Quality control method and equipment for pathological images

Abstract

The present invention discloses a quality control method and device for pathological images, including: collecting pathological images within any time period; based on a number of original questions, performing quality inspection on each collected pathological image to determine whether the image is a qualified pathological image; based on the entire process analysis of pathological image preparation, classifying a number of original questions, generating a number of classification questions containing corresponding original questions, wherein the classification questions are problems existing in any step of the pathological image preparation process; based on the original questions contained in each classification question, performing in-depth analysis on each unqualified pathological image to determine whether the image has the classification problem; counting the original questions and classification questions existing in all pathological images, and performing trend analysis to obtain analysis results. The present invention can combine the preparation process of pathological images, find the source of the problem, match and classify the original questions, and control the quality of pathological images.

Description

Technical Field

[0001] The present invention belongs to the field of medical image processing and data analysis, and specifically relates to a quality control method and device for pathological images. Background Art

[0002] In pathology, pathological images are very important resources for the examination, diagnosis and formulation of treatment plans for clinical cases. In order to obtain more accurate and reliable pathological images, hospitals generally use digital imaging technology, using digital scanners to digitize pathological specimens, and store, manage and analyze them. However, these digital images still have some problems, such as high image complexity, inconsistent quality, and the presence of noise, which may have a certain impact on the diagnosis and treatment of pathologists.

[0003] Traditional pathology image processing methods mainly rely on manual inspection and processing, which requires a lot of time and effort, and is prone to human errors and missed detections. In addition, traditional processing methods cannot meet the diverse and large-scale image processing needs, and doctors lack effective automated guidance for abnormal interference factors. Therefore, developing an intelligent pathology image processing system is the key to improving the quality and accuracy of pathology images and improving the work efficiency and accuracy of pathologists.

[0004] Traditional pathological image processing methods mainly improve the quality and accuracy of pathological images by removing noise, interference and other factors in the image, and restoring parts such as handwriting. However, these practices ignore the important guiding information contained in these factors for pathological image production and scanning. Summary of the invention

[0005] In order to solve the above technical problems, the present invention proposes a quality control method and device for pathological images.

[0006] In order to achieve the above object, the technical solution of the present invention is as follows:

[0007] In one aspect, the present invention discloses a method for quality control of pathological images, comprising:

[0008] Step S1: collecting pathological images within any time period;

[0009] Step S2: Based on a number of original questions, quality detection is performed on each collected pathological image to determine whether the image is a qualified pathological image;

[0010] Step S3: Based on the analysis of the entire process of pathological image preparation, a number of original questions are classified to generate a number of classification questions containing the corresponding original questions, wherein the classification questions are questions existing in any step of the pathological image preparation process;

[0011] Step S4: Based on the original problem contained in each classification problem, perform in-depth analysis on each unqualified pathological image to determine whether the image has the classification problem;

[0012] Step S5: Count the original problems and classification problems existing in all pathological images, perform trend analysis, and obtain analysis results.

[0013] Based on the above technical solution, the following improvements can be made:

[0014] As a preferred solution, step S1 specifically includes:

[0015] Step S1.1: collecting pathological images within any time period;

[0016] Step S1.2: Arrange the collected pathological images.

[0017] As a preferred solution, step S2 specifically includes:

[0018] Step S2.1: According to each original question, score each collected pathological image to obtain the original score S of the i-th original question. i , i=1,2,...n;

[0019] Step S2.2: Combining the original scores of each original question, the total quality score S of the pathological image is obtained by the following formula: 总 ;

[0020] ;

[0021] Where: W i is the preset weight corresponding to the i-th original question;

[0022] Step S2.3: Determine the total quality score S of the pathological image 总 Is it less than the quality passing score S? 及格 ;

[0023] If so, the image is an unqualified pathological image;

[0024] Otherwise, the image is a qualified pathological image.

[0025] As a preferred solution, step S4 specifically includes:

[0026] Step S4.1: For each classification problem, the classification score E of the j-th classification problem of the pathological image is obtained by the following formula: j , j = 1, 2, ... m;

[0027] ;

[0028] Among them: Jj is the set of original questions contained in the j-th classification problem;

[0029] Step S4.2: For each classification problem, the classification score E of the j-th classification problem of the pathological image is obtained by the following formula: j-满分 And its corresponding classification passing score E j-及格 , j = 1, 2, ... m;

[0030] ;

[0031] ;

[0032] Where: R j is the preset passing rate of the j-th classification problem;

[0033] Step S4.3: Determine the classification score E of the jth classification problem of the pathological image j Is it less than the passing score E of the corresponding category? j-及格 ;

[0034] If yes, then the image has this classification problem;

[0035] Otherwise, the image does not have this classification problem.

[0036] As a preferred solution, the quality control method also includes:

[0037] Step S6: Feedback the original problems, classification problems and analysis results of the pathological images to the pathology department.

