A blood smear quality evaluation method, system, device, medium and product

Abstract

The application discloses a blood smear quality evaluation method, system, device, medium and product, and relates to the technical field of medical examination.The method comprises the following steps: for an evaluated blood smear formed by a push-dyeing sheet, image acquisition is performed on a whole blood film area to obtain continuous field images covering the whole range of the blood film, each continuous field image is analyzed and processed, the number of single red blood cells and the number of stacked red blood cells are counted respectively, a red blood cell stacking coefficient matrix is generated, the red blood cell stacking coefficients of each row are extracted along the short edge direction and the coefficient of variation is calculated to generate blood film uniformity; the matrix rows are averagely divided into a head part, a body part and a tail part along the long edge direction, the average values of the rows are calculated to generate blood film head-body-tail thickness difference, and blood film thinness is generated; the blood film uniformity, the blood film thinness and the blood film head-body-tail thickness difference are compared with preset qualified thresholds respectively to generate a smear quality evaluation result of the evaluated blood smear.The application improves the accuracy of the evaluation result.

Description

Technical Field

[0001] This application relates to the field of medical testing technology, and in particular to a method, system, equipment, medium and product for assessing the quality of blood smear preparation. Background Technology

[0002] In clinical hematology testing, the quality of blood smears directly impacts the accuracy of subsequent cell morphology observation and disease diagnosis. A qualified blood smear must meet the core requirements of a uniform blood film, being as thin as a cicada's wing, and exhibiting varying thicknesses across the head, body, and tail regions. However, current industry assessments of blood smear preparation quality primarily rely on the subjective judgment of laboratory personnel. This method is susceptible to human factors such as differences in individual operator experience, operating habits, and visual fatigue, lacking a unified, quantitative, and stable evaluation standard. Due to the poor consistency of existing assessment methods, it is difficult to establish a standardized quality control system, easily leading to misinterpretations of cell morphology or diagnostic errors caused by substandard blood smear preparation quality, severely hindering the standardization and automation of clinical testing. Therefore, a quantitative assessment method that can objectively quantify blood smear preparation quality and eliminate the defects of subjective evaluation is urgently needed. Summary of the Invention

[0003] The purpose of this application is to provide a method, system, equipment, medium, and product for assessing the quality of blood smear preparation, in order to solve the problem that existing assessment methods rely on human experience, are highly subjective, have poor consistency, and result in low accuracy in assessing the quality of blood smear preparation.

[0004] To achieve the above objectives, this application provides the following solution.

[0005] In a first aspect, this application provides a method for evaluating the quality of blood smear preparation, comprising the following steps.

[0006] S1, for the blood smear to be evaluated formed by the smear, image acquisition is performed on the entire blood film area to obtain continuous field-of-view images covering the entire range of the blood film, and a continuous field-of-view image set is generated.

[0007] S2, based on the continuous field-of-view image set, analyze and process each continuous field-of-view image, count the number of individual red blood cells N1 and the number of stacked red blood cells N2 respectively, and calculate the red blood cell stacking coefficient S of each field of view according to the formula S=N2 / (N1+N2), and arrange the fields of view according to their spatial position on the blood film to generate the red blood cell stacking coefficient matrix M.

[0008] S3. Based on the red blood cell stacking coefficient matrix M, extract the red blood cell stacking coefficient of each row along the short side and calculate the coefficient of variation. Filter rows with a coefficient of variation of less than 20% and calculate the percentage of rows to generate blood film homogeneity.

[0009] S4, divide the matrix rows into head, body and tail along the long side and calculate the mean of each row. After comparing with the preset thickness threshold, count the proportion of rows that do not meet the set conditions and generate the blood membrane head-body-tail thickness difference P2. Then, screen the field of view in the red blood cell stacking coefficient matrix M where the red blood cell stacking coefficient S < 40% and calculate the proportion of the field of view to generate the blood membrane thinness P1.

[0010] S5. Based on the blood film uniformity, the blood film thinness P1, and the difference in thickness between the head, body, and tail of the blood film P2, compare them with preset qualified thresholds to generate the slide preparation quality evaluation result of the blood smear to be evaluated.

[0011] Secondly, this application provides a blood smear preparation quality assessment system, including the following modules.

