Management method and system of leukemia cell morphology intelligent identification microscopic device

Abstract

The invention relates to a management method and system for a leukemia cell morphology intelligent recognition microscopic device, and solves the problems that imaging quality fluctuates, suspicious cell recognition precision is limited, and clinical efficient standardization requirements are difficult to meet. Extracting sample features and calling a parameter library according to rules to select optimal scanning and imaging parameters for initialization; starting an X / Y-axis motor according to the parameters, and correcting the deviation by combining an anti-interference rule and an encoder; z-axis dual-stage focusing is carried out after stable stopping, and an FPGA time sequence control camera carries out acquisition; aI judgment results are sent to images, suspicious cells are delineated and then positioned, and high-power lens re-focusing collection and AI recognition are carried out. The method has the advantages that sample adaptation and positioning precision is improved, suspicious cell recognition is optimized, and detection efficiency and precision are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of medical equipment, in particular to a leukemia cell morphology intelligent identification microscopic device management method and system. BACKGROUND

[0002] Leukemia cell morphology detection is a key basis for clinical diagnosis of leukemia and typing, and the condition needs to be judged by observing the morphology of cell nucleus and cytoplasm and staining characteristics. The traditional detection relies on manual microscopy, and the tester needs to observe the smear sample field by field, which not only takes a long time (usually more than 30 minutes) for single sample detection, but also is easily affected by experience difference and visual fatigue, resulting in strong subjectivity and low consistency of the diagnosis result, which is difficult to meet the clinical batch detection and standardization requirements.

[0003] The current operation process of leukemia cell morphology detection mainly includes the following steps: first, fix the stained leukemia cell smear on the microscope stage, and the operator selects the fixed scanning and imaging parameter template according to experience; then start the stage, move along the fixed track by motor and encoder closed loop control, and only rely on hardware feedback for basic positioning correction; after the optical system is automatically focused, the camera is triggered to collect images, and the images are transmitted to the AI system for initial labeling of suspicious cells; finally, for the suspicious area, the operator needs to switch the high-power objective lens and manually focus to observe and confirm.

[0004] The core disadvantage of the existing operation is that the whole process is not dynamically optimized around the sample characteristics: the parameter setting relies on the fixed template and is difficult to adapt to different samples with different staining and thickness, the stage positioning accuracy and stability are insufficient, and the key link connection relies on manual operation, which finally leads to the fluctuation of imaging quality and the limitation of suspicious cell recognition accuracy, and it is difficult to meet the clinical efficient standardization requirements. SUMMARY

[0005] In order to improve sample adaptation and positioning accuracy, optimize suspicious cell recognition, improve detection efficiency and accuracy, the present application provides a leukemia cell morphology intelligent identification microscopic device management method and system.

[0006] In the first aspect, the present application provides a leukemia cell morphology intelligent identification microscopic device management method, which adopts the following technical scheme:

[0007] A leukemia cell morphology intelligent identification microscopic device management method, comprising:

[0008] Fixing the stained leukemia cell sample on the stage;

[0009] Based on the fixed leukemia cell sample, the sample features are extracted, and the optimal scanning and imaging parameters are selected by calling the preset parameter library according to the preset sample feature matching rules to complete the device initialization. The intelligent leukemia cell morphology recognition microscopy device is equipped with stage X / Y axis and Z axis drive motors.

[0010] The X / Y axis drive motors of the stage are started according to the matched scanning parameters, which drive the sample to move along the preset trajectory. The deviation is corrected according to the preset anti-interference rules and the encoder position feedback to ensure that the positioning accuracy meets the preset requirements.

[0011] After the stage comes to a stop, the Z-axis drive motor is triggered to find a clear imaging plane by following the preset dual-stage focusing algorithm and the preset contrast judgment rules, and the camera is triggered to acquire data simultaneously by the FPGA timing control algorithm.

[0012] The acquired images are transmitted to the AI ​​recognition module, which outputs the judgment result based on the preset confidence threshold.

[0013] If the cell is not suspicious, store the identification result and associate the sample features with the database;

[0014] If a cell is suspected, a preset region growth algorithm is activated to delineate the suspected area, the X / Y axis motors are driven to locate the area, a high-magnification objective lens is switched, and the Z-axis is refocused to acquire a high-resolution image, which is then transmitted to the AI ​​module for identification.

[0015] By adopting the above technical solutions, the optimal parameters are selected through sample feature matching to adapt to samples with different staining and thicknesses; anti-interference rules combined with encoders ensure the positioning accuracy of the stage; dual-stage focusing and FPGA-based management of the intelligent leukemia cell morphology identification microscope device ensure clear imaging; automatic delineation and refocusing of suspicious cells and AI-based management of the intelligent leukemia cell morphology identification microscope device significantly improve the accuracy and efficiency of detection, and contribute to standardized clinical testing.

[0016] Secondly, this application provides a management system for an intelligent microscopic device for recognizing leukemia cell morphology, employing the following technical solution:

[0017] A management system for a leukemia cell morphology intelligent identification microscopic device includes a memory, a processor, and a program stored in the memory and executable on the processor. When the program is loaded and executed by the processor, it implements the management method of the leukemia cell morphology intelligent identification microscopic device as described in the first aspect. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating a management method for a leukemia cell morphology intelligent identification microscopic device according to an embodiment of this application.

[0019] Figure 2 This is a schematic diagram of another embodiment of the present application, which is based on a fixed leukemia cell sample, extracts sample features, calls a preset parameter library to screen the optimal scanning parameters and imaging parameters according to preset sample feature matching rules, and completes the device initialization process. Detailed Implementation

[0020] The present application will be further described in detail below with reference to the accompanying drawings.

[0021] Reference Figure 1 The present application discloses a management method for a leukemia cell morphology intelligent recognition microscopic device, comprising:

[0022] Step S100: Fix the stained leukemia cell sample onto the stage.

[0023] Leukemia cell samples: Samples containing leukemia cells extracted from a patient's blood or bone marrow, used for subsequent morphological analysis. Staining: Cells are stained with specific chemical reagents (such as Wright's stain) to make different parts of the cell (such as the nucleus and cytoplasm) appear in different colors, facilitating observation under a microscope. Stage: The platform on the microscope used to place the sample, usually equipped with a precise movement device (such as an X / Y axis drive motor) for adjusting the sample's position.

[0024] The overall process is as follows:

[0025] First, the automated system extracts pre-labeled leukemia cell samples from the sample library and transfers them to the staining module. In the staining module, the samples are automatically mixed with Wright's stain and soaked for 3-5 minutes according to a preset program. Excess stain is then removed with distilled water by an automatic rinsing device. The stained samples are then automatically added to a glass slide and fixed with ethanol using a fixation device for 1-2 minutes. After fixation, the slide is automatically transferred to the microscope stage and secured with a magnetic device to ensure sample stability during subsequent operations.

[0026] Subsequently, the X / Y axis drive motors of the stage automatically adjust the position of the slide according to a preset program, aligning the observation area of ​​the sample with the center of the microscope objective. The system automatically performs preliminary observation through the low-power objective and adjusts the stage position based on the built-in image recognition algorithm to ensure that the observation area of ​​the sample is completely within the field of view. During the adjustment process, the system uses a coordinate positioning algorithm to automatically calculate and input the target coordinates (e.g., X = 5mm, Y = 3mm), and the drive motors precisely move to the designated position, ensuring that the observation area of ​​the sample is accurately aligned with the center of the objective.

[0027] Following step S100, the following steps are performed:

[0028] Step S101: After the sample is fixed on the stage, the array of a preset number of laser displacement sensors distributed along the X / Y axis of the stage is activated to collect surface height data according to the preset sampling point rules (covering the sample edge and center area).

[0029] Step S102: Call the preset data fitting algorithm to fit the collected height data into the three-dimensional contour of the sample surface, and calculate the difference between the maximum height value and the minimum height value within the contour (denoted as ΔH).

[0030] Step S103: Extract the current sample type (regular / thick smear), retrieve the preset flatness threshold for the corresponding type, and compare ΔH with the preset flatness threshold: if ΔH≤ the corresponding threshold, the sample fit is determined to be up to standard, output the "flatness qualified" signal, and enter the device initialization stage.

[0031] Step S104: If ΔH > the corresponding preset flatness threshold, locate the convex / concave area exceeding the threshold based on the three-dimensional contour, and trigger the preset multi-axis collaborative fine-tuning algorithm: control the X / Y axis motors to fine-tune the tilt angle of the stage according to the preset tilt adjustment range (within ±3°) and preset tilt step size, and simultaneously fine-tune the height of the stage according to the preset height adjustment accuracy through the Z axis motor. After each adjustment, re-collect the height data and calculate the new ΔH.

[0032] Step S105: Repeat steps S102 to S104 until the new ΔH is less than or equal to the corresponding preset flatness threshold, or the number of fine-tuning times reaches the preset upper limit of fine-tuning times; if the upper limit is reached but the standard is still not met, output a "sample placement abnormal" alarm and pause the process, prompting manual inspection.

[0033] Step S106: After the flatness meets the standard, the final ΔH, preset parameters (tilt angle, height compensation value) are associated and stored with the sample ID to provide a reference for subsequent scanning parameter correction.

[0034] Step S200: Based on the fixed leukemia cell sample, extract sample features, and according to the preset sample feature matching rules, call the preset parameter library to screen the optimal scanning parameters and imaging parameters to complete the device initialization. The intelligent leukemia cell morphology recognition microscopy device is equipped with stage X / Y axis and Z axis drive motors.

[0035] Among them, the preset sample feature matching rule is a rule set in advance based on the relationship between sample features and microscope parameters, which is used to select appropriate scanning and imaging parameters.

[0036] Preset parameter library: A database that stores a variety of possible combinations of scanning and imaging parameters, used to select the optimal parameters based on sample characteristics.

[0037] Stage X / Y and Z axis drive motors: These motors precisely control the movement of the stage in the horizontal and vertical directions, ensuring accurate focusing and positioning of the sample. The X / Y axes are responsible for the movement and positioning of the sample in the horizontal plane, enabling scanning of different fields of view; the Z axis is responsible for movement along the optical axis, enabling focus adjustment and ensuring clear imaging.

[0038] The necessary procedures can be referred to in steps S210 to S260, which will not be elaborated here.

[0039] In step S300, the X / Y axis drive motors of the stage are started according to the matched scanning parameters, which drive the sample to move along the preset trajectory. The deviation is corrected according to the preset anti-interference rules and the encoder position feedback to ensure that the positioning accuracy meets the preset requirements.