[0038] In another aspect, the present invention discloses a quality control device for pathological images, comprising:

[0039] An acquisition module, used to acquire pathological images within any time period;

[0040] A first problem detection module is used to perform quality detection on each collected pathological image based on a number of original questions to determine whether the image is a qualified pathological image;

[0041] A problem classification module is used to classify a number of original problems based on the analysis of the entire process of pathological image preparation, and generate a number of classification problems containing the corresponding original problems, wherein the classification problems are problems existing in any step of the pathological image preparation process;

[0042] A second problem detection module is used to perform in-depth analysis on each unqualified pathological image based on the original problem contained in each classification problem to determine whether the image has the classification problem;

[0043] The statistical analysis module is used to count the original problems and classification problems existing in all pathological images, and perform trend analysis to obtain analysis results.

[0044] As a preferred solution, the acquisition module specifically includes:

[0045] An acquisition unit, used for acquiring pathological images within any time period;

[0046] The sorting unit is used to sort the collected pathological images.

[0047] As a preferred solution, the first problem detection module specifically includes:

[0048] The original score acquisition unit is used to score each collected pathological image according to each original question to obtain the original score S of the i-th original question. i , i=1,2,...n;

[0049] The quality total score acquisition unit is used to integrate the original scores of each original question and obtain the total quality score S of the pathological image through the following formula: 总 ;

[0050] ;

[0051] Where: W i is the preset weight corresponding to the i-th original question;

[0052] The first judgment unit is used to judge the total quality score S of the pathological image 总 Is it less than the quality passing score S? 及格 ;

[0053] If so, the image is an unqualified pathological image;

[0054] Otherwise, the image is a qualified pathological image.

[0055] As a preferred solution, the second problem detection module includes:

[0056] The classification score acquisition unit is used to obtain the classification score E of the j-th classification problem of the pathological image for each classification problem by the following formula: j , j = 1, 2, ... m;

[0057] ;

[0058] Among them: J j is the set of original questions contained in the j-th classification problem;

[0059] The classification passing score obtaining unit is used to obtain the classification full score E of the j-th classification problem of the pathological image for each classification problem by the following formula: j-满分And its corresponding classification passing score E j-及格 , j = 1, 2, ... m;

[0060] ;

[0061] ;

[0062] Where: R j is the preset passing rate of the j-th classification problem;

[0063] The second judgment unit is used to judge the classification score of the jth classification problem of the pathological image. j Is it less than the passing score E of the corresponding category? j-及格 ;

[0064] If yes, then the image has this classification problem;

[0065] Otherwise, the image does not have this classification problem.

[0066] As a preferred solution, the quality control equipment also includes:

[0067] The feedback module is used to feed back the original problems, classification problems and analysis results of pathological images to the pathology department.

[0068] The present invention discloses a quality control method and device for pathological images, which can find the source of problems in combination with the preparation process of pathological images, match and classify the original problems, and control the quality of pathological images, and has the following beneficial effects:

[0069] First, the present invention adopts machine learning algorithms and image processing technology, which can automatically and quickly detect pathological images, and quantitatively analyze and score different original questions, avoiding errors and missed detections in the manual processing process, and improving the efficiency and reliability of pathological image detection.

[0070] Second, by counting the original problems and classification problems in all pathological images and conducting trend analysis, we can discover the commonalities and characteristics of common problems, summarize the causes, and provide data support and reference basis for the subsequent production of high-quality pathological images.

[0071] Third, based on the trend of pathology image problems, analyze the causes and solutions to help pathologists check and discover potential problems in pathology image production and scanning, and correct them, thereby improving the quality and accuracy of pathology images. Provide feedback and guidance to the pathology department.

[0072] Fourth, real-time monitoring of pathological image quality helps the pathology department improve work efficiency and diagnostic accuracy, while also providing medical institutions with safer and more reliable medical services.

[0073] The present invention can accurately and comprehensively detect and identify original problems existing in pathological images, score different original problems according to their impact on image quality, classify classification problems (i.e., step sources) corresponding to different original problems, statistically analyze problem trends in pathological images, and provide solutions to improve the process of pathology department in producing pathological images, ultimately producing pathological images with higher quality and precision, so that pathologists can make more accurate and reliable pathological diagnoses. This invention is extremely practical, reliable, and applicable. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0075] Figure 1 A flow chart of a quality control method provided by an embodiment of the present invention.

[0076] Figure 2 A process diagram of a quality control method provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0077] The preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0078] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0079] Using ordinal numbers “first,” “second,” “third,” etc. to describe common objects merely indicates that different instances of similar objects are involved and is not intended to imply that the objects so described must have a given order in time, space, order, or in any other manner.

[0080] In addition, the expression of “comprising” an element is an “open” expression, which merely means that corresponding components or steps exist, and should not be interpreted as excluding additional components or steps.