[0012] The continuous field-of-view image set generation module is used to acquire images of the entire blood film area for the blood smear to be evaluated formed by the smear, obtain continuous field-of-view images covering the entire range of the blood film, and generate a continuous field-of-view image set.

[0013] The red blood cell stacking coefficient matrix M generation module is used to analyze and process each continuous field of view image according to the continuous field of view image set, count the number of individual red blood cells N1 and the number of stacked red blood cells N2 respectively, and calculate the red blood cell stacking coefficient S of each field of view according to the formula S=N2 / (N1+N2), and arrange the fields of view according to their spatial positions on the blood film to generate the red blood cell stacking coefficient matrix M.

[0014] The blood film homogeneity generation module is used to extract the red blood cell stacking coefficient of each row along the short side direction according to the red blood cell stacking coefficient matrix M, calculate the coefficient of variation, screen the rows with a coefficient of variation of less than 20% and calculate the proportion of the number of rows, and generate blood film homogeneity.

[0015] The blood membrane thickness P1 and blood membrane head-body-tail thickness difference P2 generation module is used to divide the matrix rows into head, body and tail sections along the long side and calculate the mean of each row. After comparing with the preset thickness threshold, the module counts the proportion of rows that do not meet the set conditions and generates the blood membrane head-body-tail thickness difference P2. It also filters the field of view in the red blood cell stacking coefficient matrix M where the red blood cell stacking coefficient S < 40% and calculates the proportion of the field of view to generate the blood membrane thickness P1.

[0016] The evaluation module is used to compare the blood film uniformity, the blood film thinness P1, and the blood film head-body-tail thickness difference P2 with preset qualified thresholds to generate the slide preparation quality evaluation result of the blood smear to be evaluated.

[0017] Thirdly, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-described method for evaluating the quality of blood smear preparation.

[0018] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method for evaluating the quality of blood smear preparation.

[0019] Fifthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method for evaluating the quality of blood smear preparation.

[0020] According to the specific embodiments provided in this application, this application has the following technical effects: For blood smears to be evaluated formed by smearing, this application acquires images of the entire blood film area, obtaining continuous field-of-view images covering the entire blood film area, generating a continuous field-of-view image set, and analyzing and processing each continuous field-of-view image. The number of individual red blood cells (N1) and the number of stacked red blood cells (N2) are counted respectively, and the red blood cell stacking coefficient S of each field of view is calculated according to the formula S=N2 / (N1+N2). The fields of view are arranged according to their spatial position on the blood film to generate a red blood cell stacking coefficient matrix M. This application accurately identifies and counts the number of individual and stacked red blood cells to calculate the red blood cell stacking coefficient of each field of view, and then constructs a whole-blood film red blood cell stacking coefficient matrix according to its spatial position. This completely abandons the traditional evaluation method that relies on manual visual observation and subjective experience judgment, realizing the objectification and digitization of quality parameters from the source of data acquisition and feature extraction. Based on this, this application extracts the red blood cell stacking coefficient of each row along the short side direction according to the red blood cell stacking coefficient matrix M and calculates the coefficient of variation. Rows with a coefficient of variation of less than 20% are selected and the percentage of rows is calculated, generating a uniform blood film. The matrix is ​​divided into head, body, and tail sections along its long side, and the mean of each section is calculated. After comparing with a preset thickness threshold, the percentage of rows that do not meet the set conditions is counted, generating the head-body-tail thickness difference P2. Fields of view with a red blood cell stacking coefficient S < 40% in the red blood cell stacking coefficient matrix M are selected, and their percentage is calculated, generating the blood film thinness P1. Based on the blood film homogeneity, the blood film thinness P1, and the head-body-tail thickness difference P2, they are compared with preset acceptable thresholds to generate the slide preparation quality assessment result for the blood smear to be evaluated. This technical solution will... The abstract quality requirements of "uniform blood film, thinness, and head-body-tail variation" are transformed into calculable and traceable quantitative data indicators. This effectively eliminates the evaluation fluctuations caused by differences in personnel experience, operating habits, and visual fatigue, significantly improving the consistency, stability, and reliability of blood smear quality assessment. This enhances the accuracy of assessment results, establishes a standardized quality control basis for clinical testing, and fundamentally avoids misjudgment of cell morphology and diagnostic deviations caused by substandard slide preparation quality. At the same time, it provides solid technical support for seamless adaptation to fully automated testing equipment and high-throughput automated quality control. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a flowchart illustrating a method for evaluating the quality of blood smear preparation, provided as an embodiment of this application. Detailed Implementation

[0023] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0024] To make the objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0025] like Figure 1 As shown in the embodiment of this application, a method for evaluating the quality of blood smear preparation is provided, including the following steps.