[0040] The system includes: Preset trajectory: A pre-defined movement path based on the scanning requirements of the sample, typically a grid-like path used to cover all or part of the sample area. Encoder position feedback: Position information provided by the motor encoder, used to monitor the actual position of the stage in real time, ensuring accurate movement. Anti-interference rules: Rules or algorithms used to reduce the impact of external interference (such as vibration, temperature changes, etc.) on the stage's movement accuracy.

[0041] The necessary process is described as follows: First, based on the optimal scanning parameters determined during the initialization phase, the system starts the X / Y axis drive motors of the stage, moving the sample along a preset trajectory. During the movement, the system receives position feedback signals from the encoder in real time, monitoring the actual position of the stage. When a position deviation is detected, the system corrects it according to preset anti-disturbance rules. For example, using a PID control algorithm, the system adjusts the motor speed and direction in real time based on the position deviation feedback from the encoder, ensuring that the stage moves accurately along the preset trajectory. If the deviation exceeds a preset threshold, the system automatically pauses the movement, recalibrates the position, and then continues moving. Through this closed-loop control method, the system can effectively reduce the impact of external interference on the movement accuracy, ensuring that the positioning accuracy of the stage meets the preset requirements, and providing accurate sample position information for subsequent imaging operations.

[0042] In step S400, after the stage stops, the Z-axis drive motor is triggered to find a clear imaging plane by following the preset dual-stage focusing algorithm and the preset contrast judgment rule, and the camera is triggered to acquire data simultaneously by the FPGA timing control algorithm.

[0043] In this process, the stage, managed by the X / Y axis of the intelligent leukemia cell morphology recognition microscope, must come to a stable stop after moving to the preset sampling position. This is a prerequisite for focusing and image acquisition. The process is as follows: the stage decelerates as it moves along the trajectory towards the target position, and the motor stops driving upon arrival, allowing for a short period of mechanical damping to attenuate the vibration. Position data is collected in real time by the encoder. When the position deviation is less than a preset threshold (e.g., 1 μm) for multiple consecutive sampling cycles (e.g., within 20 ms), the stage is considered to have come to a stable stop, ensuring no displacement interference during focusing and image acquisition.

[0044] Z-axis drive motor: A motor used to control the movement of the stage in the vertical direction (Z-axis), ensuring precise vertical focusing of the sample. Dual-stage focusing algorithm: A two-step focusing algorithm that first quickly locates the approximate focus area, then fine-tunes within that area to find the optimal focus plane. Preset contrast judgment rule: A rule used to determine whether an image has achieved sharp focus, typically based on image contrast or edge strength. FPGA timing control algorithm: An algorithm for precisely controlling the camera's acquisition signals, ensuring the camera acquires images at the optimal time.

[0045] The overall necessary process can be described by referring to steps S410 to S450, which will not be repeated here.

[0046] In step S500, the acquired image is transmitted to the AI ​​recognition module, and the judgment result is output according to the preset confidence threshold.

[0047] The system includes: an AI recognition module (based on artificial intelligence technologies such as deep learning) for analyzing images acquired under a microscope and identifying cell morphology; a pre-set confidence threshold (a threshold used to judge the reliability of the recognition results; only when the confidence level of the recognition result is higher than this threshold is the recognition result considered reliable); and a database (a system for storing sample characteristics, recognition results, and other information for subsequent data analysis and traceability).

[0048] The necessary process is described as follows: After image acquisition, the system transmits the acquired images to the AI ​​recognition module. The AI ​​recognition module uses a pre-trained deep learning model (such as a convolutional neural network, CNN) to analyze the images and identify cell morphology. The model outputs a confidence score for each recognition result, indicating the reliability of the result. The system compares these confidence scores with a preset confidence threshold. If the confidence score is higher than the threshold, the recognition result is considered reliable, and the system stores the result in the database and associates it with sample features. If the confidence score is lower than the threshold, the system marks the image as "unreliable" or "requires further analysis" for subsequent manual review or other processing. In this way, the system can efficiently screen out potentially problematic cells, supporting subsequent diagnosis and research.

[0049] In step S600, if the cell is not suspicious, the identification result is stored and the sample features are associated with the database.

[0050] Among them, "unsuspected cells" are those identified as normal and without leukemia characteristics after analysis by the AI ​​recognition module. The database is a system used to store information such as sample characteristics and recognition results, facilitating subsequent data analysis and traceability. Sample characteristics refer to the physical and chemical properties of leukemia cell samples, such as staining depth, cell density, and smear thickness; these characteristics are recorded during sample fixation and scanning.

[0051] The necessary process is described as follows: After the AI ​​recognition module completes image analysis, for cells deemed not suspicious, the system stores the recognition results in the database. During storage, the system associates the recognition results with the sample's feature information, ensuring that each recognition result can be traced back to a specific sample. For example, the system records the recognition result of each not suspicious cell (such as cell type, morphological characteristics, etc.) and binds it to features such as staining depth, cell density, and smear thickness. This is done to enable rapid retrieval and analysis of cell morphology data for specific samples in subsequent data analysis, while also providing fundamental data support for potential further research. In this way, the system can efficiently manage and utilize the recognition results, providing comprehensive data support for leukemia cell morphology analysis.

[0052] In step S700, if the cell is suspicious, the preset region growth algorithm is activated to delineate the suspicious area, the X / Y axis motor is driven to locate the area, the high-magnification objective lens is switched, the Z axis is triggered to refocus and acquire a high-resolution image, and the image is then transmitted to the AI ​​module for identification.

[0053] Among them, suspicious cells are those identified by the AI ​​recognition module as potentially exhibiting characteristics of leukemia. Region growing algorithm: an image segmentation algorithm that gradually expands from a seed point to find regions similar to the seed point, used to delineate suspicious cell areas. High-magnification objective: an objective lens used in microscopes for high-magnification imaging, providing clearer and more detailed cell images.

[0054] The specific process of activating the preset region growing algorithm to delineate the suspicious area can be found in steps S710 to S760. The specific process of driving the X / Y axis motor to position the area, switching to the high-magnification objective lens, and triggering Z-axis refocusing to acquire a high-resolution image can be found in steps S7A0 to S7E0.

[0055] Reference Figure 2 Based on the fixed leukemia cell sample, sample features are extracted, and according to preset sample feature matching rules, the optimal scanning and imaging parameters are selected from the preset parameter library to complete the device initialization, including:

[0056] Step S210: Extract sample features, including staining depth, cell density, and smear thickness.

[0057] Sample characteristics refer to the physical and chemical properties of leukemia cell samples, such as staining intensity, cell density, and smear thickness. These characteristics affect the imaging parameters of the microscope. Staining intensity refers to the depth of color the cells appear under the microscope after staining, usually measured by the grayscale value of the image. Cell density is the number of cells per unit area, typically determined through image analysis. Smear thickness refers to the physical thickness of the sample on the slide, usually obtained through optical or mechanical measurement methods.

[0058] The overall necessary process is as follows:

[0059] 1. Staining Intensity Extraction: The system captures a preliminary image of the sample using a low-power objective lens of a microscope. Image analysis algorithms are then used to calculate the average grayscale value of the cell regions. A higher grayscale value indicates lighter staining, while a lower grayscale value indicates darker staining. For example, the system can set a grayscale threshold and compare the average grayscale value to this threshold to determine the degree of staining.

[0060] 2. Cell Density Extraction: The system selects one or more representative regions on the initial image. Image analysis algorithms are used to count the number of cells within these regions and calculate the cell density per unit area. For example, the system can count the number of cells within a 100×100 pixel region and then calculate the cell density based on the actual physical size (e.g., 100×100 micrometers).

[0061] 3. Smear Thickness Extraction: The system obtains the smear thickness using optical measurement methods (such as confocal microscopy) or mechanical measurement methods (such as microprobes). For example, images are captured using a confocal microscope at different focal planes, and the smear thickness is determined by analyzing the changes in image contrast. Alternatively, the physical thickness of the smear can be directly measured using a microprobe.

[0062] Step S220: Synchronously access external laboratory temperature and humidity data and the status of the X / Y / Z axis motors of the stage.

[0063] External laboratory temperature and humidity data: Temperature and humidity data of the laboratory environment, which affect the imaging effect of the microscope and the stability of the sample. Stage X / Y / Z axis motor status: The operating status of the stage motors in the X, Y, and Z axes, including the current position, operating speed, cumulative operating time, positioning deviation, and other information.

[0064] After extracting sample features, the system needs to synchronously access temperature and humidity data from an external laboratory, as well as the current status data of the X / Y / Z axis motors of the stage, to ensure the accuracy and reliability of subsequent parameter adjustments. The specific process is as follows:

[0065] 1. Temperature and Humidity Data Acquisition: The system acquires real-time temperature and humidity data through sensors connected to the laboratory environmental monitoring system. These sensors are typically installed near the microscope to ensure that the measured data accurately reflects the actual working environment of the microscope. The system compares the acquired temperature and humidity data with preset normal ranges to determine whether the current environment is suitable for sample imaging.

[0066] 2. Access to Stage Motor Status Data: The system communicates with the stage motor control system to obtain status information such as the current position, running speed, cumulative running time, and positioning deviation of the X / Y / Z axis motors. This data is crucial for subsequent parameter adjustments, as the motor's operating status directly affects the stage's movement accuracy and stability. Based on the motor's cumulative running time and positioning deviation, the system determines whether the motor requires maintenance or adjustment.

[0067] Step S230: Establish the correlation between sample features and parameter benchmarks according to preset rules. Adjust the imaging exposure time according to the staining degree, adjust the scanning step size according to the cell density, and adjust the objective lens magnification according to the smear thickness.

[0068] Among them, the parameter baseline refers to the preset microscope imaging parameter baseline values, which are determined in advance based on the characteristics of standard samples and experimental conditions, and are used to set the initial imaging parameters. Staining degree refers to the intensity of the color after staining the sample, affecting the exposure time during imaging. Cell density refers to the number of cells per unit area, affecting the scanning step size. Smear thickness refers to the physical thickness of the sample on the slide, affecting the selection of the objective lens magnification.

[0069] After acquiring sample features and external data, the system needs to establish a correlation between the sample features and parameter benchmarks based on this information to determine the initial scanning and imaging parameters. The specific process is as follows:

[0070] 1. Relationship between staining degree and imaging exposure time: The system adjusts the imaging exposure time according to the staining degree of the sample. Deeply stained samples require shorter exposure times to avoid overexposure; lightly stained samples require longer exposure times to ensure image clarity. For example, if the staining degree of a sample is measured by grayscale value, and the grayscale value is below a certain preset threshold (e.g., 128), the exposure time is set to a shorter value (e.g., 50ms); if the grayscale value is above this threshold, the exposure time is set to a longer value (e.g., 100ms).