[0081] In order to achieve the purpose of the present invention, a method and device for quality control of pathological images are provided in some embodiments, such as Figure 1 As shown, quality control methods include:

[0082] Step S1: Collect all pathological images within one month;

[0083] Step S2: Based on 9 original problems, quality inspection is performed on each collected pathological image to determine whether the image is a qualified pathological image. The 9 original problems specifically include: staining problem, overlap problem, break problem, wrinkle problem, blur problem, tissue ratio problem, cover glass edge problem, bubble problem, and handwriting problem;

[0084] Step S3: Based on the analysis of the entire process of pathological image preparation, the 9 original questions are classified to generate 3 classification questions containing the corresponding original questions. The classification questions are problems existing in any step of the pathological image preparation process. The 3 classification questions specifically include: slicing problems, scanning problems, and slide making problems;

[0085] Step S4: Based on the original problem contained in each classification problem, perform in-depth analysis on each unqualified pathological image to determine whether the image has the classification problem;

[0086] Step S5: Count the original problems and classification problems existing in all pathological images, perform trend analysis, and obtain analysis results.

[0087] like Figure 2 As shown, in order to facilitate the understanding of the present invention, each step is described in detail below.

[0088] Step S1 specifically includes:

[0089] Step S1.1: Collect pathological images within one month of the current month, including information such as the shooting date and time;

[0090] Step S1.2: Arrange the collected pathological images and sort them into pathological images of 4 weeks in chronological order.

[0091] Step S1 is to prepare for the collection and organization of pathological images, collect pathological images that need to be analyzed in the near future (e.g., within one month), so as to observe the image quality of each week of the month and summarize the problems according to the possible rules or trends, so as to facilitate the analysis and statistics of the problems by week, so as to organize the statistical rules and feedback to the pathology department later.

[0092] Step S2 specifically includes:

[0093] Step S2.1: According to each original question, score each collected pathological image to obtain the original score S of the i-th original question. i , i=1,2,...9;

[0094] Step S2.2: Combining the original scores of each original question, the total quality score S of the pathological image is obtained by the following formula: 总;

[0095] ;

[0096] Where: W i is the preset weight corresponding to the i-th original question;

[0097] Step S2.3: Determine the total quality score S of the pathological image 总 Is it less than the quality passing score S? 及格 ;

[0098] If so, the image is an unqualified pathological image;

[0099] Otherwise, the image is a qualified pathological image.

[0100] Specifically,

[0101] First, 9 filters are established, and 1 filter corresponds to 1 original problem. The 9 original problems specifically include: staining problem, overlapping problem, breakage problem, wrinkle problem, blur problem, tissue ratio problem, coverslip edge problem, bubble problem, and handwriting problem.

[0102] Secondly, 9 filters are used to detect the pathological images respectively to obtain 9 filtered images for different original problems.

[0103] Finally, the filtered images are scored using a processor or other device with computing power.

[0104] Filtered image for staining problem: Score the sample according to its uniformity and depth of staining to get the original score S 1 .

[0105] Filtered image for overlapping problem: score according to the size and number of overlapping areas to get the original score S 2 .

[0106] Filtered image for fracture problem: score according to the number, size and degree of fracture areas to get the original score S 3 .

[0107] For the image after filtering for wrinkles: the wrinkles are evaluated and scored according to the number, size, and depth of the wrinkles to obtain the original score S 4 .

[0108] For the image after filtering for blur problem: score the image according to its clarity and clear area ratio to get the original score S 5 .

[0109] Filtered image for tissue ratio problem: The tissue ratio is evaluated and scored according to the area of ​​the tissue in the image to obtain the original score S6 .

[0110] Image filtered for cover glass edge problem: score according to the size of the overlap area between the cover glass edge and the tissue to get the original score S 7 .

[0111] Filtered image for bubble problem: score according to the number and size of bubbles to get the original score S 8 .

[0112] For the handwriting problem, the filtered image is scored according to the depth of the handwriting and the proportion of the tissue occupied, and the original score S is obtained. 9 .

[0113] Assign weights to the original scores of each different original question, and summarize the weighted scores of different original questions to obtain the final quality score S of the pathological image. 总 .

[0114] The weight distribution is reasonably distributed according to the importance of different original problems, providing a basis and standard for subsequent classification and quality assessment of the entire image. For example, in the quality assessment of pathological images, the importance of factors such as tissue proportion, blur, and handwriting may be relatively high, while the importance of factors such as cover glass edge and wrinkles may be relatively low. Therefore, it is necessary to comprehensively consider the importance of various factors according to the actual situation and make reasonable weight distribution.

[0115] The weight setting can make each problem have a certain degree of importance, which can not only play a role in assessing the important quality issues of the image, but also will not be too biased towards certain issues and ignore the evaluation of other parts. However, the actual situation needs to be determined and adjusted according to the specific pathology image evaluation needs.

[0116] The total weight sum is 100%, and the full score of the entire image is S 满分 =100 points. The weights assigned to each original question are shown in Table 1. The second column is the actual score of the original question, the third column is the corresponding weight, and the fourth column is the reference value of different weights. 6 If it is set to 20%, the full score for the organization ratio is 20 points, and the actual score is S 6 , and so on.