[0026] S1, for the blood smear to be evaluated formed by the smear, image acquisition is performed on the entire blood film area to obtain continuous field-of-view images covering the entire range of the blood film, and a continuous field-of-view image set is generated.

[0027] S2, based on the continuous field-of-view image set, analyze and process each continuous field-of-view image, count the number of individual red blood cells N1 and the number of stacked red blood cells N2 respectively, and calculate the red blood cell stacking coefficient S of each field of view according to the formula S=N2 / (N1+N2), and arrange the fields of view according to their spatial position on the blood film to generate the red blood cell stacking coefficient matrix M.

[0028] S3. Based on the red blood cell stacking coefficient matrix M, extract the red blood cell stacking coefficient of each row along the short side and calculate the coefficient of variation. Filter rows with a coefficient of variation of less than 20% and calculate the percentage of rows to generate blood film homogeneity.

[0029] S4, divide the matrix rows into head, body and tail along the long side and calculate the mean of each row. After comparing with the preset thickness threshold, count the proportion of rows that do not meet the set conditions and generate the blood membrane head-body-tail thickness difference P2. Then, screen the field of view in the red blood cell stacking coefficient matrix M where the red blood cell stacking coefficient S < 40% and calculate the proportion of the field of view to generate the blood membrane thinness P1.

[0030] S5. Based on the blood film uniformity, the blood film thinness P1, and the difference in thickness between the head, body, and tail of the blood film P2, compare them with preset qualified thresholds to generate the slide preparation quality evaluation result of the blood smear to be evaluated.

[0031] This application overcomes the shortcomings of existing blood smear quality assessment methods, which are subjective and lack quantitative standards, and achieves quantitative, stable and reliable assessment of blood smear quality, providing a standardized quality basis for clinical testing.

[0032] In one exemplary embodiment, S1 specifically includes the following steps.

[0033] S11. For the blood smear to be evaluated formed by the smear, an optical microscope is used at 100x oil immersion magnification to acquire images of the entire blood film area of ​​the blood smear, obtaining continuous field-of-view images covering the entire range of the blood film to ensure no field of view is missed.

[0034] In an exemplary embodiment, based on a deep learning computer vision algorithm, each acquired field-of-view image is analyzed and processed. Specifically, step S2 includes: S21, Gaussian filtering, image equalization, and data enhancement are performed on the continuous field-of-view image set to generate a preprocessed field-of-view image set.

[0035] S22, based on the preprocessed field image set, and using the YOLOv7 basic network, optimize the feature fusion structure for small cell targets and stacked occlusion characteristics, re-adapt cell-scale anchor boxes through clustering algorithm, and use transfer learning combined with weighted loss function to complete model iterative training and parameter optimization to generate a red blood cell localization and recognition model.

[0036] S23, target detection is performed on the preprocessed field-of-view image set according to the red blood cell localization and recognition model, and the number of single red blood cells N1 and the number of stacked red blood cells N2 are identified and output statistically, wherein the stacked red blood cells are two or more overlapping red blood cell groups.

[0037] S24. According to the formula S=N2 / (N1+N2), the red blood cell stacking coefficient S of each field of view is calculated, and the matrix is ​​generated by matrix mapping according to the row and column space coordinates at the time of acquisition.

[0038] In practical applications, normal red blood cells and stacked red blood cells are labeled by professionals. Image preprocessing is completed by combining Gaussian filtering for noise reduction, image equalization, and data augmentation. Based on the YOLOv7 basic network, the feature fusion structure is optimized for small cell targets and stacked occlusion characteristics. Cell-scale anchor boxes are re-adapted through clustering algorithms. The dataset is divided proportionally, and transfer learning combined with a weighted loss function is used to complete iterative training and parameter optimization of the model, ultimately obtaining the red blood cell localization and recognition model.