[0071] 2. Correlation between cell density and scan step size: The system adjusts the scan step size based on cell density. Areas with higher cell density require smaller scan step sizes to ensure sufficient detail coverage; areas with lower cell density can use larger scan step sizes to improve scanning efficiency. For example, if the cell density is higher than a preset threshold (e.g., 1000 cells per square millimeter), the scan step size is set to a smaller value (e.g., 0.1 mm); if the cell density is lower than this threshold, the scan step size is set to a larger value (e.g., 0.2 mm).

[0072] 3. Relationship between smear thickness and objective magnification: The system adjusts the objective magnification based on the smear thickness. Thicker smears require higher objective magnification to clearly observe cell morphology; thinner smears can use lower objective magnification. For example, if the smear thickness exceeds a certain preset threshold (e.g., 10 micrometers), a high-power objective (e.g., 40×) is selected; if the smear thickness is below this threshold, a low-power objective (e.g., 10×) is selected.

[0073] Step S240: Correct parameters using a feature-environment-hardware coupling algorithm, adjust exposure time and scanning step size based on temperature and humidity deviations, adjust Z-axis focusing step size based on motor operating threshold, and generate candidate parameter set by calling preset parameter library after correction.

[0074] The algorithm comprises the following components: Feature-Environment-Hardware Coupling Algorithm: This algorithm comprehensively considers sample characteristics, laboratory environmental conditions, and hardware status to dynamically adjust the microscope's imaging parameters. Temperature and Humidity Deviation: This measures the deviation of the current laboratory temperature and humidity from preset standard values, as these deviations affect imaging results. Motor Operating Thresholds: Preset thresholds for motor operating parameters, such as cumulative running time and positioning deviation, are used to determine whether the motor requires adjustment or maintenance. Preset Parameter Library: This database stores various possible combinations of scanning and imaging parameters, used to select the optimal parameters based on sample characteristics. Candidate Parameter Set: Parameter combinations generated after algorithmic correction; these combinations will be further filtered to determine the optimal parameters.

[0075] The overall necessary process can be referred to in steps S241 to S246, which will not be repeated here.

[0076] Step S250: Drive the X / Y axis motors to run at the preset calibration points under no-load according to the candidate step length, collect the encoder positioning deviation, and select the parameter with the smallest deviation as the optimal scanning and imaging parameter.

[0077] The candidate parameter set consists of several components: Candidate parameter set: Parameter combinations generated after correction using a feature-environment-hardware coupling algorithm; these combinations represent potential optimal parameters. Encoder positioning deviation: The deviation between the actual position and the target position of the stage motor during movement, obtained through encoder feedback. Pre-set calibration points: Several pre-defined locations on the stage used to verify the effectiveness and accuracy of the candidate parameter set. Optimal scanning and imaging parameters: Parameter combinations selected from the candidate parameter set that provide the best imaging effect.

[0078] After generating the candidate parameter set, the system needs to select the optimal scanning and imaging parameters through actual testing. The specific process is as follows:

[0079] 1. Drive the X / Y axis motors: The system drives the X / Y axis motors of the stage to run unloaded according to the step size in the candidate parameter set, reaching the preset verification points. These verification points are pre-set and used to verify the validity of the parameters. For example, the system can select the four corners and the center position of the stage as verification points to ensure the movement accuracy of the stage throughout the entire working area.

[0080] 2. Collect encoder positioning deviation: At each calibration point, the system collects the encoder feedback signal and records the deviation between the actual position and the target position. For example, if the target position is (10mm, 10mm) and the actual position is (10.02mm, 10.01mm), then the deviation is (0.02mm, 0.01mm).

[0081] 3. Parameter Performance Evaluation: The system evaluates the positioning deviation of each candidate parameter set, calculating the average deviation and maximum deviation. For example, the system can calculate the average deviation of all calibration points and select the parameter set with the smallest average deviation as the optimal parameter set.

[0082] 4. Selecting the Optimal Parameters: The system selects the parameter combination with the smallest deviation from the candidate parameter set as the optimal scanning and imaging parameters. For example, if there are three candidate parameter sets with average deviations of 0.01mm, 0.02mm, and 0.03mm respectively, the system will select the parameter set with an average deviation of 0.01mm as the optimal parameters.

[0083] 5. Verification and Validation: The system performs a complete scan and imaging test under the selected optimal parameters to verify whether the actual imaging effect meets the preset requirements. If the imaging effect is not ideal, the system will readjust the parameters and repeat the above process until a parameter combination that meets the requirements is found.

[0084] Step S260: Bind the optimal parameters with the sample ID, real-time temperature and humidity, and motor status to the database, generate an initialization log, and complete the device initialization.

[0085] The system includes: Parameter Binding: Associating and storing the selected optimal scanning and imaging parameters with the sample's unique identifier (sample ID) and other relevant information (such as real-time temperature and humidity, motor status). Database: A system for storing sample features, identification results, imaging parameters, etc., facilitating subsequent data analysis and traceability. Initialization Log: A log file recording all key information during device initialization, including sample ID, optimal parameters, environmental data, and hardware status.

[0086] After selecting the optimal scanning and imaging parameters, the system binds these parameters with the sample's unique identifier and other relevant information, stores them in the database, and generates an initialization log. The specific process is as follows:

[0087] 1. Parameter Binding: The system associates optimal scanning and imaging parameters with the unique identifier of each sample (sample ID). This ensures that the imaging parameters corresponding to each sample can be accurately traced in subsequent analysis. For example, if the sample ID is "S12345", the optimal parameters include an exposure time of 50ms, a scan step size of 0.1mm, and an objective lens magnification of 40×, and this information will be bound together.

[0088] 2. Storage in Database: The system stores the bound parameters and sample information in the database. The database stores not only sample characteristic information but also imaging parameters, environmental data, and hardware status. For example, a record in the database may include information such as sample ID, staining depth, cell density, smear thickness, exposure time, scan step size, objective magnification, laboratory temperature, laboratory humidity, and motor status.

[0089] 3. Generate Initialization Log: The system generates an initialization log, recording all key information during the device initialization process. The initialization log includes sample ID, optimal parameters, real-time temperature and humidity, motor status, etc.

[0090] 4. Complete Device Initialization: Once parameter binding and storage are complete and the initialization log is generated, the system confirms that the device initialization process has been successfully completed. The system will then enter standby mode, awaiting further instructions from the user or automated processes, such as initiating sample scanning or image acquisition.

[0091] Parameters are corrected using a feature-environment-hardware coupling algorithm. Exposure time and scanning step size are adjusted based on temperature and humidity deviations. The Z-axis focusing step size is adjusted according to motor operating thresholds. After correction, a candidate parameter set is generated from a preset parameter library, including:

[0092] Step S241: Based on the collected sample characteristics, external laboratory temperature and humidity data, and the status of the stage X / Y / Z axis motors, construct a dynamic data chain of sample-environment-hardware.

[0093] Specifically, a dynamic data chain is constructed: the system integrates collected sample features, environmental data, and hardware status information into a dynamic data chain. This dynamic data chain is a real-time updated data structure that dynamically reflects the current sample features, environmental conditions, and hardware status.

[0094] Step S242: Using a feature-environment-hardware coupling algorithm, the imaging exposure time and scanning step size are first corrected based on the sample features and the temperature and humidity deviation. Then, according to the preset motor running threshold, if the cumulative running time of the motor exceeds the preset threshold or the positioning deviation exceeds the preset threshold, the Z-axis focusing step size is adjusted accordingly to form a complete parameter correction result. The preset motor running threshold includes the preset cumulative running time threshold and the preset positioning deviation threshold.

[0095] The feature-environment-hardware coupling algorithm is a comprehensive algorithm that considers sample features, laboratory environmental conditions, and hardware status to dynamically adjust the microscope's imaging parameters. Imaging exposure time: The length of time the camera is exposed during microscope imaging, affecting image brightness and sharpness. Scan step size: The distance the stage moves in each step along the X / Y axes, affecting scanning resolution and speed. Z-axis focusing step size: The distance the stage moves in each step along the Z-axis, affecting focusing accuracy and speed. Preset motor operating thresholds: Including preset cumulative motor operating time thresholds and preset motor positioning deviation thresholds, used to determine whether the motor needs adjustment or maintenance.

[0096] After constructing a dynamic data chain of sample-environment-hardware, the system employs a feature-environment-hardware coupling algorithm to correct the imaging parameters. The specific process is as follows:

[0097] 1. Initial correction based on sample characteristics: The system first makes initial corrections to the imaging exposure time and scanning step size based on sample characteristics (such as staining depth, cell density, and smear thickness).

[0098] Staining depth and exposure time: If the sample is stained deeply, reduce the exposure time to avoid overexposure; if the staining is light, increase the exposure time to ensure image clarity. For example, if the staining depth value is below 0.5 (assuming a range of 0 to 1), reduce the exposure time from the default value of 100ms to 50ms; if the staining depth value is above 0.5, increase the exposure time to 150ms.

[0099] Cell density and scan step size: If the cell density is high, decrease the scan step size to ensure sufficient detail coverage; if the cell density is low, increase the scan step size to increase scan speed. For example, if the cell density exceeds 1000 cells per square millimeter, decrease the scan step size from the default 0.2 mm to 0.1 mm; if the cell density is below 1000 cells, increase the scan step size to 0.3 mm.

[0100] 3. Further corrections will be made based on the degree of deviation in temperature and humidity:

[0101] The system further corrects the imaging exposure time and scanning step size based on the deviation between the current temperature and humidity in the laboratory and the preset standard values.

[0102] Temperature Effects: If the laboratory temperature exceeds the preset value (e.g., 25°C), it may accelerate sample drying and affect image quality. The system will appropriately increase the exposure time to compensate for this effect. For example, if the temperature exceeds 25°C, the exposure time will increase by 10 ms for every 1°C increase.

[0103] Humidity Effects: If the laboratory humidity is below the preset value (e.g., 50%), it may cause the sample to dry out, affecting image quality. The system will appropriately increase the scan step size to reduce scan time and avoid excessive sample drying. For example, if the humidity is below 50%, the scan step size increases by 0.05 mm for every 10% decrease.

[0104] 3. Adjust the Z-axis focusing step size according to the motor operating threshold:

[0105] The system checks the cumulative running time and positioning deviation of the stage's X / Y / Z axis motors. If these values ​​exceed the preset threshold, the system will adjust the Z-axis focusing step size.

[0106] Cumulative running time: If the cumulative running time of the motor exceeds a preset threshold (e.g., 10 hours), the system will increase the Z-axis focusing step size to reduce focusing time and prevent the motor from overheating. For example, if the cumulative running time exceeds 10 hours, the focusing step size will increase from the default value of 0.05mm to 0.1mm.

[0107] Positioning Deviation: If the motor positioning deviation exceeds a preset threshold (e.g., 0.1mm), the system will reduce the Z-axis focusing step size to improve focusing accuracy. For example, if the positioning deviation exceeds 0.1mm, the focusing step size will be reduced from the default value of 0.1mm to 0.05mm.