[0117] Therefore, the total quality score of pathological images is S 总 The better the image quality, the higher the score.

[0118] ;

[0119] Set the pass rate R 总(Refer to 60%), the full score S 满分 With R 总 Multiply them together to get the passing score S for image quality. 及格 , which will be used as the criterion for passing scores.

[0120] ;

[0121] like , the image quality is judged to be unsatisfactory.

[0122] Table 1 Original question, original score, and weight distribution diagram

[0123] Original question Original score Weight / % Weight reference / % Dyeing Problems <![CDATA[S 1 ]]> <![CDATA[W 1 ]]> 10 Overlapping issues <![CDATA[S 2 ]]> <![CDATA[W 2 ]]> 10 Fracture problem <![CDATA[S 3 ]]> <![CDATA[W 3 ]]> 10 Wrinkle problem <![CDATA[S 4 ]]> <![CDATA[W 4 ]]> 5 Fuzzy issues <![CDATA[S 5 ]]> <![CDATA[W 5 ]]> 15 Organizational proportion issues <![CDATA[S 6 ]]> <![CDATA[W 6 ]]> 20 Cover glass edge problem <![CDATA[S 7 ]]> <![CDATA[W 7 ]]> 5 Bubble Problem <![CDATA[S 8 ]]> <![CDATA[W 8 ]]> 10 Handwriting Problems <![CDATA[S 9 ]]> <![CDATA[W 9 ]]> 15

[0124] Each classification corresponds to a different original question. After scoring each original question, the corresponding scores should be weighted and summarized. This operation can reduce the impact of subjective evaluation factors on the classification results and improve the credibility of the evaluation. For unqualified pathological images, find out which step in the pathological image preparation process has problems, so as to facilitate subsequent remediation, such as re-slicing, re-making slides or re-scanning.

[0125] Each of the original questions is analyzed below.

[0126] (1) Tissue ratio problem: If the tissue ratio is low and the background is too much, it is very likely that the scanning parameters were not set correctly when the digital scanner scanned the image.

[0127] Therefore, the tissue ratio problem is a scanning problem and can be solved by rescanning later.

[0128] (2) Handwriting issues: Handwriting is generally used by doctors to mark and record. The following are some situations where the handwriting on the coverslip needs to be erased and rescanned: the handwriting covers up meaningful tissue structures, resulting in an inaccurate diagnosis; the handwriting is too dense, which may interfere with the algorithm's recognition and analysis during image processing; the handwriting incorrectly marks tissue structures or lesion areas, which may mislead subsequent diagnosis and treatment; the handwriting quality is poor, such as blurring, overlapping, fading, unclear colors, etc., which cannot be effectively recognized and analyzed, which may affect the image quality and the accuracy of the diagnosis results.

[0129] Since the cover glass has been placed on the tissue when marking, try not to remove the cover glass when removing the handwriting, because there is a high probability that the specimen will be displaced or the tissue specimen will be damaged. It is recommended to use a special solvent to remove the handwriting.

[0130] Therefore, the handwriting problem is a scanning problem. You can just erase the handwriting on the cover glass and scan again.

[0131] (3) Bubble problem: The reasons for the generation of bubbles basically exist in the process of making glass slides, such as: the glass slides are not thoroughly cleaned, and there are impurities such as dust and grease, which cause bubbles to appear on the surface of the glass slides; the slides are not fully flattened during bonding, and the samples are not fully attached to the slides, leaving gaps for air to enter between the slides and form bubbles; during the production process, the slides shake or move.

[0132] Therefore, the bubble problem is a problem with the slide making, and you can start over by remaking the slide.

[0133] (4) Problem with the edge of the coverslip: The edge of the coverslip is included in the digital scan. This is most likely because the cover slip was not placed correctly or was too small when it was prepared. If the problem is caused by incorrect placement, you can remove the coverslip and adjust the position of the coverslip to fix it again. If the problem is caused by the cover slip being too small, you need to prepare a coverslip that meets the size requirements or use a larger coverslip to fix the slices to avoid the edge of the coverslip appearing in the image.

[0134] Therefore, the problem with the edge of the coverslip belongs to the problem of slide making, and you can start from remaking the slide later.

[0135] (5) Wrinkle problem: The cause of wrinkles exists in the process of slicing or making glass slides. The first case is wrinkles during slicing: this includes wrinkles caused by the quality of the blade, such as the blade is not sharp enough or worn; the second is wrinkles caused by the tissue being cut too thick. The thickness of common tissue slices is generally between 3-5 microns. The thinner the tissue is cut, the easier it is to maintain the integrity and shape of the tissue structure, reducing the possibility of wrinkles and deformation. When the tissue is too thick, the central part of the slice needs to be subjected to a lot of pressure to be cut, which can easily cause deformation, fracture and wrinkles in the central part. The second case is that the force applied when pasting is different, and repeated friction or squeezing causes the tissue to deform and wrinkle.

