Auxiliary diagnosis information providing apparatus and blood analysis system

Abstract

The application is a divisional application of 202310323347.9, and provides an auxiliary diagnosis information providing device and a blood analysis system. The auxiliary diagnosis information providing device comprises: an acquisition module configured to acquire sample detection data containing a sample analysis image and / or a sample detection parameter obtained by detecting and analyzing a biological sample of a to-be-detected object; and an output module configured to output an auxiliary diagnosis report comprising auxiliary diagnosis category information and a target reference sample image according to the sample analysis image, first abnormal feature information identified based on the sample analysis image, and / or second abnormal feature information identified based on the sample detection parameter. The target reference sample image in the auxiliary diagnosis report is a reference image of a historical diagnosis case that meets a preset condition in terms of similarity to the sample analysis image and is confirmed by microscopy, so that a doctor can quickly master the diagnosis basis of the obtained auxiliary diagnosis category information to judge the reliability of the current auxiliary diagnosis category information.

Description

[0001] This application is a divisional application. The original application has the application number 202310323347.9, the application date is March 30, 2023, and the invention title is "Auxiliary Diagnostic Information Providing Device and Blood Analysis System". Technical Field

[0002] This application relates to the field of medical auxiliary technology, and in particular to an auxiliary diagnostic information providing device and a blood analysis system. Background Technology

[0003] In the field of medical diagnosis, it is very common to collect biological samples such as blood, urine, and saliva from patients for testing and analysis to obtain sample test data. This sample test data can help doctors determine whether the patient may have a certain disease.

[0004] Currently, the sample testing data obtained from the analysis of biological samples is mainly completed by doctors based on their experience and knowledge. On the one hand, the utilization of sample testing data is highly limited by the doctor's personal experience and knowledge, which can easily lead to important information not being effectively identified and utilized. On the other hand, the identification of a large amount of sample testing data consumes a significant amount of doctors' time and energy, seriously affecting diagnostic efficiency.

[0005] With the increasing technological and informational advancements in the medical industry, computer-aided diagnostic systems have gradually developed in recent years. These systems analyze and identify abnormalities in sample test data, providing this information to doctors to assist in diagnostic decisions. However, in current known solutions for these systems, doctors often lack knowledge of how the system identifies these abnormalities. This leads to significant confusion among doctors regarding the adoption of the information provided by these systems, severely limiting the practical application and development of diagnostic technology. Summary of the Invention

[0006] To address the existing technical problems, this application provides an accurate, efficient auxiliary diagnostic information providing device and a blood analysis system that facilitates the effective verification of auxiliary diagnostic results.

[0007] A first aspect of this application provides an auxiliary diagnostic information providing device, comprising:

[0008] The acquisition module is used to acquire sample detection data obtained by analyzing biological samples of the object to be tested; the sample detection data includes sample analysis images and sample detection parameters.

[0009] The first anomaly identification module is used to obtain first anomaly feature information characterizing the abnormal features in the sample analysis image based on the sample analysis image;

[0010] The second anomaly identification module is used to obtain second anomaly feature information that characterizes the abnormal parameter values ​​in the biological sample, based on the sample detection parameters.

[0011] The output module is used to output an auxiliary diagnostic report based on the sample analysis image, the first abnormal feature information, and the second abnormal feature information. The auxiliary diagnostic report includes:

[0012] Information on auxiliary diagnostic categories;

[0013] An activation map used to highlight the diagnostic basis for the first abnormal feature information is formed by superimposing a heatmap obtained based on the feature map of the sample analysis image and the biological sample analysis image; wherein, the heatmap characterizes the degree of influence of each unit region of the sample analysis image on the decision to determine the first abnormal feature information;

[0014] The sample analysis image carries a positioning marker, which is used to indicate the location of the first abnormal feature information in the sample analysis image.

[0015] A second aspect of this application provides an auxiliary diagnostic information providing device, comprising:

[0016] The acquisition module is used to acquire sample detection data obtained by analyzing biological samples of the object to be tested; the sample detection data includes sample analysis images and sample detection parameters.

[0017] The first anomaly identification module is used to obtain first anomaly feature information characterizing the abnormal features in the sample analysis image based on the sample analysis image;

[0018] The second anomaly identification module is used to obtain second anomaly feature information that characterizes the abnormal parameter values ​​in the biological sample, based on the sample detection parameters.

[0019] The output module is used to output an auxiliary diagnostic report based on the sample analysis image, the first abnormal feature information, and the second abnormal feature information. The auxiliary diagnostic report includes:

[0020] Information on auxiliary diagnostic categories;

[0021] The sample analysis image carries a positioning marker, which is used to indicate the location of the first abnormal feature information in the sample analysis image;

[0022] A target reference sample image that matches the sample analysis image, wherein the target reference sample image is a reference image in the sample analysis image library whose similarity to the sample analysis image meets a preset condition; the sample analysis image library includes sample analysis images of historical diagnostic cases confirmed by microscopic examination, and the positioning markers in the target reference sample image are used to highlight the location of the diagnostic basis for obtaining the first abnormal feature information.

[0023] A third aspect of this application provides a blood analysis system, comprising:

[0024] A sampling component for collecting and distributing biological samples, specifically blood samples, from the object to be tested.

[0025] A reaction assembly for processing the biological sample to form a test solution;

[0026] A driving component is used to drive the liquid path between the sampling component and the reaction component;

[0027] The detection component is used to classify and count the blood cells contained in the test solution to obtain sample detection data containing sample detection parameters and sample analysis images;

[0028] The auxiliary diagnostic information providing device described in any embodiment of this application is used to output an auxiliary diagnostic report based on the sample detection data.

[0029] In the above embodiments, the auxiliary diagnostic information providing device obtains first abnormal feature information characterizing abnormal features in the sample analysis image by analyzing and identifying the sample analysis image, and obtains second abnormal feature information characterizing abnormal parameter values ​​in the biological sample based on the sample detection parameters. Based on the sample analysis image, the first abnormal feature information, and the second abnormal feature information, it outputs an auxiliary diagnostic report. The auxiliary diagnostic report includes auxiliary diagnostic category information, a sample analysis image showing the location of the first abnormal feature information through positioning markers, and an activation image / target reference sample image obtained from historical diagnostic examples confirmed by microscopic examination highlighting the diagnostic basis of the first abnormal feature information obtained from the sample analysis image. Thus, users can obtain auxiliary diagnostic category information determined by comprehensive information from multiple aspects through the auxiliary diagnostic report, improving the accuracy and efficiency of the auxiliary diagnostic results. Furthermore, the activation image / target reference sample image confirmed by microscopic examination and the sample analysis image with positioning markers output in the auxiliary diagnostic report allow doctors to quickly verify the accuracy of the current auxiliary diagnostic category information, determine its reliability, and consider whether to adopt the current auxiliary diagnostic category information.

[0030] The blood analysis system provided in the above embodiments has the same technical concept and thus the same technical effect as the corresponding embodiments of each auxiliary diagnostic information providing device, and will not be described again here. Attached Figure Description

[0031] Figure 1 This is a diagram illustrating optional application scenarios for an auxiliary diagnostic information providing device provided in one embodiment of this application.