[0039] Using the red blood cell localization and recognition model, the number of single red blood cells (denoted as N1) and the number of stacked red blood cells (denoted as N2, where "stacked red blood cells" refers to a group of two or more red blood cells that overlap each other) are identified and counted.

[0040] The red blood cell stacking coefficient (denoted as S) of a single field of view is calculated using the formula: S=N2 / (N1+N2), which is the proportion of stacked red blood cells to the total number of red blood cells in the field of view.

[0041] Traverse all acquisition fields and arrange the red blood cell stacking coefficients of each field according to the spatial position of the field on the blood film to form a red blood cell stacking coefficient matrix of the whole blood film (denoted as M, the matrix dimension is the same as the number of rows and columns of the acquisition field).

[0042] In one exemplary embodiment, S3 specifically includes the following steps.

[0043] S31, taking the dimension perpendicular to the extension direction of the head, body and tail of the blood membrane as the short side direction, extract the red blood cell stacking coefficients of all rows from the red blood cell stacking coefficient matrix M row by row to generate the stacking coefficient sequence of each row.

[0044] S32, calculate the standard deviation σ and mean μ of the stacked coefficient sequence for each row, and calculate the coefficient of variation CV for the corresponding row according to the formula CV=σ / μ.

[0045] S33, based on the coefficient of variation CV, traverse all rows of the red blood cell stacking coefficient matrix M, filter rows where the coefficient of variation CV is less than 20%, count the number of filtered rows and calculate the proportion of the number of filtered rows to the total number of rows in the short side direction, and generate the blood film homogeneity.

[0046] In an exemplary embodiment, step S4 involves screening fields of view where the red blood cell stacking coefficient S is less than 40% in the red blood cell stacking coefficient matrix M and calculating the field of view percentage to generate the blood film thinness P1. This step specifically includes the following steps.

[0047] S41, traverse all red blood cell stacking coefficients in the red blood cell stacking coefficient matrix M, compare the red blood cell stacking coefficient S of each field of view with the thinness determination threshold one by one, and generate a set of thinness compliance status identifiers.

[0048] S42, based on the set of thinness compliance status identifiers, count the number of fields of view that meet the red blood cell stacking coefficient S < 40%, calculate the proportion of the number of fields of view that meet the red blood cell stacking coefficient S < 40% to the total number of fields of view of the whole blood membrane, and generate the blood membrane thinness P1.

[0049] In an exemplary embodiment, in step S4, the matrix rows are divided into head, body and tail along the long side and the mean of each row is calculated. After comparing with a preset thickness threshold, the percentage of rows that do not meet the set conditions is counted to generate the thickness difference P2 of the blood membrane head-body-tail. This step specifically includes the following steps.

[0050] S43, taking the extension direction from the initial end of blood application to the end of application as the long side direction, divide all rows in the red blood cell stacking coefficient matrix M along the long side direction into three segments: head, body and tail, to generate a set of head-body-tail partitioned rows.

[0051] S44. Based on the set of rows in the head, body and tail partitions, calculate the row average of the red blood cell stacking coefficient for each row in the head, body and tail, compare the row average with the preset thickness threshold of the corresponding segment, and generate a thickness compliance status identifier for each row.

[0052] S45, based on the thickness compliance status indicators of each row, count the number of rows that do not meet the conditions of head row average value > 70%, body row average value > 30% and < 70%, and tail row average value < 30%, calculate the proportion of the number of rows that do not meet the above conditions to the total number of rows on the long side, and generate the blood membrane head-body-tail thickness difference P2.

[0053] In practical applications, based on the core quality requirements of blood smears being "uniform, thin, and showing head-to-tail differences," the following three key evaluation indicators are calculated using statistical methods based on the stacking coefficient matrix M.

[0054] Indicator 1: Blood film homogeneity.

[0055] Using the short side direction of the blood membrane as the analysis dimension (i.e., the dimension perpendicular to the "head-body-tail" extension direction of the blood membrane), the stacking coefficients of all rows in the short side direction of the stacking coefficient matrix M are extracted, and the coefficient of variation (CV) of the data set is calculated. The proportion of rows with CV less than 20% is considered as the blood membrane homogeneity; the greater the blood membrane homogeneity, the better the thickness uniformity of the blood membrane in the short side direction.