[0136] Therefore, the wrinkle problem is a slicing problem, and you can start from re-slicing later.

[0137] (6) Ambiguity problem: There are three main reasons for ambiguity.

[0138] 1: The uneven tissue surface causes blurred scanning;

[0139] 2: Staining problems: Insufficient staining will result in unclear details in cells and tissues; excessive staining will cause distortion or blurring of details in cells and tissues; uneven staining and different shades of color will cause the image to look blurry and unclear;

[0140] 3: Scanning problem: Dust or stains on the scanner lens or sensor cause the scanned image to be blurry; insufficient light during scanning causes blurriness; the scanner or the scanned object moves during scanning, causing the image to be blurry.

[0141] Therefore, the blur problem can be attributed to either a slicing problem or a scanning problem. If the blur is caused by an uneven tissue surface or staining problem, you can start by re-slicing. If the blur is caused by a scanning problem, you can just re-scan it.

[0142] (7) Overlap problem: The overlap is most likely caused during the slicing stage, but a small chance is that the tissue is not flattened during patching.

[0143] Therefore, the overlapping problem belongs to the slicing problem, and it is necessary to start from re-slicing afterwards.

[0144] (8) Fracture problem: The cause of fracture basically occurs during the slicing stage, which is caused by the quality of the blade, such as wear or dullness.

[0145] Therefore, the breakage problem belongs to the slicing problem, and it is necessary to start from re-slicing afterwards.

[0146] (9) Dyeing problems: There are three main types of dyeing problems.

[0147] 1. Uneven staining and staining intensity problems: some parts are overstained, while others are understained, resulting in large color differences between different cell tissues.

[0148] 2. Staining heterogeneity problem: Different cells in the same lesion area stain inconsistently, showing heterochromia, which affects the results of pathological judgment.

[0149] 3. Technical error problems: including quality problems of dyes, inaccurate dyeing temperature or lower than standard requirements, inconsistent dyeing time, and improper dyeing techniques, such as uneven dripping of dyeing liquid, too long dyeing time, poor quality of dyeing liquid, etc., which will affect the dyeing effect.

[0150] Therefore, the staining problem belongs to the slicing problem, and the subsequent operation can be started from re-slicing.

[0151] In summary, combined with the above prior knowledge, based on the analysis of the entire process of pathological image preparation, the management and control personnel can classify the 9 original questions and generate 3 classified questions containing the corresponding original questions. The classified questions are problems existing in any step of the pathological image preparation process. The 3 classified questions specifically include: slicing problems, scanning problems, and slide making problems.

[0152] Specifically, each classification problem includes its corresponding original problem, as follows:

[0153] Sectioning problems: staining problems, overlapping problems, breakage problems, wrinkle problems and blurring problems;

[0154] Scanning issues: fuzziness and tissue proportion issues;

[0155] Slide making problems: coverslip edge problems, bubble problems, and pen marking problems.

[0156] The control personnel preset the correspondence between the original question and the classification question, and input it into the processor for subsequent processing.

[0157] Step S4 specifically includes:

[0158] Step S4.1: For each classification problem, the classification score E of the j-th classification problem of the pathological image is obtained by the following formula: j , j=1,2,3;

[0159] ;

[0160] Among them: J j is the set of original questions contained in the j-th classification problem;

[0161] Step S4.2: For each classification problem, the classification score E of the j-th classification problem of the pathological image is obtained by the following formula: j-满分 And its corresponding classification passing score E j-及格 , j=1,2,3;

[0162] ;

[0163] ;

[0164] Where: R j is the preset passing rate of the j-th classification problem;

[0165] Step S4.3: Determine the classification score E of the jth classification problem of the pathological image j Is it less than the passing score E of the corresponding category? j-及格 ;

[0166] If yes, then the image has this classification problem;

[0167] Otherwise, the image does not have this classification problem.

[0168] Specifically, the classification of slicing problems (including staining problems, overlapping problems, break problems, fold problems and fuzzy problems) is divided into E 1 , Full score in classification: E 1-满分 And classification pass score E 1-及格 ,as follows:

[0169] ;

[0170] ;

[0171] .

[0172] Similarly: Classification of scanning problems (collection has fuzzy problems and organizational proportion problems) 2 , Full score in classification: E 2-满分 And classification pass score E 2-及格 ,as follows:

[0173] ;

[0174] ;

[0175] .

[0176] Classification of slide making problems (including cover glass edge problems, bubble problems and handwriting problems) 3 as follows:

[0177] ;

[0178] ;

[0179] .

[0180] The above R 1 , R 2 , R 3 The value can be 60%.

[0181] Step S5 specifically includes counting the original problems and classification problems existing in all pathological images, and performing trend analysis to obtain analysis results.