[0032] Figure 2 A diagram illustrating optional application scenarios for an auxiliary diagnostic information providing device provided in another embodiment of this application;

[0033] Figure 3 A schematic diagram of the structure of an auxiliary diagnostic information providing device provided in one embodiment;

[0034] Figure 4 A schematic diagram of the interface for an auxiliary diagnostic report provided in one embodiment;

[0035] Figure 5 A schematic diagram of the structure of an auxiliary diagnostic information providing device provided in another embodiment;

[0036] Figure 6 This is a schematic diagram of a blood analysis system provided in one embodiment. Detailed Implementation

[0037] The technical solution of this application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0038] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein in the specification of this application is for the purpose of describing particular embodiments only and is not intended to limit the ways in which this application may be implemented. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0039] In the following description, the expression “some embodiments” is used, which describes a subset of possible embodiments. However, it should be understood that “some embodiments” can be the same subset or different subsets of all possible embodiments and can be combined with each other without conflict.

[0040] In the following description, the terms “first,” “second,” etc., are used merely to distinguish similar objects and do not represent a specific ordering of objects. It is understood that “first,” “second,” and “third” may be interchanged in a specific order or sequence where permitted, so that the embodiments of this application described herein can be implemented in an order other than that illustrated or described herein.

[0041] Please see Figure 1 The diagram shows an optional application scenario of the auxiliary diagnostic information providing device provided in an embodiment of this application. The auxiliary diagnostic information providing device 11 can refer to a computer program product that implements auxiliary diagnostic functions based on computer program flow, such as various application programs; or it can refer to an auxiliary diagnostic device that loads a corresponding computer program product to implement auxiliary diagnostic functions, such as various intelligent devices with storage and computing capabilities.

[0042] In practical applications, the auxiliary diagnostic information providing device 11 refers to an auxiliary diagnostic device loaded with a corresponding computer program product. It can be a physically independent intelligent device or integrated with a known intelligent device; for example... Figure 1 As shown, the auxiliary diagnostic information providing device 11 is a smart device physically separate from the sample analyzer 21, such as a smartphone, personal computer, medical diagnostic instrument, cloud server, etc.; please refer to Figure 2 The auxiliary diagnostic information providing device 11 is integrated with the sample analyzer 21, such as a sample analyzer loaded with corresponding computer programs.

[0043] The auxiliary diagnostic information providing device 11 can be configured to communicate with the output interface of the application system within the sample analyzer 21 that performs sample detection and analysis, and is used to obtain sample detection data obtained after performing detection and analysis on the biological sample of the object to be tested from the sample analyzer 21. The sample analyzer 21 can refer to a device used for intelligent detection and analysis of collected biological samples. Biological samples can be samples taken from the body of the object to be tested and containing various biological cell information or other biological information, such as blood samples, urine samples, other body fluids (pleural effusion, ascites, cerebrospinal fluid, serous cavity effusion, synovial fluid), etc. The biological cell types can be at least one of the following: neutrophils, lymphocytes, monocytes, eosinophils, and basophils; or they can be immature granulocytes, tumor cells, lymphoblasts, plasma cells, atypical lymphocytes, proerythroblasts, basal erythroblasts and polychromatic erythroblasts, normal erythroblasts, promegalyte cells, basal giant cells, polychromatic giant cells, and nucleated erythroblasts selected from normal giant erythroblasts and macronucleated globules.

[0044] The auxiliary diagnostic information providing device 11 acquires sample test data obtained by the sample analyzer 21 from the biological sample. It can combine the clinical information of the subject, the sample test data, and a case database formed based on historical diagnostic records to determine abnormal characteristics of the biological sample. Based on the identified abnormal characteristics, it outputs an auxiliary diagnostic report. This report provides auxiliary diagnostic category information determined based on the abnormal characteristics in the biological sample, as well as the diagnostic basis for obtaining this category. This allows users to more intuitively understand the possible or potential symptoms of the subject by viewing the auxiliary diagnostic report. Furthermore, in cases where abnormalities are confirmed, the diagnostic basis provided in the report informs decision-making, enabling accurate auxiliary diagnostic results without relying on the individual experience of the laboratory physician. This improves testing efficiency and accuracy, and makes the diagnostic conclusions more interpretable. The sample test images and parameters obtained by the sample analyzer can be transformed into descriptive auxiliary diagnostic reports, reducing the workload of laboratory physicians in screening test data. Simultaneously, the results in the auxiliary diagnostic reports can better align with tiered medical service policies.

[0045] Optionally, in the process of determining the auxiliary diagnostic category information, the auxiliary diagnostic information providing device 11 can comprehensively consider multiple aspects of the data of the subject to be tested, such as clinical information data that can reflect the individual differences of the subject to be tested, generally including gender, age, medical history, etc. The utilization of clinical information data by the auxiliary diagnostic information providing device 11 can be reflected in multiple stages: one stage is the process of obtaining sample test data by the sample analyzer 21 through testing and analysis of the biological sample of the subject to be tested, in which the sample analyzer 21 considers the clinical information data of the subject to be tested to adjust the obtained sample test data; another stage is that the auxiliary diagnostic information providing device 11 directly acquires the clinical information data of the subject to be tested, and in the process of determining the auxiliary diagnostic category information based on the sample analysis image, the identified first abnormal feature information and the second abnormal feature information, the auxiliary diagnostic category information is further adjusted by comprehensively considering the clinical information data of the subject to be tested. In some embodiments, the use of the clinical information data of the subject to be tested by the auxiliary diagnostic information providing device 11 includes the second stage described above. The auxiliary diagnostic information providing device 11 is communicatively connected to a Laboratory Information System (LIS). The LIS typically includes application terminals located in different locations in the hospital, such as the information desk and laboratory department, which can be used to receive test data, input and save patient test information, and assist the hospital in information management. The auxiliary diagnostic information providing device 11 can directly obtain the specified category of clinical information data of the subject to be tested from the LIS.