[0108] 4. Generate complete parameter correction results: After the above correction steps, the system generates complete parameter correction results, including the corrected imaging exposure time, scanning step size and Z-axis focusing step size.

[0109] Step S243: If multiple factors conflict during parameter correction, the preset historical imaging quality database is called according to the preset imaging clarity priority principle, and the preset optimal parameter adaptation cases of the same type of sample are referenced to determine the final correction direction.

[0110] Among the key challenges are: Multi-factor adjustment conflicts: During parameter correction, the influence of multiple factors such as sample characteristics, environmental conditions, and hardware status may lead to conflicts between different correction directions. Imaging sharpness priority principle: In parameter correction, priority is given to ensuring imaging sharpness, even if it requires sacrificing other performance aspects (such as scanning speed). Preset historical imaging quality database: This database stores a large amount of historical imaging data and corresponding parameters, used for reference to optimal parameter adaptation cases for similar samples. Optimal parameter adaptation cases: Parameter configuration cases that successfully obtained high-quality imaging under similar sample characteristics and environmental conditions.

[0111] During parameter correction, conflicts may arise due to multiple factors. For example, the exposure time might be adjusted to compensate for deviations in ambient temperature and humidity, but simultaneously, the motor's cumulative running time may be approaching a threshold, requiring a reduction in exposure time to prevent motor overheating. In such cases, the system needs to resolve the conflict according to preset rules. The specific process is as follows:

[0112] 1. Detection of Multi-Factor Adjustment Conflicts: When performing parameter corrections, the system monitors in real time for the existence of multi-factor adjustment conflicts. For example, the system may detect that the exposure time needs to be increased to compensate for temperature and humidity deviations, but the cumulative motor running time is approaching the threshold, requiring a reduction in exposure time. The system records the specific details of the conflict, including the factors involved, current parameter values, and adjustment direction.

[0113] 2. Applying the Image Sharpness Priority Principle: When multiple adjustment conflicts are detected, the system first applies the image sharpness priority principle. This means that in conflict situations, the system prioritizes the adjustment direction that ensures image sharpness. For example, if increasing the exposure time can significantly improve image sharpness, even if this causes the motor running time to approach the threshold, the system will prioritize increasing the exposure time.

[0114] 3. Accessing the Historical Imaging Quality Database: If the principle of prioritizing image sharpness cannot completely resolve the conflict, the system will access the preset historical imaging quality database to find the optimal parameter matching cases similar to the current sample characteristics and environmental conditions. For example, the system will search the database for successful imaging parameter configurations for similar samples (such as the same staining depth, cell density, and smear thickness) under similar temperature and humidity conditions.

[0115] 4. Determine the final correction direction: The system refers to the optimal parameter adaptation cases in the historical imaging quality database, and combines the current sample characteristics and environmental conditions to determine the final correction direction. For example, if a case in the historical database shows that high-quality imaging can be obtained under similar conditions by increasing the exposure time and decreasing the scan step size, the system will select this adjustment direction.

[0116] 5. Recording and Feedback: The system records the final correction direction and adjusted parameter values ​​for subsequent verification and optimization.

[0117] Step S244: Based on the complete parameter correction results corresponding to the final correction direction, call the preset parameter library and filter parameter combinations according to the rule that the deviation between the corrected parameters and the parameters in the library does not exceed the preset deviation threshold.

[0118] The system comprises the following components: Preset Parameter Library: A database storing various possible combinations of scanning and imaging parameters, pre-defined based on historical data and experimental experience. Corrected Parameters: Imaging parameters corrected using a feature-environment-hardware coupling algorithm, including exposure time, scan step size, and Z-axis focusing step size. Preset Deviation Threshold: The maximum allowable deviation between the corrected parameters and those in the preset parameter library, used to select suitable parameter combinations. Fit Scoring Rules: A set of rules for evaluating whether a parameter combination is suitable for the current sample characteristics and environmental conditions, typically based on multiple dimensions such as image sharpness and scanning efficiency.

[0119] The general process is as follows: After determining the final correction direction, the system calls a preset parameter library and compares the deviation between the corrected parameters and the parameters in the library. Based on a preset deviation threshold, parameter combinations with deviations within the threshold are selected. Then, a score is calculated for each parameter group according to the fit scoring rules and labeled. Finally, parameter combinations with scores not lower than the preset threshold are retained, sorted from highest to lowest total score, forming a candidate parameter set with fit scores, providing a basis for subsequent selection of the optimal parameters.

[0120] Step S245: For each set of parameters after screening, calculate and mark the score according to the preset fit scoring rules.

[0121] After the parameters are corrected using a feature-environment-hardware coupled algorithm, and suitable parameter combinations are selected from a pre-set parameter library, the system needs to score the suitability of these parameter combinations. The purpose of this step is to evaluate the applicability of each parameter combination under the current sample features and environmental conditions, in order to further select the optimal parameter combination. The following is a detailed disclosure of this step:

[0122] 1. Fit Scoring Rules: Fit scoring rules are a set of evaluation criteria based on multiple dimensions such as image quality and scanning efficiency. These rules typically include the following aspects:

[0123] Image sharpness: Evaluates whether the combination of parameters can produce a sharp image under the current conditions.

[0124] Scanning efficiency: Evaluate whether the combination of parameters can efficiently complete the scanning task while ensuring image quality.

[0125] Environmental adaptability: assess the stability of the parameter combination under current laboratory environmental conditions.

[0126] Hardware compatibility: Evaluate whether the parameter combination is compatible with the current hardware status to avoid hardware overload or damage.

[0127] 2. Calculate the fit score: The system calculates and labels a score for each filtered parameter combination according to preset fit scoring rules. The specific process is as follows:

[0128] Initialization score: The system initializes a fit score for each parameter combination, with the initial value usually being 0.

[0129] Image Sharpness Assessment: The system evaluates image sharpness based on the corrected exposure time and scan step size. For example, a moderate exposure time and a reasonable scan step size result in a higher image sharpness score. Image sharpness typically has a high weighting, such as accounting for 60% of the total score. Scanning Efficiency Assessment: The system evaluates scanning efficiency based on the scan step size and Z-axis focusing step size. For example, a large scan step size and a moderate focusing step size result in a higher scanning efficiency score. Scanning efficiency typically has a lower weighting, such as accounting for 40% of the total score.

[0130] Environmental adaptability assessment: The system evaluates the environmental adaptability of parameter combinations based on the current temperature and humidity conditions of the laboratory. For example, if a parameter combination can operate stably under the current temperature and humidity conditions, the environmental adaptability score is higher. The weight of environmental adaptability is usually moderate, for example, accounting for 30% of the total score.

[0131] Hardware compatibility assessment: The system evaluates the hardware compatibility of parameter combinations based on the current hardware status. For example, if a parameter combination will not cause motor overload or excessive positioning deviation, the hardware compatibility score is high. The weight of hardware compatibility is usually moderate, for example, accounting for 30% of the total score.

[0132] Overall Score: Based on the above evaluation results, the system calculates an overall fit score for each parameter combination. The overall score is a weighted sum of the scores from each dimension.

[0133] For example, the formula for overall fit score can be expressed as: Overall fit score = 0.6 × Image sharpness score + 0.4 × Scanning efficiency score.

[0134] 3. Scoring: The system labels the calculated fit score on each parameter combination for subsequent filtering and comparison. For example, the fit score for a parameter combination might be as follows:

[0135] Parameter combination 1: Overall compatibility score 0.75; Parameter combination 2: Overall compatibility score 0.80; Parameter combination 3: Overall compatibility score 0.65.

[0136] 4. Retain high-scoring parameter combinations: The system retains parameter combinations with scores no lower than a preset scoring threshold. This preset threshold is typically set based on historical data and experimental experience, for example, 0.70. The retained parameter combinations will proceed to the next step for further selection of optimal parameters.

[0137] Step S246: retain parameter combinations with scores not lower than a preset score threshold, sort them from high to low according to the total score, retain a preset number of parameter combinations, and form a candidate parameter set with fit score.

[0138] 1. Scoring Threshold Screening: The system first filters parameter combinations with scores no lower than a preset fitness score threshold. The preset score threshold is usually set based on historical data and experimental experience to ensure that the selected parameter combinations have high fitness.

[0139] 2. Sorting and Retention: The filtered parameter combinations are sorted from highest to lowest according to their suitability score, and a preset number of parameter combinations are retained. This step ensures that the system can quickly select the optimal parameter combination in subsequent operations, improving the system's operating efficiency.

[0140] 3. Generate candidate parameter set: Combine the retained parameter combinations and their fitness scores to form a candidate parameter set.

[0141] After the stage comes to a stop, the Z-axis drive motor is triggered to perform a preset dual-stage focusing algorithm. Using a preset contrast determination rule, a clear imaging plane is found, and simultaneously, the camera is triggered to acquire data via an FPGA timing control algorithm.

[0142] Step S410: After the stage comes to a stop, the previously collected external laboratory temperature and humidity data and the stage Z-axis motor status data are retrieved, and the Z-axis coarse and fine adjustment step sizes are corrected in combination with the preset dual-stage focusing algorithm initial parameters.

[0143] After the stage has come to a stable stop, the system needs to adjust the Z-axis focusing step size based on the previously collected external laboratory temperature and humidity data and the stage Z-axis motor status data. The purpose of this step is to ensure that, under the current environmental and hardware conditions, the Z-axis can quickly and accurately locate a clear imaging plane. The following is a detailed disclosure of this step:

[0144] 1. Retrieving Previous Data: The system first retrieves previously collected external laboratory temperature and humidity data, as well as the Z-axis motor status data of the stage. This data includes: Laboratory temperature and humidity: the current temperature and humidity of the laboratory, which affect the stability of the sample and the imaging effect. Z-axis motor status: including the current position of the motor, cumulative running time, positioning deviation, etc., which affect the focusing accuracy and stability of the Z-axis.

[0145] 2. Using preset dual-stage focusing algorithm initial parameters: The system uses the preset dual-stage focusing algorithm initial parameters to correct the coarse and fine adjustment step sizes of the Z-axis. The preset dual-stage focusing algorithm typically includes two stages: Stage 1 (coarse adjustment): Rapidly moves the Z-axis with large step sizes to locate the region of contrast abrupt change. Stage 2 (fine adjustment): Finely adjusts the Z-axis with smaller step sizes to lock onto the sharp imaging plane.