[0182] Statistics and analysis are conducted on the problems that have occurred recently. Trend curves are drawn based on the different problems that occur in the images at each stage (such as each week). The trends of various problems that have occurred recently are observed, and the causes of the problems are summarized.

[0183] By statistically scoring, we can understand the frequency and severity of different original problems, and provide more information for the classification and cause analysis of the problems. For example, if a certain original problem occurs frequently and is of high severity, it means that this original problem deserves special attention and needs to be corrected accordingly. In addition, by statistically scoring and classifying different original problems, we can understand which problems are relatively prominent and which are more common, thus providing a basis for subsequent improvements. In summary, we can analyze the trends of the two dimensions of classification and problems, respectively, and provide corresponding references and guidance for relevant institutions and personnel.

[0184] By counting the classification trends in each week of the month, we can understand the number and occurrence of each classification in the batch of pathological images, and then derive the trend of the classification problem, such as the changing trend of the frequency of occurrence of a specific classification problem in the month.

[0185] By analyzing the trend of classified problems and following the principle of problem priority, we prioritize corrective measures for classified problems with higher frequency and severity, thus improving the accuracy and trust of pathological diagnosis results. For example, the number of cases requiring slicing problems in the month is increasing week by week, showing an upward trend, while the number of problems with slide making or scanning is relatively low. This means that it is highly likely that the doctor responsible for slicing is not competent enough or the knife used for slicing is worn or blunt.

[0186] In addition to statistical classification trends, the number and change trends of different original problem types are also the same. Through statistical analysis of the original problem trends, we can deeply understand the occurrence and causes of various problems and propose corresponding improvement measures. At the same time, we can also draw on professional knowledge and experience in other fields, introduce new detection technologies and methods, and improve the quality of pathological images and diagnostic effects.

[0187] Furthermore, the quality control method of the present invention further comprises:

[0188] Step S6: Feedback the original problems, classification problems and analysis results of the pathological images to the pathology department.

[0189] Based on the analysis results after statistical trends, the status of the problems and analysis reports are fed back to the pathology department, and improvement plans and suggestions as well as the possible reasons for each type of problem are put forward in a timely manner.

[0190] Based on the feedback, the Pathology Department inspected and reviewed the production process in the past month, corrected and improved existing problems, reduced or avoided the occurrence of similar problems, and produced high-quality pathology images in the future, ultimately improving the accuracy of pathology diagnosis and treatment effectiveness.

[0191] Feedback to the pathology department means that after the pathology image is detected, the detection and trend statistical results are fed back to the pathologists, so that they can understand the problems of the pathology images and the corresponding corrective measures.

[0192] The content of the feedback to the pathology department includes specific and clear information such as the test results of the pathology images, the number and severity of the problems. For example, the frequency, number, type of a certain problem in the pathology image, and the corresponding corrective measures. For different types of problems, corresponding suggestions and measures can be given to improve the pathologists' understanding and problem-solving capabilities. At the same time, relevant data support should also be provided, including statistical data on the types of problems, treatment plans, and pathology analysis results, so that pathologists can quickly recognize the severity of the disease and take corresponding measures.

[0193] Feedback to the pathology department can be implemented through a variety of channels, such as regular business meetings, door-to-door explanations and diagnostic reports. Different feedback methods should be targeted at different types of problems and groups of pathologists. For example, for pathologists with a common problem, they can be given short-term training on the problem and a defect registration form can be opened to facilitate pathologists to make more consistent diagnoses of pathology images. At the same time, feedback on the implementation of corrective measures should be shared with pathologists on a regular basis, as well as the specific results and measures of the feedback.

[0194] Feedback on the improvement effect of the pathology department should be the ultimate goal of continuous improvement of the entire system. The slow response of pathologists to specific problems and the lack of timely feedback mechanism lead to the inability to solve the problems in a timely manner. Therefore, the management and training of pathologists should be strengthened according to specific problems, and better film production and scanning tools should be introduced. In addition, through the analysis and evaluation of the improvement effect after feedback, the deficiencies of the feedback mechanism can be reversely detected and corrected, and the mechanism can be evaluated and adjusted regularly.

[0195] In summary, the main process of the present invention is as follows: 1) Collect and organize pathological images; 2) Establish multiple filters to locate the original problems in the pathological images, score them according to the severity of the problems, assign weights to each original problem according to its importance to the image, and comprehensively score the entire pathological image; 3) Based on the entire process analysis of pathological image preparation, the original problems are classified to obtain classified problems; 4) Conduct in-depth analysis; 5) Calculate the number of different original problems and classified problems each week and statistically calculate the proportion, draw trend curves, and analyze recent problems; 6) Feedback the analysis results to the pathology department, and suggest which process step (i.e., classification problem) needs to be restarted, such as rescanning, remaking slides or re-slicing, give possible reasons for the problem, and then put forward rectification suggestions to solve the problems in the production and scanning process of pathological images, and ultimately improve the quality of subsequent pathological images, providing pathologists with more accurate and reliable pathological diagnoses.