[0046] To facilitate understanding of the technical implementation of the auxiliary diagnostic information providing device 11 provided in the embodiments of this application, the specific examples in this application are mainly illustrated using blood samples as an example for detailed description, and the sample analyzer 21 refers to a blood cell analyzer. Known blood cell analyzers are a typical application of flow cytometry. The impedance channel used for cell counting utilizes the Coulter impedance principle, while the hemoglobin concentration measurement utilizes the colorimetric principle. More advanced blood cell analyzers, used for white blood cell classification, reticulocyte identification, nucleated red blood cell identification, basophil identification, low-value platelet identification, and primitive cell identification, utilize laser scattering and nucleic acid fluorescence staining techniques in their optical channels. The goal of these technologies is to convert the biological characteristics of blood cells, such as size, complexity of cell contents, and nucleic acid content, into electrical pulse signals. These electrical pulse signals can be the raw data collected by the blood cell analyzer. By analyzing the collected electrical pulse signals, sample analysis images characterizing blood cell features and numerical sample detection parameters are output. However, it should be noted that although the description of the embodiments in this application uses the analysis data of blood samples as an example, it should not be construed as limiting the scope of protection of this application. For example, in some other embodiments, the sample analyzer 21 can also be a biochemical analyzer. The biochemical analyzer analyzes different biological samples depending on different testing needs. For example, for liver function testing, the biochemical analyzer analyzes alanine aminotransferase (ALT / GPT), aspartate aminotransferase (AST / GOT), alkaline phosphatase (ALP), total bilirubin (T.BIL), direct bilirubin (D.BIL), total protein (TP), and albumin (ALB) in the biological sample of the test subject. For kidney function testing, the biochemical analyzer analyzes urea nitrogen (BUN), creatinine (Cre), carbon dioxide combining power (CO2), and uric acid (UA) in the biological sample of the test subject. For example, in lipid testing, the biochemical analyzer analyzes total cholesterol (CHO), triglycerides (TG), high-density lipoprotein cholesterol (HDL-C), and low-density lipoprotein cholesterol (LDL-C) in the biological sample of the test subject. For blood glucose testing, the biochemical analyzer analyzes glucose (GLU) in the biological sample. In other embodiments, the sample analyzer 21 can also be an immunoassay analyzer, used to analyze tumor markers, thyroid function-related characteristics, reproductive / endocrine-related characteristics, cardiovascular-related characteristics, and / or congenital disease-related characteristics in the biological sample to obtain corresponding sample test data. In other embodiments, the sample test data of the test subject can come from multiple different sample analyzers 21, such as the hematology analyzer, the biochemical analyzer, and the immunoassay analyzer. Accordingly, the sample test data acquired by the auxiliary diagnostic information providing device 11 can include one or more of hematology analysis data, biochemical analysis data, and immunoassay data.

[0047] Please see Figure 3 This is a schematic diagram of the structure of an auxiliary diagnostic information providing device provided in an embodiment of this application. The auxiliary diagnostic information providing device includes an acquisition module 111, a first anomaly identification module 112, a second anomaly identification module 113, and an output module 114a.

[0048] The acquisition module 111 is used to acquire sample detection data obtained by detecting and analyzing biological samples of the object to be detected; the sample detection data includes sample analysis images and sample detection parameters.

[0049] In this context, the "object to be tested" refers to the owner of the biological sample. Taking a human blood sample as an example, the object to be tested is typically the patient who provided the blood sample. Sample testing data refers to the analytical data obtained by counting and detecting various biological samples containing cellular or other biological information. Sample analysis images are image-based detection data that characterizes the features of a biological sample, obtained after the sample analyzer performs counting and analysis on the biological sample. Specifically, sample analysis images can be graphics obtained from analyzing pulse signals during sample testing, or data matrices obtained from pulse signal analysis. Data matrices can be displayed graphically on the instrument interface. Sample testing parameters are numerical detection data that characterizes the features of a biological sample, obtained after the sample analyzer performs testing and analysis on the biological sample. Biological sample characteristics can be the healthy or unhealthy characteristics exhibited by different types of cells in the biological sample, such as the presence of abnormal cell types, cell quantity characteristics, cell size characteristics, cell composition ratio characteristics, cell contents characteristics, and nucleic acid content characteristics. Among them, obtaining sample detection data by analyzing biological samples of the object to be tested can refer to connecting with a sample analyzer and importing sample detection data obtained by the sample analyzer from analyzing biological samples of the object to be tested.

[0050] Optionally, the acquisition module 111 is further configured to acquire clinical information data of the subject to be tested. The auxiliary diagnostic information providing device can be configured to provide a human-computer interactive information input interface, allowing the user to input corresponding information in various configuration items of the information input interface to obtain the clinical information data of the subject to be tested; or, the auxiliary diagnostic information providing device can be configured to communicate with a laboratory information system to import the clinical information data of each subject to be tested from the laboratory information system. The auxiliary diagnostic information providing device, by acquiring the clinical information data of the subject to be tested, comprehensively considers the clinical information data and sample test data to identify abnormal features in the biological sample. In the process of obtaining auxiliary diagnostic category information, it comprehensively considers the age, gender, medical history, and other attribute information of the subject to be tested to analyze the sample analysis image, thereby obtaining more accurate auxiliary diagnostic results while taking into account the individual differences of different subjects to be tested.

[0051] The first anomaly recognition module 112 is used to obtain first anomaly feature information that characterizes the abnormal features in the sample analysis image based on the sample analysis image.

[0052] Abnormal features refer to image features in a sample analysis image that reflect an unhealthy state of the object being tested. For example, in the PLT histogram of blood sample detection data, abnormal features refer to image features in the PLT histogram that reflect an abnormal distribution of platelet counts of different sizes in the blood of the object being tested. The first abnormal feature information refers to the output information based on the identification results of abnormal features in the sample analysis image. It should be noted that the type of output information corresponding to the identification results of abnormal features in the sample analysis image can vary depending on the image recognition algorithm used. In this embodiment, the first abnormal feature information at least includes the category of abnormal features in the sample analysis image, such as the presence of small red blood cell interference or a low platelet count.

[0053] The second anomaly identification module 113 is used to obtain second anomaly feature information that characterizes the abnormal parameter values ​​in the biological sample, based on the sample detection parameters.

[0054] The acquisition of the second abnormal feature information mainly takes two forms. First, the auxiliary diagnostic information providing device obtains sample testing parameters from a third party, comprehensively considers the clinical information data of the subject under test and the sample testing parameters, and determines the abnormal parameter values ​​by comparing them with the corresponding standard parameter table. Second, the third party comprehensively considers the clinical information data of the subject under test and the sample testing parameters, and determines the abnormal parameter values ​​by comparing them with the corresponding standard parameter table. The auxiliary diagnostic information providing device can directly obtain the identification results of the abnormal parameter values ​​from the third party. Here, the third party refers to a logical entity relatively independent of the auxiliary diagnostic information providing device, such as a sample analyzer or a laboratory information system.

[0055] It should be noted that the form of the second abnormal feature information, which characterizes abnormal parameter values ​​in the biological sample based on the sample detection parameters, can be set in various formats to meet the reading needs of different types of users. For example, for patient users, a format that is easier to read is usually a parameter value that is higher or lower than the standard parameter; for doctor users, it is more desirable to see the numerical values ​​of a parameter and the corresponding standard parameter values.

[0056] The output module 114a is configured to output an auxiliary diagnostic report based on the sample analysis image, the first abnormal feature information, and the second abnormal feature information. The auxiliary diagnostic report includes: auxiliary diagnostic category information; an activation map for highlighting the diagnostic basis of the first abnormal feature information, wherein the activation map is formed by superimposing a heat map obtained based on the feature map of the sample analysis image and the sample analysis image; wherein the heat map characterizes the degree of influence of each unit region of the sample analysis image on the decision to determine the first abnormal feature information; and the sample analysis image carrying a positioning marker, wherein the positioning marker is used to indicate the position of the first abnormal feature information in the sample analysis image.