[0146] 3. Correct Z-axis focus step size: The system corrects the Z-axis focus step size based on the called data and preset algorithm. The specific process is as follows:

[0147] The environmental conditions are modified as follows:

[0148] Temperature Correction: If the laboratory temperature exceeds the preset value (e.g., 25°C), it may cause the sample to expand or shrink, affecting focusing. The system will adjust the focusing step size according to the temperature deviation. For example, for every 1°C increase, the coarse adjustment step size increases by 0.01mm, and the fine adjustment step size decreases by 0.005mm.

[0149] Humidity Correction: If the laboratory humidity is lower than the preset value (e.g., 50%), it may cause the sample to dry out, affecting focus. The system will adjust the focus step size according to the humidity deviation. For example, for every 10% decrease, the coarse adjustment step size decreases by 0.01mm, and the fine adjustment step size increases by 0.005mm.

[0150] The hardware status has been corrected as follows:

[0151] Cumulative running time: If the cumulative running time of the Z-axis motor exceeds a preset threshold (e.g., 10 hours), it may cause a decrease in motor performance and affect focusing accuracy. The system will adjust the focusing step size based on the cumulative running time. For example, for every hour exceeding 1 hour, the coarse adjustment step size increases by 0.01mm, and the fine adjustment step size decreases by 0.005mm.

[0152] Positioning Deviation: If the positioning deviation of the Z-axis motor exceeds a preset threshold (e.g., 0.05mm), it may cause inaccurate focusing. The system will adjust the focusing step size according to the positioning deviation. For example, for every 0.01mm deviation, the coarse adjustment step size decreases by 0.01mm, and the fine adjustment step size increases by 0.005mm.

[0153] 4. Generate the corrected focus step size: The system generates the corrected Z-axis focus step size according to the above correction rules. For example, the corrected focus step size may be as follows: Corrected coarse adjustment step size: 0.5mm;

[0154] Corrected fine-tuning step size: 0.05mm.

[0155] Step S420: Perform dual-stage focusing with the corrected step size. In the first stage, move the Z-axis with a preset large step size to quickly locate the contrast change range. In the second stage, move the Z-axis with a preset small step size and lock the clear imaging plane through the preset contrast determination rule.

[0156] Two-stage focusing: A focusing method consisting of two stages, coarse adjustment and fine adjustment, used to quickly and accurately find the optimal focus plane. Contrast judgment rule: A rule used to determine whether the image contrast meets the standard for sharp imaging, usually based on image grayscale variations or edge intensity. Preset large step size: A large Z-axis movement step size used in the first stage of focusing (coarse adjustment) to quickly locate contrast abrupt changes. Preset small step size: A smaller Z-axis movement step size used in the second stage of focusing (fine adjustment) to precisely lock onto the sharp imaging plane.

[0157] The necessary process is described as follows: After the stage stabilizes, the system enters a two-stage focusing process. First, the system moves rapidly along the Z-axis with a preset large step size to quickly locate the region of contrast abrupt change. This stage is called coarse adjustment, and its goal is to quickly approach the optimal focus plane and reduce focusing time. Next, the system switches to a preset small step size to finely adjust the Z-axis position and lock in the sharp imaging plane using a preset contrast determination rule. This stage is called fine adjustment, and its goal is to accurately find the optimal focus plane to ensure image sharpness. The contrast determination rule is usually based on the edge intensity or grayscale changes of the image. For example, an edge detection algorithm is used to calculate the edge intensity of the image, and when the edge intensity exceeds a preset threshold, a sharp imaging plane is considered to have been found. Through this two-stage focusing method, the system can efficiently and accurately find the optimal focus plane, providing a clear imaging foundation for subsequent image acquisition.

[0158] In step S430, while locking onto the clear imaging plane, the FPGA timing control algorithm is started, and the acquisition signal is synchronously sent to the camera according to the preset hardware trigger delay.

[0159] Among them, the FPGA timing control algorithm is an algorithm based on a field-programmable gate array (FPGA) used to precisely control the timing of image acquisition by the camera, ensuring the synchronization and accuracy of image acquisition. Hardware trigger delay: The time delay between the trigger signal being sent to the camera and the actual image acquisition needs to be precisely controlled to ensure the synchronization of image acquisition.

[0160] The necessary process is described as follows: After locking onto a clear imaging plane, the system activates the FPGA timing control algorithm, synchronously sending acquisition signals to the camera according to a preset hardware trigger delay. The FPGA timing control algorithm precisely controls the camera's acquisition timing based on the preset delay time, ensuring image acquisition on the optimal focusing plane. For example, if the preset hardware trigger delay is 100 microseconds, the FPGA will send the acquisition signal within 100 microseconds after focusing is complete, allowing the camera to capture the image at the moment of peak focus. Through this precise timing control, the system can acquire high-quality, clear images, providing a reliable foundation for subsequent analysis and processing.

[0161] Step S440: Receive images captured by the camera in real time, determine whether the image meets the standard through a preset sharpness evaluation model, and if it does not meet the standard, readjust the fine-tuning step size by combining the current temperature and humidity fluctuations with the real-time positioning deviation of the Z-axis motor.

[0162] Among them, the sharpness assessment model is a model used to evaluate image sharpness, typically based on features such as edge strength and texture sharpness. Temperature and humidity fluctuation refers to the deviation between the current laboratory temperature and humidity and preset standard values; these fluctuations may affect imaging results. Z-axis motor real-time positioning deviation refers to the deviation between the actual position of the Z-axis motor and the target position during focusing; these deviations may affect focusing accuracy.

[0163] The necessary process is described as follows: The system receives images captured by the camera in real time and determines whether the images meet the standards using a preset sharpness evaluation model. This model evaluates the images based on features such as edge strength and texture sharpness. If the image does not meet the standards, the system readjusts the fine-tuning step size by considering the current temperature and humidity fluctuations and the real-time positioning deviation of the Z-axis motor. For example, if the temperature and humidity fluctuations exceed the preset range, the system will adjust the step size to compensate for these fluctuations; if the Z-axis motor positioning deviation is too large, the system will adjust the step size accordingly to improve focusing accuracy. Through this dynamic adjustment, the system ensures that clear images can be acquired under different conditions, providing high-quality data support for subsequent image analysis.

[0164] Step S450: Repeat the dual-stage focusing and FPGA synchronous acquisition process according to the revised step size until a qualified image is obtained, and store the final focusing parameters, acquisition timing data and sample association.

[0165] The system includes: Dual-stage focusing: a focusing method consisting of coarse and fine adjustments to quickly and accurately find the optimal focus plane. FPGA timing control algorithm: an algorithm based on a Field-Programmable Gate Array (FPGA) to precisely control the timing of image acquisition by the camera, ensuring synchronization and accuracy. Sharpness evaluation model: a model for evaluating image sharpness, typically based on features such as edge strength and texture sharpness. Final focus parameters: the optimal focus parameters determined after multiple adjustments, including the Z-axis position and focus step size. Acquisition timing data: timing information related to image acquisition, including the trigger signal transmission time and the camera's actual acquisition time.

[0166] The system receives images captured by the camera in real time and determines whether the image meets the standards using a preset sharpness evaluation model. If the image does not meet the standards, it readjusts the fine-tuning step size by considering the current temperature and humidity fluctuations and the real-time positioning deviation of the Z-axis motor.

[0167] Step S441: Collect image quality data, real-time environmental fluctuation data, hardware status data, and previous sample feature data to construct a four-dimensional dynamic data chain of image quality, environmental fluctuation, hardware status, and sample features.

[0168] The four-dimensional dynamic data chain is a data structure that integrates data from four dimensions: image quality, environmental fluctuations, hardware status, and sample characteristics. This data is used to comprehensively evaluate and adjust focusing parameters. Image quality data includes image sharpness scores, edge strength, and texture sharpness, used to assess image quality. Real-time environmental fluctuation data includes current temperature and humidity fluctuations in the laboratory, collected in real-time by environmental sensors. Hardware status data includes real-time positioning deviations and cumulative running time of the Z-axis motor, acquired in real-time by the motor control system.

[0169] The necessary process is described as follows: The system first collects current image quality data, including sharpness score and edge strength, while simultaneously acquiring real-time environmental fluctuation data, such as temperature and humidity changes, and hardware status data, such as the positioning deviation of the Z-axis motor. Combined with previously stored sample feature data, the system constructs a four-dimensional dynamic data chain. For example, if the image sharpness score is below a preset threshold, it indicates that the image quality is substandard, and the current temperature and humidity fluctuations and motor positioning deviation are recorded. This data will serve as the basis for subsequent adjustments to the focus step size, ensuring that clear images are acquired under various conditions.

[0170] Step S442: A four-dimensional coupling correction algorithm is adopted, which takes image quality data as the main factor and triggers correction when the sharpness score is lower than the preset threshold. The step size is adjusted according to the preset coefficient when the temperature and humidity fluctuations in the environmental fluctuation data exceed the preset range. The step size is compensated according to the preset amount when the Z-axis motor positioning deviation in the hardware status data exceeds the preset micrometer value. The feature data of the synchronous reference sample is adjusted according to the preset increment rule to form a fine-tuning step size correction scheme.

[0171] The algorithm includes: a four-dimensional coupling correction algorithm (which comprehensively considers image quality, environmental fluctuations, hardware status, and sample features) used to dynamically adjust the focus step size; a sharpness score calculated based on features such as edge strength and texture sharpness to evaluate image sharpness; a preset threshold value that the image sharpness score must reach to determine if the image is sharp; and preset coefficients used to adjust the step size, dynamically calculated based on temperature and humidity fluctuations and hardware status.

[0172] Preset compensation step size: The compensation step size is adjusted according to the hardware status (such as motor positioning deviation).

[0173] The necessary process is described as follows: The system employs a four-dimensional coupled correction algorithm, primarily based on image quality data, to assess whether image sharpness meets the standards. If the sharpness score is lower than a preset threshold, the algorithm triggers a correction mechanism. Combining environmental fluctuation data, if temperature and humidity fluctuations exceed a preset range, the step size is adjusted by a preset coefficient; combining hardware status data, if the Z-axis motor positioning deviation exceeds a preset micrometer value, the step size is compensated by a preset amount. Simultaneously, referencing sample feature data, the step size is adjusted according to a preset incremental rule, forming a fine-tuning step size correction scheme. For example, if the image sharpness score is 0.6 (lower than the preset threshold of 0.8) and temperature and humidity fluctuations exceed the range, the system adjusts the step size by a preset coefficient of 0.05; if the Z-axis motor positioning deviation is 0.03mm (exceeding the preset 0.02mm), the system compensates with a step size of 0.01mm. Finally, the system further adjusts the step size according to the sample characteristics and the preset incremental rule to ensure that image sharpness meets the standards.

[0174] Step S443: If there are multiple factors that conflict in the adjustment of the correction plan, the preset historical correction database is called according to the preset priority rules, and the preset optimal adaptation case with the same sample characteristics and the same environmental fluctuation range is referenced to determine the final correction direction.