[0196] In addition, an embodiment of the present invention further discloses a quality control device for pathological images, including:

[0197] An acquisition module, used to collect pathological images within one month;

[0198] The first problem detection module is used to perform quality detection on each collected pathological image based on 9 original questions to determine whether the image is a qualified pathological image;

[0199] A problem classification module is used to classify the 9 original problems based on the analysis of the entire process of pathological image preparation, and generate 3 classification problems containing the corresponding original problems. The classification problems are problems existing in any step of the pathological image preparation process;

[0200] A second problem detection module is used to perform in-depth analysis on each unqualified pathological image based on the original problem contained in each classification problem to determine whether the image has the classification problem;

[0201] The statistical analysis module is used to count the original problems and classification problems existing in all pathological images, and perform trend analysis to obtain analysis results.

[0202] Furthermore, the acquisition module specifically includes:

[0203] A collection unit, used to collect pathological images within one month;

[0204] The sorting unit is used to sort the collected pathological images.

[0205] Further, the first problem detection module specifically includes:

[0206] The original score acquisition unit is used to score each collected pathological image according to each original question to obtain the original score S of the i-th original question. i , i=1,2,...9;

[0207] The quality total score acquisition unit is used to integrate the original scores of each original question and obtain the total quality score S of the pathological image through the following formula: 总 ;

[0208] ;

[0209] Where: W i is the preset weight corresponding to the i-th original question;

[0210] The first judgment unit is used to judge the total quality score S of the pathological image 总 Is it less than the quality passing score S? 及格 ;

[0211] If so, the image is an unqualified pathological image;

[0212] Otherwise, the image is a qualified pathological image.

[0213] Further, the second problem detection module includes:

[0214] The classification score acquisition unit is used to obtain the classification score E of the j-th classification problem of the pathological image for each classification problem by the following formula: j , j=1,2,3;

[0215] ;

[0216] Among them: J j is the set of original questions contained in the j-th classification problem;

[0217] The classification passing score obtaining unit is used to obtain the classification full score E of the j-th classification problem of the pathological image for each classification problem by the following formula: j-满分 And its corresponding classification passing score E j-及格 , j=1,2,3;

[0218] ;

[0219] ;

[0220] Where: R j is the preset passing rate of the j-th classification problem;

[0221] The second judgment unit is used to judge the classification score of the jth classification problem of the pathological image. j Is it less than the passing score E of the corresponding category? j-及格 ;

[0222] If yes, then the image has this classification problem;

[0223] Otherwise, the image does not have this classification problem.

[0224] Furthermore, the quality control equipment also includes:

[0225] The feedback module is used to feed back the original problems, classification problems and analysis results of pathological images to the pathology department.

[0226] The contents of quality control equipment are similar to those of quality control ideas and will not be repeated here.

[0227] The present invention discloses a pathological image quality control method and device that can combine the preparation process of the pathological image, find the source of the problem, match and classify the original problem, and control the quality of the pathological image, which has the following beneficial effects:

[0228] First, the present invention adopts machine learning algorithms and image processing technology, which can automatically and quickly detect pathological images, and quantitatively analyze and score different original questions, avoiding errors and missed detections in the manual processing process, and improving the efficiency and reliability of pathological image detection.

[0229] Second, by counting the original problems and classification problems in all pathological images and conducting trend analysis, we can discover the commonalities and characteristics of common problems, summarize the causes, and provide data support and reference basis for the subsequent production of high-quality pathological images.

[0230] Third, based on the trend of pathology image problems, analyze the causes and solutions to help pathologists check and discover potential problems in pathology image production and scanning, and correct them, thereby improving the quality and accuracy of pathology images. Provide feedback and guidance to the pathology department.

[0231] Fourth, real-time monitoring of pathological image quality helps the pathology department improve work efficiency and diagnostic accuracy, while also providing medical institutions with safer and more reliable medical services.

[0232] The present invention can accurately and comprehensively detect and identify original problems existing in pathological images, score different original problems according to their impact on image quality, classify classification problems (i.e., step sources) corresponding to different original problems, statistically analyze problem trends in pathological images, and provide solutions to improve the process of pathology department in producing pathological images, ultimately producing pathological images with higher quality and accuracy, improving the accuracy and efficiency of pathological detection, and enabling pathologists to make more accurate and reliable pathological diagnoses. The present invention is practical, reliable, and applicable.