[0057] The auxiliary diagnostic report can be set to various formats, such as simplified and detailed versions. Multiple display sections can be set up within the report to display different content as needed. In this embodiment, the auxiliary diagnostic report includes at least an analysis display area that specifically displays auxiliary diagnostic category information, and an image display area that specifically displays activation maps and sample analysis images with positioning markers. Please refer to [link to relevant documentation]. Figure 4This is a schematic diagram of the interface of an auxiliary diagnostic report provided in one embodiment. Auxiliary diagnostic category information refers to the conclusion information of the unhealthy type of the subject to be tested, determined by the auxiliary diagnostic information providing device after comprehensive analysis of the obtained sample detection data and the identified first and second abnormal feature information. Specifically, the auxiliary diagnostic information providing device obtains the auxiliary diagnostic category information based on the sample analysis image, the first abnormal feature information, and the second abnormal feature information by inputting the sample analysis image, the first abnormal feature information, and the second abnormal feature information into a diagnostic database constructed based on historical diagnostic cases, and obtaining the auxiliary diagnostic category information through the correspondence between diagnostic categories and conditions provided in the diagnostic database. Optionally, the auxiliary diagnostic category information also includes treatment suggestions corresponding to the currently obtained conclusion information of the unhealthy type.

[0058] A heatmap can be a feature map obtained from the feature extraction network in a neural network model. It is generated by calculating the weight values ​​of each channel of the feature map output from the feature extraction network to the classification output layer, and then using a machine vision visualization algorithm to linearly weight and sum the image features output from each channel based on these weight values. The resulting image is displayed according to the linear weighted sum, i.e., the image of the region of interest corresponding to the classification result output by the neural network model. In an optional specific example, the machine vision visualization algorithm refers to the Grad-CAM algorithm. The Grad-CAM algorithm utilizes the backpropagation characteristic between adjacent network layers in a neural network model to achieve training and learning. It is compatible with different types of neural network models (e.g., deep convolutional neural networks VGG, residual neural networks ResNet) and can obtain heatmaps without retraining. In this embodiment, the Grad-CAM algorithm calculates the weight values ​​of each channel of the feature map using the backpropagation gradient of the classification convolutional neural network (CNN), as shown in Formula 1 below:

[0059]

[0060] In Formula 1, c represents the category, yc is the logits corresponding to that category (i.e., the value before passing through the classification output layer (Softmax layer)), A represents the feature map output by the convolution (image features output by the feature extraction network), k represents the number of channels in the feature map, ij represents the horizontal and vertical coordinates of the feature map, and Z represents the size of the feature map (i.e., length multiplied by width). The Grad-CAM algorithm's process of obtaining a heatmap based on the input image is equivalent to calculating the mean of the gradients on the feature map of the input image, performing a global average pooling operation. After obtaining the weights, the channels of the feature map are linearly weighted and fused together to obtain the heatmap, as shown in Formula 2 below:

[0061]

[0062] In Formula 2, Grad-CAM adds a linear weighting operation (ReLU function) to the fused heatmap, retaining only the regions that have a positive effect on category c, thus obtaining the image of the region of interest corresponding to the classification result output by the neural network model. The activation map is obtained by overlaying the heatmap and the sample analysis image. In this way, the activation map can more intuitively show the importance of different regions in the sample analysis image for obtaining the first abnormal feature information. It can not only make full use of the neural network model to identify and analyze the sample analysis image to compensate for the bias caused by individual differences of doctors, but also make full use of the neural network model to realize more image features that are not discernible to the human eye to improve the accuracy and precision of recognition. More importantly, it can obtain the basis for the neural network model to obtain the first abnormal feature information based on the sample analysis image, and present this basis in a visual way in the auxiliary diagnostic report.

[0063] The sample analysis image carrying positioning markers indicates the location of the first abnormal feature information in the sample analysis image. This makes it easier to identify the key areas of focus in the sample analysis image based on the image area defined by the positioning markers, and also facilitates a quick assessment of the accuracy of the first abnormal feature information obtained.

[0064] In this embodiment, the sample analysis image refers to an image formed by counting and detecting cells in a biological sample taken from the object to be tested. Depending on the counting and detection principle used, the image represents the changes in electrical or optical parameters that characterize the differences among cells in the biological sample. It identifies different cell types and obtains cell counts, distribution, and other cellular characteristics through these differences. The sample analysis image differs from cell morphology images, cardiovascular images, and human tissue images obtained from cameras, microscopes, or endoscopes, which represent the morphology of substances. These cell imaging images are still a form of simulated representation data of the biological sample. In this embodiment, the sample analysis image, through the detection and analysis of the biological sample, can obtain quantitative representation results that more accurately characterize the differences, quantity, and distribution of cells in the biological sample, completing the conversion of the biological sample's representation from analog to digital. The sample analysis image carries information about the digitized biological sample characteristics. The auxiliary diagnostic information providing device uses the sample analysis image as input to determine the first abnormal feature information, which can uncover abnormal features in the biological sample that cannot be identified solely through cell imaging images, thereby effectively improving the accuracy and reference value of the auxiliary diagnostic results. In the auxiliary diagnostic information providing device, the first abnormal feature information obtained through the first abnormality recognition module can identify abnormalities in biological sample features that cannot be obtained from cell imaging images in the sample analysis image. The activation map obtained by overlaying the heatmap and the sample analysis image can characterize the degree of influence of each image region in the sample analysis image on the identification result of abnormalities in biological sample features, converting information that is difficult for the human eye to discern but is extracted intelligently by computer into a visual result. The auxiliary diagnostic report displays the auxiliary diagnostic category information of the biological sample, the activation map, and the sample analysis image carrying positioning markers. This allows for the rapid location of biological sample features corresponding to the abnormal feature information, such as the cell type, morphology, and quantity distribution in the biological sample. Furthermore, the reliability of the currently obtained auxiliary diagnostic category information can be analyzed and judged based on the characterization of the decision influence degree of the activation map.

[0065] In the above embodiments, the auxiliary diagnostic information providing device obtains first abnormal feature information characterizing abnormal features in the sample analysis image by analyzing and recognizing the sample analysis image, and obtains second abnormal feature information characterizing abnormal parameter values ​​in the biological sample based on the sample detection parameters. Based on the sample analysis image, the first abnormal feature information, and the second abnormal feature information, it outputs an auxiliary diagnostic report. The auxiliary diagnostic report includes auxiliary diagnostic category information, a sample analysis image with the location marked by the first abnormal feature information, and an activation map highlighting the diagnostic basis of the first abnormal feature information obtained from the sample analysis image. Thus, users can obtain auxiliary diagnostic category information determined by comprehensive information from multiple aspects through the auxiliary diagnostic report, fully utilizing computer intelligence to recognize the sample analysis image to uncover more in-depth image information containing biological sample characteristics, improving the accuracy and efficiency of the auxiliary diagnostic results. Furthermore, the activation map and the location-marked sample analysis image output in the auxiliary diagnostic report allow doctors to quickly verify the accuracy of the current auxiliary diagnostic category information, thereby considering whether to adopt the current auxiliary diagnostic category information.

[0066] In some embodiments, the auxiliary diagnostic report further includes:

[0067] A target reference sample image that matches the sample analysis image; the target reference sample image is a reference image in the sample analysis image library whose similarity to the sample analysis image meets preset conditions; the sample analysis image library contains sample analysis images of historical diagnostic cases confirmed by microscopic examination.