[0175] Among these, the following are key considerations: Multi-factor adjustment conflicts: When adjusting the focus step size, factors such as environmental fluctuations, hardware status, and sample characteristics can lead to conflicting adjustment directions or values. Preset priority rules: Pre-defined rules determine which factor has higher adjustment priority when multiple factors conflict. Preset historical correction database: A database storing past successful correction examples for reference to find the optimal adjustment scheme under similar conditions. Optimal adaptation cases: Parameter configuration cases that successfully obtained high-quality imaging under similar sample characteristics and environmental conditions.

[0176] The necessary process is described as follows: When the correction scheme generated by the four-dimensional coupling correction algorithm has multiple factor adjustment conflicts, the system, according to preset priority rules, calls a preset historical correction database to find the optimal adaptation case under similar sample features and environmental fluctuation conditions, thereby determining the final correction direction. For example, if the image sharpness score is low and the temperature and humidity fluctuate greatly, while the Z-axis motor positioning deviation exceeds the threshold, the system will prioritize image sharpness, refer to similar cases in the historical database, and select the optimal adjustment direction. This process ensures that the system can make reasonable decisions, optimize the focusing step size, and improve image quality under complex conditions.

[0177] Step S444: Output the fine-tuning step size after correction according to the final correction direction, call the preset step size verification library, and verify the feasibility according to the rule that the deviation between the corrected step size and the adapted step size in the library does not exceed the preset deviation threshold and the corresponding hardware load rate is lower than the preset safety value.

[0178] Among them, the corrected step size is the Z-axis focusing step size adjusted according to the four-dimensional coupling correction algorithm. The preset step size verification library is a database storing various possible focusing step sizes and their verification results, used to verify the feasibility of the corrected step size. The preset deviation threshold is the maximum allowable deviation between the corrected step size and the adapted step size in the preset step size verification library. The hardware load rate is the load condition of the Z-axis motor under the current focusing step size, used to assess whether the motor is in a safe operating state.

[0179] The necessary process is described as follows: The system outputs the corrected fine-tuning step size based on the final correction direction and calls the preset step size verification library for verification. The verification process includes checking whether the deviation between the corrected step size and the adapted step size in the library is within a preset threshold, and whether the corresponding hardware load rate is lower than a preset safety value. For example, if the corrected step size is 0.07mm, and the deviation from the adapted step size of 0.06mm in the verification library is 0.01mm, which is lower than the preset deviation threshold of 0.02mm, and the hardware load rate is 30%, which is lower than the preset safety value of 50%, then the verification passes. Through this verification step, the system ensures that the corrected focusing step size not only improves image quality but also does not overload the hardware, ensuring the safe operation of the device.

[0180] Step S445: After verification, the correction step size is bound and stored with the four-dimensional data. The preset image quality compliance probability and preset environmental tolerance range under this step size are simultaneously marked to form a traceable fine-tuning step size correction record.

[0181] Among them, the preset image quality compliance probability is the probability that the image quality will reach the preset standard based on historical data and current conditions. The preset environmental tolerance range is the range of temperature and humidity fluctuations that the system can tolerate under current environmental conditions, ensuring that the image quality is not affected.

[0182] The necessary process is described as follows: After successful verification, the system binds and stores the correction step size with the four-dimensional dynamic data chain, simultaneously annotating the preset image quality compliance probability and preset environmental tolerance range under that step size, forming a traceable fine-tuning step size correction record. For example, if the correction step size is 0.07mm, the system records the current image quality data, temperature and humidity fluctuations, motor positioning deviation, and sample characteristics, and annotates the image quality compliance probability as 90% and the environmental tolerance range as temperature 22-28℃ and humidity 40-60%. This record not only provides data support for subsequent analysis but also facilitates the rapid retrieval of optimal parameters under similar conditions, improving system operating efficiency.

[0183] Activating the preset region growing algorithm to delineate suspicious areas includes:

[0184] Step S710: Using the center point of the suspicious cell initially identified by the AI ​​recognition module as the initial seed point, call the previously collected sample features, real-time environmental data and the positioning deviation value of the stage X / Y axis motor, and set the confidence weight of the seed point in combination with the initial parameters of the preset region growth algorithm.

[0185] The system comprises the following components: AI Recognition Module: An image recognition system based on artificial intelligence technology, used for initial identification of suspicious cells. Initial Seed Point: The center point of the suspicious cells identified by the AI ​​Recognition Module, serving as the starting point for the region growth algorithm. Confidence Weight: A numerical value representing the degree of confidence in the initial seed point, affecting the range and speed of region growth. Sample Features: Including staining depth, cell density, smear thickness, etc., these features influence cell identification and region growth. Real-time Environmental Data: Current laboratory environmental conditions such as temperature and humidity, which may affect cell morphology and identification. Stage X / Y Axis Motor Positioning Deviation: The deviation between the actual position and the target position of the stage in the X / Y axis directions, affecting the accuracy of region growth.

[0186] The necessary process is described as follows: The system uses the center point of the suspicious cells initially identified by the AI ​​recognition module as the initial seed point, and calls upon previously collected sample features, real-time environmental data, and the positioning deviation values ​​of the stage's X / Y axis motors. Combined with the preset initial parameters of the region growth algorithm, the confidence weight of the seed point is set. For example, if the sample is deeply stained, the seed point weight may be increased to expand the growth range; if the cell density is high, the weight may be decreased to narrow the range. Simultaneously, considering environmental data and motor deviation, the weights are adjusted to adapt to the current conditions. This process ensures that the region growth algorithm can dynamically adjust according to specific conditions, improving the accuracy and reliability of the identification.

[0187] Step S720: Adjust the weights based on sample features. In the deep staining region, increase the weight of the seed point to a preset ratio. In the high cell density region, reduce the neighborhood range of the initial seed point to a preset pixel value.

[0188] Among them, weight: In the region growing algorithm, this is a parameter used to adjust the growth range and speed of the seed point. Preset ratio: A fixed ratio used when adjusting the weights based on sample features, used for dynamically adjusting the weight values. Preset pixel value: In high-cell-density regions, the size of the neighborhood range set for the initial seed point.

[0189] The necessary process is described below: The system adjusts the weight of the initial seed point based on the sample characteristics. If the sample is deeply stained, the system increases the seed point weight to a preset proportion, such as 20%, to expand the growth range and ensure coverage of more possible cell regions. Conversely, in high-cell-density areas, the system reduces the neighborhood of the initial seed point to a preset pixel value, such as from 5 pixels to 3 pixels, to improve growth accuracy and avoid overgrowth. This dynamic adjustment based on sample characteristics enables the region growing algorithm to more accurately identify and delineate suspicious cell regions, providing a more reliable foundation for subsequent detailed analysis.

[0190] Step S730: Establish growth criteria that integrate external factors. The basic criteria are set as follows: gray level difference between adjacent pixels and seed point ≤ preset baseline threshold, gradient direction consistency ≥ preset percentage. The environmental adaptation criteria are set as follows: when temperature and humidity fluctuations exceed the preset range, the gray level difference threshold is tightened to the preset lower limit. The hardware compensation criteria are set as follows: when X / Y axis positioning deviation > preset micrometer value, the growth neighborhood is expanded to the deviation compensation range.

[0191] The growth criteria are as follows: Growth Criteria: Rules used in the region growing algorithm to determine whether a pixel should be added to the growing region. Basic Criteria: Growth rules based on image grayscale and gradient direction. Environmental Adaptation Criteria: Rules that adjust the growth criteria according to temperature and humidity fluctuations. Hardware Compensation Criteria: Rules that adjust the growth criteria according to the positioning deviation of the stage's X / Y axis motors.

[0192] The necessary process is described as follows: The system establishes growth criteria that integrate external factors to ensure the accuracy and adaptability of the region growing algorithm. The basic criteria set that the grayscale difference between adjacent pixels and the seed point does not exceed a preset threshold (e.g., 10), and the gradient direction consistency reaches a preset percentage (e.g., 80%) to ensure the continuity and consistency of the grown region. When temperature and humidity fluctuations exceed a preset range, the environmental adaptation criterion tightens the grayscale difference threshold to a preset lower limit (e.g., 5) to reduce the impact of environmental changes on growth. If the X / Y axis positioning deviation exceeds a preset micrometer value (e.g., 0.05 mm), the hardware compensation criterion expands the growing neighborhood to the deviation compensation range to compensate for hardware errors. The comprehensive application of these criteria enables the region growing algorithm to adapt to different sample characteristics and environmental conditions, improving the identification accuracy of suspicious regions.

[0193] Step S740: Iteratively execute the growth process. After each round of growth, call the preset historical database of suspicious areas of the same type of sample and compare the deviation between the current area and the database. If the deviation exceeds the preset threshold, adjust the growth step size in combination with the real-time running current of the X / Y axis motors of the stage. If the current fluctuates greatly, reduce the step size to the preset value.

[0194] The process includes iterative execution: repeatedly executing the region growth algorithm to gradually expand the growth area. Growth step size: the number of pixels expanded in each growth iteration, affecting the growth speed and accuracy. Preset historical database of suspicious region morphology for similar samples: stores previously successfully identified suspicious region morphology data for comparison and adjustment. Real-time operating current of the stage's X / Y axis motors: the real-time current value of the motors during operation, reflecting the motor's load.

[0195] The necessary process is described as follows: The system iteratively executes a region growing algorithm. After each round of growing, it calls a preset database of suspicious regions from similar historical samples and compares the current region with the morphological deviations in the database. If the deviation exceeds a preset threshold, the growing step size is adjusted based on the real-time operating current of the stage's X / Y axis motors. For example, if the current fluctuates significantly, indicating an increased motor load, the system will reduce the growing step size to a preset value, such as from 2 pixels to 1 pixel, to ensure the stability and accuracy of the growing process. Through this dynamic adjustment, the system can accurately delineate suspicious regions under complex conditions, improving the reliability of identification.

[0196] Step S750: When no new pixels meeting the criteria are added for a consecutive preset number of times, or when the area reaches the preset maximum threshold corresponding to the sample smear thickness, growth is terminated.

[0197] Among them, the preset number of consecutive times is defined as the number of consecutive times during region growth when no new pixels meeting the growth criteria are added, serving as one of the conditions for terminating growth. Region area: The size of the current growth region, usually expressed in terms of pixel count or actual physical size. Preset maximum threshold: A maximum area threshold for region growth set based on the sample smear thickness; exceeding this value terminates growth.