[0233] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for illustrating the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which shall fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A quality control method for pathological images, characterized in that: include: Step S1: collecting pathological images within any time period; Step S2: Based on 9 original problems, quality inspection is performed on each collected pathological image to determine whether the image is a qualified pathological image. The 9 original problems specifically include: staining problem, overlap problem, break problem, wrinkle problem, blur problem, tissue ratio problem, cover glass edge problem, bubble problem, and handwriting problem; Step S3: Based on the analysis of the entire process of pathological image preparation, several original questions are classified to generate three classification questions containing the corresponding original questions. The classification questions are problems existing in any step of the pathological image preparation process. The three classification questions specifically include: slicing problems, scanning problems, and slide making problems; Step S4: Based on the original problem contained in each classification problem, perform in-depth analysis on each unqualified pathological image to determine whether the image has the classification problem; Step S5: Count the original problems and classification problems existing in all pathological images, and perform trend analysis to obtain analysis results; Step S6: Feedback the original problems, classification problems and analysis results of the pathological images to the pathology department; The step S4 specifically includes: Step S4.1: For each classification problem, the classification score E of the j-th classification problem of the pathological image is obtained by the following formula: j , j = 1, 2, ... m; Among them: J j is the set of original questions contained in the j-th classification problem; S k is the original score for the kth original question; W k is the preset weight corresponding to the kth original question; Step S4.2: For each classification problem, the classification score E of the j-th classification problem of the pathological image is obtained by the following formula: j-满分 And its corresponding classification passing score E j-及格 , j = 1, 2, ... m; AND j-及格 =And j-满分 *R j ; Where: R j is the preset passing rate of the j-th classification problem; Step S4.3: Determine the classification score E of the jth classification problem of the pathological image j Is it less than the passing score E of the corresponding category? j-及格 ; If yes, then the image has this classification problem; Otherwise, the image does not have this classification problem.

2. The quality control method of pathological images according to claim 1, characterized in that: The step S1 specifically includes: Step S1.1: collecting pathological images within any time period; Step S1.2: Arrange the collected pathological images.

3. The quality control method of pathological images according to claim 1, characterized in that: The step S2 specifically includes: Step S2.1: According to each original question, score each collected pathological image to obtain the original score S of the i-th original question. i , i=1,2,…n; Step S2.2: Combining the original scores of each original question, the total quality score S of the pathological image is obtained by the following formula: 总 ; Where: W i is the preset weight corresponding to the i-th original question; Step S2.3: Determine the total quality score S of the pathological image 总 Is it less than the quality passing score S? 及格 ; If so, the image is an unqualified pathological image; Otherwise, the image is a qualified pathological image.

4. A quality control device for pathological images, characterized in that: include: An acquisition module, used to acquire pathological images within any time period; The first problem detection module is used to perform quality detection on each collected pathological image based on 9 original problems to determine whether the image is a qualified pathological image. The 9 original problems specifically include: staining problem, overlap problem, break problem, wrinkle problem, blur problem, tissue ratio problem, cover glass edge problem, bubble problem, and handwriting problem; A problem classification module is used to classify a number of original problems based on the analysis of the entire process of pathological image preparation, and generate three classification problems containing the corresponding original problems. The classification problems are problems existing in any step of the pathological image preparation process. The three classification problems specifically include: slicing problems, scanning problems, and slide making problems; A second problem detection module is used to perform in-depth analysis on each unqualified pathological image based on the original problem contained in each classification problem to determine whether the image has the classification problem; The statistical analysis module is used to count the original problems and classification problems existing in all pathological images, and to perform trend analysis to obtain analysis results; Feedback module, used to feed back the original problems, classification problems and analysis results of pathological images to the pathology department; The second problem detection module includes: The classification score acquisition unit is used to obtain the classification score E of the j-th classification problem of the pathological image for each classification problem by the following formula: j , j = 1, 2, ... m; Among them: J j is the set of original questions contained in the j-th classification problem; S k is the original score for the kth original question; W k is the preset weight corresponding to the kth original question; The classification passing score obtaining unit is used to obtain the classification full score E of the j-th classification problem of the pathological image for each classification problem by the following formula: j-满分 And its corresponding classification passing score E j-及格 , j = 1, 2, ... m; AND j-及格 =And j-满分 *R j ; Where: R j is the preset passing rate of the j-th classification problem; The second judgment unit is used to judge the classification score of the jth classification problem of the pathological image. j Is it less than the passing score E of the corresponding category? j-及格 ; If yes, then the image has this classification problem; Otherwise, the image does not have this classification problem.

5. The quality control device for pathological images according to claim 4, characterized in that: The acquisition module specifically includes: An acquisition unit, used for acquiring pathological images within any time period; The sorting unit is used to sort the collected pathological images.

6. The pathological image quality control device according to claim 4, characterized in that: The first problem detection module specifically includes: The original score acquisition unit is used to score each collected pathological image according to each original question to obtain the original score S of the i-th original question. i , i=1,2,…n; The quality total score acquisition unit is used to integrate the original scores of each original question and obtain the total quality score S of the pathological image through the following formula: 总 ; Where: W i is the preset weight corresponding to the i-th original question; The first judgment unit is used to judge the total quality score S of the pathological image 总 Is it less than the quality passing score S? 及格 ; If so, the image is an unqualified pathological image; Otherwise, the image is a qualified pathological image.

Images