[0068] The auxiliary diagnostic report also includes a target reference sample image displayed in the image display area. The sample analysis image library is a pre-established image database. Historical diagnostic cases refer to diagnostic records that determine whether a biological sample is abnormal and to identify the type of unhealthy condition of the tested individual based on sample testing data, clinical information data, and corresponding standard parameter tables. For example, in the historical diagnostic records of user A, if the biological sample collected from user A is a blood sample, and based on the whole blood test performed by user A at time 1, the whole blood test record a1 includes the blood cell analysis data output by the blood analyzer for user A's blood sample at time 1, the clinical information data entered into the laboratory information system when user A's blood sample was tested at time 1, and the comprehensive judgment made by the laboratory physician using the aforementioned data. After microscopic examination confirms that user A underwent a whole blood test at time 1, the whole blood test record a1 can be used as a historical diagnostic case in the diagnostic case database. By pre-establishing a sample analysis image library from sample analysis images included in historical diagnostic cases confirmed by microscopic examination, the auxiliary diagnostic information providing device obtains sample analysis images of biological samples of the object to be tested during the auxiliary diagnostic process. Based on the similarity matching between the sample analysis images and reference images in the sample analysis image library, the reference images that meet the preset similarity conditions are determined as target reference sample images.

[0069] The reference image that meets the preset similarity criteria can be the reference image with the highest similarity among all reference images in the sample analysis image library; or it can be the reference image in the sample analysis image library that first reaches a similarity value higher than a threshold during the matching process with the sample analysis image. Optionally, there can be multiple target reference sample images. Determining the target reference sample images that match the current sample analysis image refers to determining the multiple reference images in the sample analysis image library that have the highest relative similarity to the sample analysis image.

[0070] Optionally, the target reference sample image is a reference image carrying location markers. The methods for adding location markers to the target reference sample image can include: First, historical diagnostic cases serve as sample data upon which the auxiliary diagnostic information providing device is constructed. The sample analysis images contained within these cases serve as training data for the neural network model and, after annotation, inherently carry location markers; that is, each reference image in the sample analysis image library inherently carries a location marker. Second, the sample analysis images contained in the historical diagnostic cases do not inherently carry location markers. During the auxiliary diagnostic process, after obtaining sample analysis images carrying location markers, the auxiliary diagnostic information providing device searches for reference images in the sample analysis image library that meet preset similarity conditions, and then adds location markers to the corresponding positions in the matching reference images based on the positions of the location markers in the sample analysis images. Third, the activation map forming module of the auxiliary diagnostic information providing device performs gradient backpropagation analysis on the target reference sample image based on the classification labels in the first abnormal feature information to obtain a heatmap, and then uses a location marker module to form location markers on the target reference sample image based on the heatmap. Similarly, the location markers for the target state reference image can also be created using the above methods.

[0071] In the above embodiments, the auxiliary diagnostic report further displays a target reference sample image, which facilitates comparison between the target reference sample image and the currently obtained sample analysis image carrying the positioning mark. Since the target reference sample image is a sample analysis image contained in historical diagnostic cases that have been confirmed by microscopic examination, it can serve as a reliable basis for judgment, thereby making it easier for doctors to refer to and judge the accuracy of the currently obtained first abnormal feature information.

[0072] In some embodiments, the output module 114a further includes:

[0073] When the category of the first abnormal feature information is the first type, the auxiliary diagnostic report includes at least two target reference sample images whose similarity meets preset conditions, and the at least two target reference samples have the same degree of abnormality.

[0074] When the category of the first abnormal feature information is the second type, the auxiliary diagnosis includes the target reference sample image with the highest similarity and at least one target state reference image, wherein the abnormality level of the target reference sample image and the target state reference image is different.

[0075] The sample analysis image library can be divided into different image sub-libraries according to different categories of the first abnormal feature information. These sub-libraries can be specifically configured to contain reference images with multiple abnormality severity level labels or reference images containing the same abnormality severity level label. For cases where the first abnormal feature information belongs to the first type, if the image sub-library corresponding to the first type contains reference images with the same abnormality severity level label, then multiple target reference sample images with the highest relative similarity and the same abnormality severity level are retrieved from the sample analysis image library and displayed in the auxiliary diagnostic report. For cases where the first abnormal feature information belongs to the second type, if the image sub-library corresponding to the second type contains reference images with different abnormality severity level labels, then the target reference sample image with the highest relative similarity and target state reference images with different abnormality severity levels are retrieved from the sample analysis image library and displayed in the auxiliary diagnostic report.

[0076] For example, the first type includes platelet aggregation. When the first abnormal feature information corresponding to the sample analysis image is platelet aggregation, the auxiliary diagnostic information providing device can search for multiple target reference sample images with the highest similarity to the sample analysis image from the sample analysis image library and display the searched multiple target reference sample images in the auxiliary diagnostic report. The second type includes other abnormal feature information besides the first type. Optionally, the second type includes other abnormal information besides platelet aggregation. For example, the second type may include the presence of reactive lymphocytes, primitive cells, immature granulocytes, etc.

[0077] In the above embodiments, the sample analysis image library is configured to match the corresponding reference images with the recognition results of first abnormal feature information of different categories. For some case types, there may be situations where the degree of abnormality is not distinguished, or the currently collected historical diagnostic cases are insufficient to distinguish multiple degrees of abnormality, or doctors do not need to refer to multiple degrees of abnormality, etc. Therefore, the image sub-library corresponding to the first abnormal feature information of some categories contains reference images that do not distinguish the degree of abnormality. At the same time, for some case types, the image sub-library of the first abnormal feature information of the corresponding category that needs to be considered when diagnosing a suspected case type contains reference images of different degrees of abnormality. This allows the auxiliary diagnostic information providing device to match multiple reference images of different degrees of abnormality in the sample analysis image library, which is beneficial to improving the accuracy of the auxiliary diagnostic results and effectively enhancing the reference value of the auxiliary diagnostic report for doctors.

[0078] Optionally, the auxiliary diagnostic report may also include:

[0079] A target state reference image that matches the sample analysis image but has an anomaly level different from the target reference sample image;

[0080] The anomaly level of the target reference sample image is one of severe, moderate, and mild, and the anomaly level of the target state reference image includes the other two of severe, moderate, and mild.

[0081] As described in the previous embodiment, in practical applications, the reference image in the auxiliary diagnostic report for matching the sample analysis image may only include the target reference sample image without distinguishing the degree of abnormality. In this embodiment, the reference image in the auxiliary diagnostic report for matching the sample analysis image includes reference images with abnormality levels of severe, moderate, and mild. Specifically, within the sample analysis image library pre-established based on the sample analysis images included in historical diagnostic cases, corresponding image sub-libraries are established for different categories of first abnormality feature information. The corresponding image sub-libraries for the preset category of first abnormality feature information contain sample analysis images with abnormality levels of severe, moderate, and mild as reference images. In the auxiliary diagnostic report, in addition to displaying the target reference sample image in the image display area, the auxiliary diagnostic information providing device further displays target state reference images with different abnormality levels, providing more comparison angles for the currently obtained sample analysis image carrying positioning markers. This allows doctors to make more reliable judgments directly based on the reference images with different abnormality levels output by the auxiliary diagnostic report.

