offset trend value in the reaction stage; The trend trajectory generation submodule identifies segments with continuous similar change directions according to the reaction stage segment offset trend value, accumulates the trends of adjacent segments with the same direction, classifies and identifies the trend directions, and obtains the coagulation process time series trajectory trend.

7. The automated coagulation detection system according to claim 1, characterized in that: The system also includes a risk trend identification module: The risk trend identification module classifies the trajectory trends of samples in the same round based on the temporal trajectory trend of the coagulation process, collects the sample feature categories recorded in the sample label information group, determines the stability of the associated intervals between the categories and the trajectory classification, and generates an automated coagulation detection state mapping group; The automated coagulation detection state mapping group includes trajectory classification results, sample feature correlation, classification stability interval, and detection state label.

8. The automated coagulation detection system according to claim 7, characterized in that: The risk trend identification module includes: The trajectory interval classification submodule is based on the coagulation process time series trajectory trend, identifies the trajectory trend vector according to the trajectory change start and end points and the amplitude difference, delineates the trajectory change interval according to the position of adjacent inflection points in the time series, marks the sample trajectory corresponding to the interval number, and generates a trajectory interval marking table; The category interval association submodule calls the trajectory interval marking table, collects the feature category items in the sample label information, analyzes the category frequency by the sample distribution quantity between the interval sequence number and the feature category, and selects the association combination with a frequency higher than the category interval matching threshold to obtain the category interval correspondence relationship group; The process status configuration submodule sets the process label mapping rules between the trajectory interval and the feature category according to the category interval correspondence group, configures the detection process status field for the interval samples that meet the mapping rules, automatically marks the sample process management field, and generates an automated coagulation detection state mapping group.

Images