[0198] The necessary process is described below: During region growth, the system continuously monitors changes in the growth area. When no new pixels meeting the growth criteria are added for a preset number of consecutive cycles (e.g., 3 times), or when the area of ​​the growth region reaches a preset maximum threshold corresponding to the sample smear thickness, the system terminates growth. For example, if the sample smear thickness is 10 micrometers and the preset maximum threshold is 1000 pixels, the system stops growth when the growth area reaches this area. This mechanism ensures that region growth does not over-expand, effectively delineating suspicious cell areas and providing accurate boundaries for subsequent detailed analysis.

[0199] Step S760: Perform double verification on the delineated area. Verify the center sharpness of the area by using the Z-axis motor to ensure it is greater than or equal to a preset value. Correct deviations by associating imaging data from stable temperature and humidity periods. Then, extract the boundary coordinates of the suspicious area using a preset polygon fitting algorithm. Store the boundary data in association with sample features, environment, and motor status data.

[0200] The system includes: Dual verification: The designated area is verified twice to ensure its accuracy and reliability. Z-axis motor focus sharpness: The Z-axis motor is used to adjust the focus and verify whether the sharpness of the center of the area meets the preset value. Preset polygon fitting algorithm: An algorithm used to extract the boundary coordinates of suspicious areas and fit them into polygon shapes. Boundary data: The boundary coordinates of the designated area, used for subsequent analysis and storage.

[0201] The necessary process is described as follows: The system performs double verification on the identified suspicious areas. First, it verifies whether the sharpness of the area's center reaches a preset value through Z-axis motor focusing, ensuring image quality. Next, the system correlates imaging data from periods of stable temperature and humidity to correct for possible deviations. Finally, a preset polygon fitting algorithm is used to extract the boundary coordinates of the suspicious areas, and these boundary data are correlated and stored with sample features, environmental data, and motor status data. For example, if the preset sharpness value is 0.8, the system will adjust the Z-axis focus until the sharpness reaches this value, then fit the boundary and store the relevant data. This process ensures the accuracy and reliability of the identified areas, providing a solid foundation for subsequent detailed analysis.

[0202] Drive the X / Y axis motors to position the area, switch to the high-magnification objective lens, and trigger Z-axis refocusing to acquire high-resolution images, including:

[0203] Step S7A0: Retrieve the boundary coordinates of the suspicious area, the historical positioning deviation of the X / Y axis of the stage, real-time temperature and humidity data, and sample smear thickness data.

[0204] Among them, the following data are included: Suspicious area boundary coordinates: The boundary coordinates of the suspicious area delineated by the region growing algorithm, used for precise positioning. Stage X / Y axis historical positioning deviation: Records of past positioning deviations of the stage in the X / Y axis directions, used to compensate for current positioning. Real-time temperature and humidity data: Current temperature and humidity data in the laboratory, which may affect the morphology of the sample and imaging results. Sample smear thickness data: The thickness of the sample smear, affecting the selection of focusing and imaging parameters.

[0205] The necessary process is described below: The system first retrieves the boundary coordinates of the suspicious area, the historical X / Y axis positioning deviation of the stage, real-time temperature and humidity data, and sample smear thickness data. This data will be used for subsequent positioning and imaging parameter adjustments. For example, the historical X / Y axis positioning deviation data of the stage is used to compensate for possible positioning errors, ensuring that the stage can accurately move to the target position. Real-time temperature and humidity data are used to assess the impact of environmental conditions on imaging, allowing for necessary adjustments. Sample smear thickness data is used to select appropriate focusing parameters to ensure image quality. The comprehensive application of this data provides a precise positioning and parameter basis for subsequent high-magnification imaging.

[0206] Step S7B0: Based on the retrieved data, a preset three-dimensional coupling compensation algorithm is used. The preset compensation amount is added according to the temperature and humidity deviation from the reference value and the historical deviation exceeding the preset deviation threshold. After correcting the positioning target value, the X / Y axis motor is driven by a preset segmented speed regulation strategy. The position is fed back through the encoder synchronously. The position stops when the deviation is ≤ the preset positioning accuracy threshold.

[0207] The algorithm includes: a three-dimensional coupling compensation algorithm (which comprehensively considers historical positioning deviations of the stage's X / Y axes, real-time temperature and humidity data, and sample smear thickness data) to correct the positioning target value; a preset compensation amount (set based on temperature and humidity deviations from the baseline and historical deviations exceeding a preset deviation threshold) to adjust the positioning target value; a preset segmented speed control strategy (adjusting motor speed in segments based on the stage's current and target positions to improve positioning accuracy); and encoder feedback position (provided by the encoder, providing real-time position feedback from the stage motor for real-time monitoring and adjustment of the stage's position).

[0208] The necessary process is described as follows: Based on the retrieved boundary coordinates of the suspicious area, historical positioning deviations of the stage's X / Y axes, real-time temperature and humidity data, and sample smear thickness data, the system uses a preset three-dimensional coupling compensation algorithm to correct the positioning target value. For example, if the real-time temperature and humidity deviate from the baseline value, the system will adjust the positioning target value according to the preset compensation amount; if the historical deviation exceeds the preset threshold, the system will add additional compensation. The corrected target value drives the X / Y axis motors through a preset segmented speed regulation strategy, while the stage position is monitored in real time through encoder feedback to ensure that the deviation is within the preset positioning accuracy threshold. This process ensures that the stage can move accurately to the target position, providing accurate positioning for subsequent high-magnification imaging.

[0209] In step S7C0, after the positioning is stable, the preset high-magnification objective switching rate is adapted according to the comparison result between the sample smear thickness and the preset smear thickness threshold. During the switching, the motor current is monitored in real time. When the current exceeds the preset current threshold, the process is paused and the objective is finely adjusted until the current drops back down. After the process is complete, the preset high-magnification objective dedicated focusing parameter library is called.

[0210] Among them, the high-magnification objective switching rate is adjusted based on the comparison between the sample smear thickness and the preset smear thickness threshold. Motor current is the real-time current value of the motor during objective switching, used to monitor the motor load. The preset current threshold is the safe upper limit of the motor current; if this value is exceeded, the process must be paused and adjusted. A dedicated high-magnification objective focusing parameter library stores the focusing parameters for quick retrieval of suitable focusing parameters.

[0211] The necessary process is described as follows: After positioning and stabilization, the system adapts the preset high-magnification objective switching rate based on the comparison between the sample smear thickness and the preset smear thickness threshold. During the switching process, the system monitors the motor current in real time. If the current exceeds the preset threshold, the switching is paused and the objective is fine-tuned until the current returns to a safe range. Once in position, the system calls upon the high-magnification objective's dedicated focusing parameter library to ensure the accuracy of the focusing parameters. For example, if the sample smear thickness is 10 micrometers, exceeding the preset threshold of 8 micrometers, the system will reduce the switching rate to ensure stability. This process ensures the smoothness of high-magnification objective switching and the accuracy of focusing, providing a reliable guarantee for subsequent high-definition imaging.

[0212] In step S7D0, by combining the comparison results of real-time temperature and humidity fluctuations with the preset temperature and humidity fluctuation range, the Z-axis focusing step size is corrected, the preset dual-stage focusing algorithm is executed, and the signal is generated simultaneously through the FPGA according to the preset trigger delay to drive the camera to acquire high-definition images according to the preset high-magnification imaging parameters.

[0213] The preset temperature and humidity fluctuation range refers to the range of temperature and humidity fluctuations the system can tolerate; exceeding this range requires adjustment of the focus parameters. The dual-stage focusing algorithm is a focusing method consisting of coarse and fine adjustments to quickly and accurately find the optimal focus plane. The FPGA trigger delay is the time delay between the trigger signal being sent to the camera and the actual image acquisition; it needs precise control to ensure image acquisition synchronization. High-magnification imaging parameters include exposure time and gain, used for high-magnification objective imaging to ensure image sharpness and quality.

[0214] The necessary process is described below: The system compares real-time temperature and humidity fluctuations with preset temperature and humidity fluctuation ranges to adjust the Z-axis focusing step size. For example, if the temperature and humidity fluctuations exceed the preset range, the system adjusts the focusing step size to compensate for this fluctuation. Subsequently, the system executes a preset dual-stage focusing algorithm to ensure focusing accuracy. After focusing is complete, the system generates a signal via the FPGA according to a preset trigger delay, driving the camera to acquire high-resolution images according to preset high-magnification imaging parameters. This process ensures high-quality images can be acquired under different environmental conditions, providing a clear imaging foundation for subsequent analysis.

[0215] In step S7E0, a preset edge strength-texture clarity multi-feature model is used to determine the high-definition image. If the image meets the standard, it is sent to the AI ​​recognition module for secondary recognition. If the image does not meet the standard, the Z-axis step size is readjusted and the focus is repeated for acquisition. Finally, if the image meets the standard, the positioning / focus / environment data is bound to the sample and stored.

[0216] Among them, the edge strength-texture sharpness multi-feature model is a model that comprehensively evaluates image quality by analyzing features such as edge strength and texture sharpness to determine whether an image meets the standards.

[0217] The necessary process is described as follows: The system uses a preset edge intensity-texture sharpness multi-feature model to evaluate the quality of the acquired high-definition images. If the image quality meets the standards, it is transmitted to the AI ​​recognition module for secondary recognition; if it does not meet the standards, the Z-axis focusing step size is readjusted and the acquisition process is repeated. Finally, the positioning, focusing, and environmental data are bound and stored with the sample information to form a complete data record. For example, if both the edge intensity and texture sharpness of the image reach the preset threshold, the system will transmit the image to the AI ​​module for secondary cell type recognition; if the image quality does not meet the standards, the system will adjust the focusing step size and re-acquire the image until a clear image is obtained. This process ensures the reliability of image quality, providing accurate data support for subsequent cell analysis.

[0218] Based on the same inventive concept, embodiments of the present invention provide a management system for an intelligent microscopic device for recognizing leukemia cell morphology, including a memory and a processor, wherein the memory stores information that can run on the processor to implement, as described above. Figures 1 to 2 The procedure for any method.

[0219] The embodiments described in this specific implementation are preferred embodiments of this application and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.

Claims

1. A management method for a microscopic device for intelligent identification of leukemia cell morphology, characterized in that, include: The stained leukemia cell sample was fixed onto the stage; Based on the fixed leukemia cell sample, the sample features are extracted, and the optimal scanning and imaging parameters are selected by calling the preset parameter library according to the preset sample feature matching rules to complete the device initialization. The intelligent leukemia cell morphology recognition microscopy device is equipped with stage X / Y axis and Z axis drive motors. The X / Y axis drive motors of the stage are started according to the matched scanning parameters, which drive the sample to move along the preset trajectory. The deviation is corrected according to the preset anti-interference rules and the encoder position feedback to ensure that the positioning accuracy meets the preset requirements. After the stage comes to a stop, the Z-axis drive motor is triggered to find a clear imaging plane by following the preset dual-stage focusing algorithm and the preset contrast judgment rules, and the camera is triggered to acquire data simultaneously by the FPGA timing control algorithm. The acquired images are transmitted to the AI ​​recognition module, which outputs the judgment result based on the preset confidence threshold. If the cell is not suspicious, store the identification result and associate the sample features with the database; If a cell is suspected, a preset region growth algorithm is activated to delineate the suspected area, the X / Y axis motors are driven to locate the area, a high-magnification objective lens is switched, and the Z-axis is refocused to acquire a high-resolution image, which is then transmitted to the AI ​​module for identification.