[0082] The target state reference map includes a first target state reference map and a second target state reference map corresponding to two different levels of anomaly. If the anomaly level of the target reference sample map is severe, then the anomaly levels of the first target state reference map and the second target state reference map are moderate and mild, respectively. If the anomaly level of the target reference sample map is moderate, then the anomaly levels of the first target state reference map and the second target state reference map are severe and mild, respectively. If the anomaly level of the target reference sample map is mild, then the anomaly levels of the first target state reference map and the second target state reference map are severe and moderate, respectively.

[0083] It should be noted that the target state reference image can also be a reference image carrying location markers. The method of adding location markers to the target state reference image can be the same as the method of adding location markers to the target reference sample image. That is, the sample analysis images included in the historical diagnostic cases on which the sample analysis image library was based originally contained location markers; or, after finding a matching reference image in the sample analysis image library based on the currently obtained sample analysis image carrying location markers, location markers can be added to the matching reference image according to the position of the location markers in the sample analysis image.

[0084] In the above embodiments, the auxiliary diagnostic report further displays a target state reference image. The target state reference image and the target reference sample image together form a combination of reference images with abnormality levels of severe, moderate and mild. This allows doctors to make more reliable judgments based on the reference images with different abnormality levels output by the auxiliary diagnostic report. This maximizes the value of the process of obtaining auxiliary diagnostic conclusions by identifying sample analysis images based on machine learning, while ensuring that reliable references are provided.

[0085] Optionally, the auxiliary diagnostic information providing device further includes:

[0086] The image matching module is used to determine a matching image set in the sample analysis image library based on the category of the first abnormal feature information; determine a target reference sample image whose similarity meets a preset condition based on the similarity between the sample analysis image and each reference image in the matching image set; and determine a target state reference image in the sample analysis image library that has the same mode category as the target reference sample image but a different abnormality level than the target reference sample image based on the mode category to which the target reference sample image belongs.

[0087] Among them, for the same category of first abnormal feature information, sample analysis images that obtain the same first abnormal feature information may contain the same biometric information, but they may also have display differences due to differences in the output images of different sample analyzers and / or individual differences of the objects to be detected, such as differences in color depth. In order to eliminate the influence of such display differences on the error when searching for target state reference images and target reference sample images in the sample analysis image library, the sample analysis image library is set as an image set divided according to the category of first abnormal feature information. Reference images under different image sets are further distinguished according to pattern categories, and reference images with different abnormality levels are contained under the same pattern category.

[0088] An image set can be a collection of reference images formed based on different classification conditions, such as different sample attributes or features. For the same reference image within the sample analysis image library, it may belong to image set A under classification condition 1, and simultaneously to image set B under classification condition 2. Determining the matching image set in the sample analysis image library based on the category of the first abnormal feature information can refer to using the category of the first abnormal feature information as the classification condition to identify the image set from the sample analysis image library. Pattern category can be another dimension for classifying reference images within the sample analysis image library. Pattern category can be based on the feature similarity between different reference images in the image set. The features used for classification can be one or more, such as classification based on the attribute features (clinical information, microscopic information) of the historical diagnostic cases from which the reference image originates; or classification based on image features, determining the classification result under the pattern category by calculating the distance between image features of each reference image; or classification by combining attribute features and image feature distance, etc.

[0089] In an optional specific example, the sample analysis image library includes image sub-libraries established based on first anomaly feature information of different categories. The image sub-libraries corresponding to the first anomaly feature information of the same category are divided into multiple pattern sub-libraries based on feature similarity. Each pattern sub-library includes reference images of different anomaly severity levels. The image sub-libraries corresponding to the first anomaly feature information of the same category are further divided into multiple pattern sub-libraries by determining the pattern categories through feature similarity analysis of the reference images within the image sub-libraries. Within the sample analysis image library, the image sub-library can be used as a first-level directory, and the pattern sub-library as a next-level directory under the image sub-library. Before outputting the auxiliary diagnostic information providing device, the auxiliary diagnostic information first determines the matching image sub-library based on the category of the first abnormal feature information through the first-level directory. Within the scope of the matching image sub-library, a reference image whose similarity meets the preset conditions is searched as the target reference sample image. Then, based on the pattern sub-library where the target reference sample image is located, if the abnormality level of the target reference sample image is A1 (A1 is one of severe, moderate, and mild), then reference images with abnormality levels of A2 and A3 (A2 and A3 are the other two of severe, moderate, and mild) are selected from the pattern sub-library where the target reference sample image is located as the target state reference image. In this embodiment, the image matching module first determines a target reference sample image based on the maximum value of the overall image similarity between the currently obtained sample analysis image and reference images in the matching image sub-library; then, within the matching pattern sub-library where the target reference sample image is located, it determines two other target state reference images with an anomaly level different from the target reference sample image based on the maximum value of the local image similarity of the currently obtained sample analysis image and reference images in the matching pattern sub-library within a preset interest region. In an optional embodiment, the preset interest region can be the region where the positioning marker is located in the sample analysis image. In another optional example, the image matching module first determines a target reference sample image based on the maximum value of the overall image similarity between the currently obtained sample analysis image and reference images in the matching image sub-library; then, within the matching pattern sub-library where the target reference sample image is located, it randomly selects two other target state reference images with an anomaly level different from the target reference sample image.

[0090] In the above embodiments, by setting the sample analysis image library to divide the image set according to the category of the first abnormal feature information and dividing the image set according to the pattern category, it is beneficial to improve the accuracy of the reference image combination (target reference sample image and target state reference image combination) provided in the auxiliary diagnostic report for severe, moderate and mild conditions.

[0091] It should be noted that in the various embodiments where the target reference sample image and the target state reference image are output in the auxiliary diagnostic report, at least one of the target reference sample image and the target state reference image carries a positioning marker. That is, the output display of the reference image matching the sample analysis image in the auxiliary diagnostic report includes the following situations: outputting only one or more target reference sample images, with or without positioning markers; outputting one target reference sample image and one or more target state reference images, with or without positioning markers in the target reference sample image and the target state reference image. In the embodiments where the target reference sample image and the target state reference image contain positioning markers, the method of adding the positioning markers has been described in the foregoing embodiments and will not be repeated here.

[0092] In some embodiments, the first anomaly identification module includes: an anomaly feature classification module, which determines the classification label of the anomaly features in the sample analysis image through an image classification neural network model to obtain the first anomaly feature information;

[0093] The auxiliary diagnostic information providing device further includes:

[0094] The activation map forming module is used to perform gradient backpropagation analysis based on the classification labels in the first abnormal feature information to obtain the heat map. The heat map is then magnified and superimposed on the sample analysis image to obtain the activation map.

[0095] A positioning marker module is used to form positioning markers on the sample analysis image based on the highlighted areas of the activation map.

[0096] In one optional example, the auxiliary diagnostic information providing device obtains an activation map and a sample analysis image carrying localization markers from an auxiliary diagnostic report based on a sample analysis image. This includes: using an image classification neural network model to identify the sample analysis image; outputting corresponding classification labels based on the identified abnormal features in the sample analysis image; using the Grad-CAM algorithm to utilize the weight values ​​of each channel of the feature map output to the classification output layer from the feature extraction network of the image classification neural network model; using a machine vision visualization algorithm to linearly weight and sum the image features output from each channel based on the weight values; obtaining a heatmap based on the linear weighted summation result; performing image matching between the heatmap and the sample analysis image and then superimposing them to obtain the activation map; identifying the bright areas in the heatmap; and constructing a minimum bounding rectangle based on the contours of the bright areas to form localization markers. The classification neural network model includes a feature extraction layer and a classification prediction layer. The sample analysis image is input into the classification neural network model, and the feature extraction layer extracts features from the sample analysis image. The feature extraction layer can have multiple layers, with the last layer connected to the classification prediction layer. The classification prediction layer performs classification prediction based on the feature vectors output by the feature extraction layer to determine the corresponding classification label.