[0056] ,in, is the standard deviation, and μ is the mean.

[0057] Indicator 2: Blood film thickness (denoted as P1).

[0058] The stacking coefficient S of all fields of view in the stacking coefficient matrix M is statistically analyzed. The proportion of fields of view with "stacking coefficient S < 40%" to the total number of fields of view is calculated, which is P1. The larger P1 is, the thinner the blood membrane is as a whole, which meets the requirement of "thin as a cicada's wing".

[0059] Indicator 3: Difference in thickness of the blood film head, body and tail (denoted as P2).

[0060] Using the long side of the blood membrane as the "head-body-tail" extension direction (the head is the initial end of blood application and the tail is the final end of application), all rows in the long side of the stacking coefficient matrix M are divided into three equal segments, corresponding to the "head" (first third row), "body" (middle third row), and "tail" (last third row) of the blood membrane.

[0061] The stacking coefficient of all rows in the header must be greater than 70%, the stacking coefficient of all rows in the body must be greater than 30% and less than 70%, and the stacking coefficient of all rows in the tail must be less than 30%.

[0062] The proportion of rows that do not meet the above requirements to the total number of rows on the long side is P2; the smaller P2 is, the more the difference in thickness between the head, body and tail of the blood membrane meets the standard.

[0063] In one exemplary embodiment, S5 specifically includes the following steps.

[0064] S51, receive the blood film uniformity, the blood film thinness P1, and the blood film head-body-tail thickness difference P2, and compare the blood film uniformity with a first threshold, the blood film thinness P1 with a second threshold, and the blood film head-body-tail thickness difference P2 with a third threshold, respectively, to generate the compliance comparison results of the three indicators; the first threshold is 0.8, the second threshold is 0.5, and the third threshold is 0.3.

[0065] S52, based on the comparison results, determine whether the blood film uniformity > 0.8, the blood film thinness P1 > 0.5, and the blood film head-body-tail thickness difference P2 < 0.3 are all met simultaneously. If they are all met simultaneously, generate an evaluation result indicating that the slide preparation quality is qualified.

[0066] S53, if any item in the comparison results does not meet the corresponding numerical comparison conditions, an evaluation result of unqualified film production quality is generated.

[0067] In practical applications, the blood smear is considered to be of qualified quality when the three key indicators of the blood smear to be evaluated meet the following conditions simultaneously.

[0068] 1. Blood film homogeneity > 0.8 (blood film homogeneity meets the standard).

[0069] 2. P1 > 0.5 (blood film thinness meets the standard).

[0070] 3. P2 < 0.3 (the difference in thickness of the blood film head, body and tail meets the standard).

[0071] If any one of the indicators is not met, the quality is deemed unqualified.

[0072] The technical solution of this application is further illustrated below with specific embodiments.

[0073] Example 1 Sample preparation: Select one blood smear sample to be evaluated, prepared from a smear slide in clinical testing, ensuring that the blood smear is undamaged and evenly stained.

[0074] Image acquisition: Using a fully automated digital pathology slide scanner equipped with a 100x oil immersion lens, the entire blood membrane area of ​​the blood smear was scanned and acquired: the field resolution was set to 2048×2048 pixels, and the images were continuously acquired in a row-column order, for a total of 34,200 fields (180 rows in the long side and 190 columns in the short side), forming a field image set covering the entire blood membrane.

[0075] Stacking coefficient calculation: A deep learning image analysis model based on ResNet50 is called to process 34,200 field-view images one by one.

[0076] Identify the number of individual red blood cells N1 and the number of stacked red blood cells N2, and calculate the stacking coefficient S for each field of view.

[0077] Arrange the 34,200 S values ​​into a 180×190 stacked coefficient matrix M according to the positional order of "180 rows on the long side and 190 columns on the short side".

[0078] Evaluation index calculation: Blood film homogeneity calculation: Extract all S values ​​(180 in total) from the 190 columns of the short side of matrix M, calculate their coefficient of variation and statistically obtain the blood film homogeneity as 0.85 (>0.8).

[0079] Calculate P1: The number of fields of view with S < 40% in the statistical matrix M is 20520, which accounts for 0.6 (> 0.5) of the total number of fields of view.