2. The management method of the intelligent identification microscopic device for leukemia cell morphology according to claim 1, characterized in that, Based on the fixed leukemia cell sample, sample features are extracted, and optimal scanning and imaging parameters are selected from a preset parameter library according to preset sample feature matching rules. Device initialization is then completed, including: Extract sample features, including staining intensity, cell density, and smear thickness; Synchronously access external laboratory temperature and humidity data and the status of the X / Y / Z axis motors of the stage; The correlation between sample characteristics and parameter benchmarks is established according to preset rules. The degree of staining corresponds to the adjustment of imaging exposure time, cell density corresponds to the adjustment of scanning step size, and smear thickness corresponds to the adjustment of objective lens magnification. The parameters are corrected by a feature-environment-hardware coupling algorithm. The exposure time and scanning step size are adjusted in combination with temperature and humidity deviations. The Z-axis focusing step size is adjusted according to the motor operating threshold. After correction, a candidate parameter set is generated by calling the preset parameter library. The X / Y axis motors are driven to run at the preset calibration points under no-load according to the candidate step size. The encoder positioning deviation is collected, and the parameter with the smallest deviation is selected as the optimal scanning and imaging parameter. The optimal parameters are bound to the sample ID, real-time temperature and humidity, and motor status and stored in the database to generate an initialization log, thus completing the device initialization.

3. The management method of the intelligent identification microscopic device for leukemia cell morphology according to claim 2, characterized in that, Parameters are corrected using a feature-environment-hardware coupling algorithm. Exposure time and scanning step size are adjusted based on temperature and humidity deviations. The Z-axis focusing step size is adjusted according to motor operating thresholds. After correction, a candidate parameter set is generated from a preset parameter library, including: Based on the collected sample characteristics, external laboratory temperature and humidity data, and the status of the stage X / Y / Z axis motors, a dynamic data chain of sample-environment-hardware is constructed. The feature-environment-hardware coupling algorithm is adopted. First, the sample features are used as the main factor. The imaging exposure time and scanning step size are corrected by combining the temperature and humidity deviation. Then, based on the preset motor running threshold, if the cumulative running time of the motor exceeds the preset threshold or the positioning deviation exceeds the preset threshold, the Z-axis focusing step size is adjusted accordingly to form a complete parameter correction result. The preset motor running threshold includes the preset cumulative running time threshold and the preset positioning deviation threshold. If multiple factors conflict during parameter correction, the preset historical imaging quality database is called according to the principle of prioritizing image clarity, and the preset optimal parameter adaptation cases of similar samples are referenced to determine the final correction direction. Based on the complete parameter correction results corresponding to the final correction direction, the preset parameter library is called, and parameter combinations are selected according to the rule that the deviation between the corrected parameters and the parameters in the library does not exceed the preset deviation threshold. For each set of parameters after screening, a score is calculated and marked according to the preset fit scoring rules; Parameter combinations with scores not lower than a preset score threshold are retained, sorted from highest to lowest total score, and a preset number of parameter combinations are retained to form a candidate parameter set with a suitability score.

4. The management method of the intelligent identification microscopic device for leukemia cell morphology according to claim 3, characterized in that, After the stage comes to a stop, the Z-axis drive motor is triggered to perform a preset dual-stage focusing algorithm. Using a preset contrast determination rule, a clear imaging plane is found, and simultaneously, the camera is triggered to acquire data via an FPGA timing control algorithm. After the stage comes to a stop, the external laboratory temperature and humidity data and the stage Z-axis motor status data collected in the early stage are called up, and the Z-axis coarse and fine adjustment step size is corrected by combining the initial parameters of the preset dual-stage focusing algorithm. The dual-stage focusing is performed with the corrected step size. In the first stage, the Z-axis is moved with a preset large step size to quickly locate the contrast change range. In the second stage, the Z-axis is moved with a preset small step size and the sharp imaging plane is locked by a preset contrast judgment rule. While locking onto a clear imaging plane, the FPGA timing control algorithm is activated to synchronously send the acquisition signal to the camera according to the preset hardware trigger delay. The system receives images captured by the camera in real time and determines whether the image meets the standard through a preset clarity evaluation model. If the image does not meet the standard, it readjusts the fine-tuning step size by combining the current temperature and humidity fluctuations with the real-time positioning deviation of the Z-axis motor. Repeat the dual-stage focusing and FPGA synchronous acquisition process according to the revised step size until a qualified image is obtained, and then store the final focusing parameters, acquisition timing data and sample association.

5. The management method of the intelligent identification microscopic device for leukemia cell morphology according to claim 4, characterized in that, The system receives images captured by the camera in real time and determines whether the image meets the standards using a preset sharpness evaluation model. If the image does not meet the standards, it readjusts the fine-tuning step size by considering the current temperature and humidity fluctuations and the real-time positioning deviation of the Z-axis motor. Collect image quality data, real-time environmental fluctuation data, hardware status data, and previous sample feature data to construct a four-dimensional dynamic data chain of image quality, environmental fluctuation, hardware status, and sample features; A four-dimensional coupling correction algorithm is adopted, which takes image quality data as the main factor and triggers correction when the sharpness score is lower than the preset threshold. It combines the temperature and humidity fluctuations in the environmental fluctuation data to adjust the step size according to the preset coefficient when they exceed the preset range. It combines the Z-axis motor positioning deviation in the hardware status data to compensate the step size according to the preset amount when it exceeds the preset micron value. It also combines the synchronous reference sample feature data according to the preset increment rule to form a fine-tuning step size correction scheme. If the correction plan has multiple conflicting adjustment factors, the preset historical correction database is called according to the preset priority rules, and the preset optimal adaptation case with the same sample characteristics and the same environmental fluctuation range is referenced to determine the final correction direction. Based on the final correction direction, output the fine-tuning step size after correction, call the preset step size verification library, and verify the feasibility according to the rule that the deviation between the corrected step size and the adapted step size in the library does not exceed the preset deviation threshold and the corresponding hardware load rate is lower than the preset safety value. After verification, the correction step size is bound to the four-dimensional data and stored. The preset image quality compliance probability and preset environmental tolerance range under the step size are simultaneously marked to form a traceable fine-tuning step size correction record.

6. The management method of the intelligent identification microscopic device for leukemia cell morphology according to claim 1, characterized in that, Activating the preset region growing algorithm to delineate suspicious areas includes: Using the center point of the suspicious cell initially identified by the AI ​​recognition module as the initial seed point, the sample features collected in the early stage, real-time environmental data and the positioning deviation value of the X / Y axis motor of the stage are called, and the confidence weight of the seed point is set in combination with the initial parameters of the preset region growth algorithm. Based on sample features, the weights are adjusted. In areas with deep staining, the weight of the seed point is increased to a preset ratio, and in areas with high cell density, the neighborhood of the initial seed point is reduced to a preset pixel value. Establish growth criteria that integrate external factors. The basic criteria are set as follows: gray difference between adjacent pixels and seed point ≤ preset baseline threshold, gradient direction consistency ≥ preset percentage. The environmental adaptation criteria are set as follows: when temperature and humidity fluctuations exceed the preset range, the gray difference threshold is tightened to the preset lower limit. The hardware compensation criteria are set as follows: when X / Y axis positioning deviation > preset micrometer value, the growth neighborhood is expanded to the deviation compensation range. The growth process is executed iteratively. After each round of growth, a preset historical database of suspicious areas of similar samples is called to compare the current area with the database. If the deviation exceeds the preset threshold, the growth step size is adjusted in combination with the real-time running current of the X / Y axis motors of the stage. When the current fluctuates greatly, the step size is reduced to the preset value. Growth is terminated when no new pixels meeting the criteria are added for a consecutive preset number of times, or when the area of ​​the region reaches the preset maximum threshold corresponding to the thickness of the sample smear. The delineated area is double-verified. The sharpness of the center of the area is verified by the Z-axis motor focusing sharpness verification. The deviation is corrected by associating imaging data during stable temperature and humidity periods. Then, the boundary coordinates of the suspicious area are extracted by the preset polygon fitting algorithm. The boundary data is associated with and stored with sample features, environment and motor status data.

7. The management method of the intelligent identification microscopic device for leukemia cell morphology according to claim 6, characterized in that, Drive the X / Y axis motors to position the area, switch to the high-magnification objective lens, and trigger Z-axis refocusing to acquire high-resolution images, including: Retrieve the boundary coordinates of the suspicious area, the historical positioning deviation of the X / Y axis of the stage, real-time temperature and humidity data, and sample smear thickness data; Based on the retrieved data, a preset three-dimensional coupling compensation algorithm is adopted. The preset compensation amount is added according to the temperature and humidity deviation from the reference value and the historical deviation exceeding the preset deviation threshold. After correcting the positioning target value, the X / Y axis motor is driven by a preset segmented speed regulation strategy. The position is fed back through the encoder at the same time. The position stops when the deviation is ≤ the preset positioning accuracy threshold. After the positioning is stable, the preset high-magnification objective switching rate is adapted according to the comparison results between the sample smear thickness and the preset smear thickness threshold. During the switching, the motor current is monitored in real time. When the current exceeds the preset current threshold, the process is paused and the objective is finely adjusted until the current drops back down. After the positioning is stable, the preset high-magnification objective dedicated focusing parameter library is called. By combining the real-time temperature and humidity fluctuations with the preset temperature and humidity fluctuation range, the Z-axis focusing step size is corrected, the preset dual-stage focusing algorithm is executed, and the signal is generated simultaneously through the FPGA according to the preset trigger delay to drive the camera to acquire high-definition images according to the preset high-magnification imaging parameters. A preset edge strength-texture clarity multi-feature model is used to judge high-definition images. If the image meets the standard, it is sent to the AI ​​recognition module for secondary recognition. If the image does not meet the standard, the Z-axis step size is readjusted and the focus is repeated for acquisition. Finally, if the image meets the standard, the positioning / focus / environment data is bound and stored with the sample.

8. A management system for an intelligent microscopic device for recognizing leukemia cell morphology, characterized in that, The device includes a memory, a processor, and a program stored in the memory and executable on the processor, which, when loaded and executed by the processor, implements a management method for a leukemia cell morphology intelligent identification microscopic device as described in any one of claims 1 to 7.

Images