[0097] In some embodiments, the activation map forming module may also employ algorithms such as GAP, CAM, Grad-CAM++ to perform gradient backpropagation analysis based on the classification labels in the first anomaly feature information to obtain a heat map.

[0098] It should be noted that the auxiliary diagnostic information providing device can also acquire sample analysis images carrying location markers in different ways. For example, in another optional example, the auxiliary diagnostic information providing device uses an object detection neural network to identify the sample analysis image, outputs corresponding classification labels based on the identified abnormal features in the sample analysis image, and outputs the sample analysis image carrying location markers. The object detection neural network model includes convolutional layers and RPN (region proposal networks) network layers. The convolutional layers extract features from the input image to obtain feature maps, and the RPN (region proposal networks) network layers use prior anchors to output a set of rectangular candidate regions with objectness information. Based on the set of rectangular candidate regions, the image region where the corresponding abnormal feature is located is selected. In yet another optional example, the auxiliary diagnostic information providing device uses an image segmentation neural network to detect whether there are abnormal features in the sample analysis image, obtains the target detection result of the local image where the abnormal feature is located, and determines the position of the location marker in the sample analysis image based on the target detection result of the local image. In another alternative example, a location marker can be formed based on a target detection algorithm for the region of interest, peak finding, valley finding, etc. In this example, the activation map can be generated based on the particle density in the third dimension within the region where the location marker is located. Here, the third dimension is another dimension that is different from the display plane of the activation map. For example, when the sample analysis image includes three dimensions: forward scattered light, side scattered light, and side fluorescence, if the activation map displayed in the auxiliary diagnostic report has two dimensions: side scattered light and side fluorescence, then the third dimension can be forward scattered light. If the activation map has two dimensions: forward scattered light and side fluorescence, then the third dimension can be side scattered light.

[0099] It should be noted that the location markers are used to highlight the key areas of interest in the sample analysis image from which the first anomalous feature information is obtained. These markers can take the form of bounding boxes of various regular or irregular geometric shapes, or they can be image regions marked with different colors. In an optional specific example, the location marker refers to a rectangular bounding box.

[0100] In the above embodiments, a neural network model is used to process the sample analysis images, making full use of the advantages of the neural network model in image recognition, completing the intelligent recognition of sample analysis images to replace the brain of professional laboratory physicians, and at the same time making the significant features of the images in the output results visible.

[0101] In some embodiments, the sample analysis image includes at least one of a blood cell histogram, a blood cell scatter plot, a pulse signal waveform, a pulse signal feature plot, a thromboelastography, and a biochemical or immune response curve.

[0102] The auxiliary diagnostic information providing device can communicate and connect with different types of sample analyzers to obtain different types of sample analysis images as input. Among these, blood cell histograms can include histograms corresponding to different blood cells, such as platelet histograms, erythrocyte histograms, leukocyte histograms, and histograms formed by pulse signal characteristic values ​​such as peak height, front peak width, back peak width, and half-peak width (e.g., RBC, PLT, and WBC histograms formed by pulse signal peak height). Blood cell scatter plots can include different types of scatter plots obtained by detecting light scattering in different directions of each blood cell, converting the light signal into electrical pulses, such as leukocyte DIFF scatter plots, RET scatter plots, two-dimensional, three-dimensional, or even multi-dimensional scatter plots composed of characteristic values ​​of pulse signals in several dimensions (e.g., a DIFF three-dimensional scatter dataset composed of the peak heights of SFL, SSC, and FSC three-channel pulse signals), and two-dimensional scatter plot projections formed from three-dimensional and high-dimensional scatter data. Pulse signal waveforms can be, for example, waveforms formed by the raw pulse signals collected by a hematology analyzer during blood sample testing, or waveforms formed by pulse signals filtered by a pulse recognition algorithm. Pulse signal feature maps can be, for example, vector feature maps composed of pulse feature values ​​such as peak height, initial peak width, subsequent peak width, and half-peak width. Sample analyzers can also include coagulation analyzers, immunoassay analyzers, etc., and the corresponding sample analysis images can also include thromboelastography, biochemical or immune response curves. The auxiliary diagnostic information providing device can use sample analysis images output by one or more sample analyzers as input, identify the sample analysis images, and output the first abnormal feature information of the abnormal features carried in each sample analysis image.

[0103] In the above embodiments, the auxiliary diagnostic information providing device works in conjunction with the sample analyzer to obtain and analyze the sample test data output by the sample analyzer, and can accurately determine whether there are abnormalities in the biological sample of the object to be tested, so as to provide auxiliary diagnosis for laboratory physicians.

[0104] Please see Figure 5Another embodiment of this application provides an auxiliary diagnostic information providing device, including: an acquisition module 111, a first anomaly recognition module 112, a second anomaly recognition module 113, and an output module 115. The acquisition module 111 is used to acquire sample detection data obtained by analyzing biological samples of the object to be tested; the sample detection data includes sample analysis images and sample detection parameters; the first anomaly recognition module 112 is used to obtain first anomaly feature information characterizing abnormal features in the sample analysis image based on the sample analysis image; the second anomaly recognition module 113 is used to obtain second anomaly feature information characterizing abnormal parameter values ​​in the biological sample based on the sample detection parameters; the output module 114b is used to output an auxiliary diagnostic report based on the sample analysis image, the first anomaly feature information and the second anomaly feature information, the auxiliary diagnostic report including: auxiliary diagnostic category information; the sample analysis image carrying a positioning mark, the positioning mark being used to indicate the position of the first anomaly feature information in the sample analysis image; a target reference sample image matching the sample analysis image, the target reference sample image being a reference image in the sample analysis image library whose similarity to the sample analysis image meets preset conditions; the sample analysis image library includes sample analysis images of historical diagnostic cases confirmed by microscopic examination, the positioning mark in the target reference sample image being used to highlight the location of the diagnostic basis for obtaining the first anomaly feature information.

[0105] Among them, with Figure 1 The main difference in the illustrated embodiment is that the auxiliary diagnostic report output by the output module 114b can omit the activation map and output a target reference sample image that matches the currently obtained sample analysis image and is confirmed by microscopic examination.