[0080] Calculate P2: Divide the 180 rows on the long side into 60 rows for the head, 60 rows for the body, and 60 rows for the tail. Check the mean of each row: the mean of the head is >70%, the mean of the body is between 30% and 70%, and the mean of the tail is <30%. The number of rows that do not meet the requirements is 0, so P2 = 0 (<0.3).

[0081] Quality assessment: The blood film uniformity, P1, and P2 of this blood smear all meet the qualified standards, and the slide preparation quality is deemed qualified.

[0082] Example 2 Sample preparation: Select two blood smear samples to be evaluated (there is a problem of excessive local thickness of blood film).

[0083] Image acquisition and coefficient calculation: Same as in Example 1, 30,000 fields of view are acquired and a 150×200 stacked coefficient matrix M is formed.

[0084] Evaluation index calculation: Blood film homogeneity: calculated to be 0.28, <0.8; P1: the proportion of visual field with S <40% is 0.45, <0.5; P2: the number of rows that do not meet the head-body-tail mean requirement is 45, accounting for 0.27 of the total number of rows, <0.3.

[0085] Quality assessment: The blood film uniformity and P1 of this blood smear do not meet the qualified standards, and the slide preparation quality is deemed unqualified.

[0086] This application also provides a blood smear preparation quality assessment system, including the following modules.

[0087] The continuous field-of-view image set generation module is used to acquire images of the entire blood film area for the blood smear to be evaluated formed by the smear, obtain continuous field-of-view images covering the entire range of the blood film, and generate a continuous field-of-view image set.

[0088] The red blood cell stacking coefficient matrix M generation module is used to analyze and process each continuous field of view image according to the continuous field of view image set, count the number of individual red blood cells N1 and the number of stacked red blood cells N2 respectively, and calculate the red blood cell stacking coefficient S of each field of view according to the formula S=N2 / (N1+N2), and arrange the fields of view according to their spatial positions on the blood film to generate the red blood cell stacking coefficient matrix M.

[0089] The blood film homogeneity generation module is used to extract the red blood cell stacking coefficient of each row along the short side direction according to the red blood cell stacking coefficient matrix M, calculate the coefficient of variation, screen the rows with a coefficient of variation of less than 20% and calculate the proportion of the number of rows, and generate blood film homogeneity.

[0090] The blood membrane thickness P1 and blood membrane head-body-tail thickness difference P2 generation module is used to divide the matrix rows into head, body and tail sections along the long side and calculate the mean of each row. After comparing with the preset thickness threshold, the module counts the proportion of rows that do not meet the set conditions and generates the blood membrane head-body-tail thickness difference P2. It also filters the field of view in the red blood cell stacking coefficient matrix M where the red blood cell stacking coefficient S < 40% and calculates the proportion of the field of view to generate the blood membrane thickness P1.

[0091] The evaluation module is used to compare the blood film uniformity, the blood film thinness P1, and the difference in thickness between the head, body, and tail of the blood film P2 with preset qualified thresholds to generate the slide preparation quality evaluation result of the blood smear to be evaluated.

[0092] This application achieves quantitative evaluation of blood smear preparation quality through a process of "whole blood membrane image acquisition - red blood cell stacking coefficient calculation - multi-dimensional statistical index evaluation - standardized quality judgment": avoiding the subjectivity of manual evaluation, relying on objective data for evaluation results, and significantly improving consistency and reliability.

[0093] The quantitative standards for "uniform blood film, thinness, and head-body-tail difference" were clearly defined, providing a standardized basis for quality control in clinical testing.

[0094] Based on deep learning computer vision algorithms, it can be adapted to fully automated inspection equipment to automate the evaluation process and improve inspection efficiency.

[0095] In an exemplary embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments. The computer device can be a server or a terminal. The computer device includes a processor, a memory, an input / output interface (I / O), and a communication interface. The processor, memory, and I / O interface are connected via a system bus, and the communication interface is connected to the system bus via the I / O interface. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The database of the computer device stores data to be processed. The I / O interface of the computer device is used for exchanging information between the processor and external devices. The communication interface of the computer device is used for communicating with an external terminal via a network connection. When the computer program is executed by the processor, it implements the above-described methods.

[0096] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0097] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0098] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0099] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by hardware related to computer program instructions. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0100] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0101] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0102] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for evaluating the quality of blood smear preparation, characterized in that, include: S1, For the blood smear to be evaluated formed by the smear, the entire blood film area is image acquired to obtain continuous field-of-view images covering the entire range of the blood film and generate a continuous field-of-view image set; S2, based on the continuous field-of-view image set, analyze and process each continuous field-of-view image, count the number of individual red blood cells N1 and the number of stacked red blood cells N2 respectively, and calculate the red blood cell stacking coefficient S of each field of view according to the formula S=N2 / (N1+N2), and arrange the fields of view according to their spatial position on the blood film to generate the red blood cell stacking coefficient matrix M. S3, Based on the red blood cell stacking coefficient matrix M, extract the red blood cell stacking coefficient of each row along the short side and calculate the coefficient of variation, screen rows with a coefficient of variation of less than 20% and calculate the proportion of the number of rows to generate blood film homogeneity. S4, divide the matrix rows into head, body and tail along the long side and calculate the mean of each row. After comparing with the preset thickness threshold, count the proportion of rows that do not meet the set conditions and generate the blood film head-body-tail thickness difference P2. Then, screen the field of view in the red blood cell stacking coefficient matrix M where the red blood cell stacking coefficient S < 40% and calculate the proportion of the field of view to generate the blood film thinness P1. In step S4, the matrix rows are divided into head, body, and tail sections along the long side, and the mean of each row is calculated. After comparing with a preset thickness threshold, the percentage of rows that do not meet the set conditions is counted, generating the blood membrane head-body-tail thickness difference P2, which specifically includes: S43, taking the extension direction from the initial end of blood application to the end of application as the long side direction, divide all rows in the red blood cell stacking coefficient matrix M along the long side direction into three segments: head, body and tail, to generate a set of head-body-tail partitioned rows. S44, based on the set of rows in the head, body and tail partitions, calculate the row average of the red blood cell stacking coefficient for each row in the head, body and tail, compare the row average with the preset thickness threshold of the corresponding segment, and generate a thickness compliance status identifier for each row. S45, based on the thickness compliance status indicators of each row, count the number of rows that do not meet the conditions of head row average value > 70%, body row average value > 30% and < 70%, and tail row average value < 30%, calculate the proportion of the number of rows that do not meet the above conditions to the total number of rows on the long side, and generate the blood membrane head-body-tail thickness difference P2. S5. Based on the blood film uniformity, the blood film thinness P1 and the blood film head-body-tail thickness difference P2, compare them with the preset qualified thresholds respectively to generate the slide preparation quality evaluation result of the blood smear to be evaluated. S5 specifically includes: S51, receive the blood film uniformity, the blood film thinness P1, and the blood film head-body-tail thickness difference P2, and compare the blood film uniformity with a first threshold, the blood film thinness P1 with a second threshold, and the blood film head-body-tail thickness difference P2 with a third threshold, respectively, to generate the compliance comparison results of the three indicators; the first threshold is 0.8, the second threshold is 0.5, and the third threshold is 0.3; S52, based on the comparison results, determine whether the blood film uniformity > 0.8, the blood film thinness P1 > 0.5, and the blood film head-body-tail thickness difference P2 < 0.3 are simultaneously met. If they are simultaneously met, generate an evaluation result that the slide preparation quality is qualified. S53, if any item in the compliance comparison results does not meet the corresponding numerical comparison conditions, an evaluation result of unqualified film production quality is generated. The abstract quality requirements of "uniform and thin blood film with head-body-tail differences" are transformed into calculable and traceable quantitative data indicators.

2. The method for evaluating the quality of blood smear preparation according to claim 1, characterized in that, S2 specifically includes: S21, Gaussian filtering for noise reduction, image equalization, and data enhancement are performed on the continuous field-of-view image set to generate a preprocessed field-of-view image set; S22, based on the preprocessed field image set, and using the YOLOv7 basic network, optimize the feature fusion structure for small cell targets and stacked occlusion characteristics, re-adapt cell-scale anchor boxes through clustering algorithm, and use transfer learning combined with weighted loss function to complete model iterative training and parameter optimization to generate a red blood cell localization and recognition model. S23, target detection is performed on the preprocessed field-of-view image set according to the red blood cell localization and recognition model, and the number of single red blood cells N1 and the number of stacked red blood cells N2 are identified and output statistically, wherein the stacked red blood cells are two or more overlapping red blood cell groups; S24. According to the formula S=N2 / (N1+N2), the red blood cell stacking coefficient S of each field of view is calculated, and the matrix is ​​mapped according to the row and column space coordinates at the time of acquisition to generate the red blood cell stacking coefficient matrix M.

3. The method for evaluating the quality of blood smear preparation according to claim 1, characterized in that, S3 specifically includes: S31, taking the dimension perpendicular to the extension direction of the head, body and tail of the blood membrane as the short side direction, extract the red blood cell stacking coefficients of all rows from the red blood cell stacking coefficient matrix M row by row to generate the stacking coefficient sequence of each row. S32, calculate the standard deviation σ and mean μ of the stacked coefficient sequence for each row, and calculate the coefficient of variation CV for the corresponding row according to the formula CV=σ / μ; S33, based on the coefficient of variation CV, traverse all rows of the red blood cell stacking coefficient matrix M, filter rows where the coefficient of variation CV is less than 20%, count the number of filtered rows and calculate the proportion of the number of filtered rows to the total number of rows in the short side direction, and generate the blood film homogeneity.

4. The method for evaluating the quality of blood smear preparation according to claim 1, characterized in that, In step S4, the field of view in which the red blood cell stacking coefficient S is less than 40% in the red blood cell stacking coefficient matrix M is selected and the field of view percentage is calculated to generate the blood film thinness P1, specifically including: S41, traverse all red blood cell stacking coefficients in the red blood cell stacking coefficient matrix M, compare the red blood cell stacking coefficient S of each field of view with the thinness determination threshold one by one, and generate a thinness compliance status identifier set. S42, based on the set of thinness compliance status identifiers, count the number of fields of view that meet the red blood cell stacking coefficient S < 40%, calculate the proportion of the number of fields of view that meet the red blood cell stacking coefficient S < 40% to the total number of fields of view of the whole blood membrane, and generate the blood membrane thinness P1.

5. A blood smear preparation quality assessment system, characterized in that, The method for evaluating the quality of blood smear preparation according to any one of claims 1-4 includes: The continuous field-of-view image set generation module is used to acquire images of the entire blood film area for the blood smear to be evaluated formed by the smear, obtain continuous field-of-view images covering the entire range of the blood film, and generate a continuous field-of-view image set. The red blood cell stacking coefficient matrix M generation module is used to analyze and process each continuous field of view image according to the continuous field of view image set, count the number of individual red blood cells N1 and the number of stacked red blood cells N2 respectively, and calculate the red blood cell stacking coefficient S of each field of view according to the formula S=N2 / (N1+N2), and generate the red blood cell stacking coefficient matrix M according to the spatial position of the field of view on the blood film. The blood film homogeneity generation module is used to extract the red blood cell stacking coefficient of each row along the short side direction according to the red blood cell stacking coefficient matrix M and calculate the coefficient of variation, filter the rows with a coefficient of variation of less than 20% and calculate the proportion of the number of rows, and generate blood film homogeneity. The blood membrane thickness P1 and blood membrane head-body-tail thickness difference P2 generation module is used to divide the matrix rows into head, body and tail along the long side and calculate the mean of each row. After comparing with the preset thickness threshold, the proportion of rows that do not meet the set conditions is counted to generate blood membrane head-body-tail thickness difference P2. The module also filters the field of view in the red blood cell stacking coefficient matrix M where the red blood cell stacking coefficient S < 40% and calculates the proportion of the field of view to generate blood membrane thickness P1. The evaluation module is used to compare the blood film uniformity, the blood film thinness P1, and the blood film head-body-tail thickness difference P2 with preset qualified thresholds to generate the slide preparation quality evaluation result of the blood smear to be evaluated.

6. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the blood smear preparation quality assessment method according to any one of claims 1-4.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the blood smear preparation quality assessment method according to any one of claims 1-4.

8. A computer program product, comprising a computer program, characterized in that, When executed by a processor, the computer program implements the blood smear preparation quality assessment method according to any one of claims 1-4.

Images