[0106] In the above embodiments, the auxiliary diagnostic information providing device obtains first abnormal feature information characterizing abnormal features in the sample analysis image by analyzing and identifying the sample analysis image, and obtains second abnormal feature information characterizing abnormal parameter values ​​in the biological sample based on the sample detection parameters. Based on the sample analysis image, the first abnormal feature information, and the second abnormal feature information, it outputs an auxiliary diagnostic report. The auxiliary diagnostic report includes auxiliary diagnostic category information, a sample analysis image showing the location of the first abnormal feature information through positioning markers, and a target reference sample image obtained from historical diagnostic cases confirmed by microscopic examination. Thus, users can obtain auxiliary diagnostic category information determined by comprehensive information from multiple aspects through the auxiliary diagnostic report, improving the accuracy and efficiency of the auxiliary diagnostic results. Furthermore, the target reference sample image confirmed by microscopic examination and the sample analysis image with positioning markers output in the auxiliary diagnostic report allow doctors to quickly verify the accuracy of the current auxiliary diagnostic category information, determine its reliability, and consider whether to adopt the current auxiliary diagnostic category information.

[0107] In some embodiments, the auxiliary diagnostic report further includes:

[0108] A target state reference image that is matched with the sample analysis image but has an anomaly level different from the target reference sample image; wherein the anomaly level includes severe, moderate and mild.

[0109] The anomaly level of the target reference sample image is one of severe, moderate, and mild, and the anomaly level of the target state reference image includes the other two of severe, moderate, and mild.

[0110] In the above embodiments, the auxiliary diagnostic report further displays a target state reference image. The target state reference image and the target reference sample image together form a combination of reference images with abnormality levels of severe, moderate and mild. This allows doctors to make more reliable judgments based on the reference images with different abnormality levels output by the auxiliary diagnostic report. This maximizes the value of the process of obtaining auxiliary diagnostic conclusions by identifying sample analysis images based on machine learning, while ensuring that reliable references are provided.

[0111] In some embodiments, the auxiliary diagnostic information providing device further includes:

[0112] The reference image determination module is used to determine a matching image set in the sample analysis image library based on the category of the first abnormal feature information; and to determine a target reference sample image whose similarity meets a preset condition based on the similarity between the sample analysis image and the reference image in the matching image set; or, it is used to perform a search in the sample analysis image library based on similarity based on the sample analysis image, and determine a matching image set based on the category of the first abnormal feature information, and perform a second search in the matching image set based on similarity based on the two search results to determine a target reference sample image whose similarity meets a preset condition.

[0113] An image set can be a collection of reference images from a sample analysis image library, categorized according to different classification criteria. These criteria can include different sample attributes, sample features, etc. For the same reference image within the sample analysis image library, it can be assigned to different image sets based on different classification criteria. Determining the matching image set in the sample analysis image library based on the category of the first anomaly feature information can mean using the category of the first anomaly feature information as the classification criterion, selecting the image set from the sample analysis image library, and determining the reference image with the highest similarity or the top few reference images with the highest relative similarity as the target reference sample image based on the similarity between the current sample analysis image and each reference image in the matching image set.

[0114] Optionally, the auxiliary diagnostic information providing device may determine the target reference sample image by searching the sample analysis image library according to similarity based on the current sample analysis image, obtaining the first search result of the similarity between the sample analysis image and each reference image in real time, determining a matching image set according to the category of the first abnormal feature information, and searching the matching image set according to similarity based on the current sample analysis image, obtaining the second search result of the similarity between the sample analysis image and each reference image in real time, and determining one or more reference images whose similarity first reaches a set threshold as the target reference sample image based on the first search result and the second search result.

[0115] In the above embodiments, the sample analysis image library is configured to support different methods for finding target reference sample images, which is beneficial for supporting the needs of more application scenarios and improving search efficiency.

[0116] Please see Figure 6 In another aspect, this application also provides a blood analysis system, including:

[0117] Sampling component 211 is used to collect and distribute biological samples of the object to be tested, wherein the biological sample is a blood sample;

[0118] Reaction component 212 is used to process the biological sample to form a test solution;

[0119] Drive component 213 is used to drive the liquid path between the sampling component and the reaction component;

[0120] The detection component 214 is used to classify and count the blood cells contained in the test solution to obtain sample detection data including sample detection parameters and sample analysis images.

[0121] The auxiliary diagnostic information providing device 11 described in any embodiment of this application is used to output an auxiliary diagnostic report based on the sample detection data.

[0122] In this embodiment, the blood analysis system is a medical auxiliary diagnostic device that integrates the analysis of blood samples of the test subject to obtain sample detection images and parameters, and the identification of abnormal features in biological samples based on the sample detection images and parameters obtained from the analysis of blood samples, and outputs an auxiliary diagnostic report with visualized diagnostic evidence. The blood analysis system can be an upgrade of a blood analyzer, including a sampling component, a reaction component, a driving component, and a detection component for analyzing blood samples to obtain sample detection data. The auxiliary diagnostic information providing device, as a computer program product that implements auxiliary diagnostic functions based on computer program flow, can be stored in the memory of the blood analysis system and executed by the processor to implement the auxiliary diagnostic functions of the auxiliary diagnostic information providing device described in this embodiment.

[0123] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. An auxiliary diagnostic information providing device, characterized in that, include: The acquisition module is used to acquire sample detection data obtained by analyzing biological samples of the object to be tested; the sample detection data includes sample analysis images and / or sample detection parameters. The first anomaly detection module is used to obtain first anomaly feature information characterizing the abnormal features in the sample analysis image based on the sample analysis image, and / or The second anomaly identification module is used to obtain second anomaly feature information that characterizes the abnormal parameter values ​​in the biological sample, based on the sample detection parameters. The output module is configured to output an auxiliary diagnostic report based on the sample analysis image, the first abnormal feature information, and / or the second abnormal feature information, the auxiliary diagnostic report including: Information on auxiliary diagnostic categories; A target reference sample image that matches the sample analysis image, wherein the target reference sample image is a reference image in the sample analysis image library that meets a preset condition in terms of similarity to the sample analysis image; the sample analysis image library includes sample analysis images of historical diagnostic cases confirmed by microscopic examination; The auxiliary diagnostic report also includes: the sample analysis image carrying a positioning marker, the positioning marker being used to indicate the location of the first abnormal feature information in the sample analysis image; The auxiliary diagnostic information providing device generates the location markers for the sample analysis image using any of the following methods: Gradient backpropagation analysis is performed based on the classification labels in the first abnormal feature information to obtain a heatmap, and the highlighted areas in the heatmap are identified. A minimum bounding rectangle is constructed based on the contours of the highlighted areas to form the location markers for the sample analysis image; or The sample analysis image is identified using a target detection neural network. Based on the identified abnormal features in the sample analysis image, corresponding classification labels and sample analysis images carrying location markers are output; or An image segmentation neural network is used to detect whether there are abnormal features in the sample analysis image, and the target detection results of the local image corresponding to the abnormal features are obtained. Based on the target detection results of the local image, the position of the local marker in the sample analysis image is determined; or The localization markers of the sample analysis image are formed based on at least one of the target detection algorithm, peak finding algorithm, and valley finding algorithm of region of interest.

2. The auxiliary diagnostic information providing device as described in claim 1, characterized in that, When the category of the first abnormal feature information is the first type, the auxiliary diagnostic report includes at least two target reference sample images whose similarity meets preset conditions, and the at least two target reference sample images have the same level of abnormality. When the category of the first abnormal feature information is the second type, the auxiliary diagnosis includes the target reference sample image with the highest similarity and at least one target state reference image, wherein the abnormality level of the target reference sample image and the target state reference image is different.

3. The auxiliary diagnostic information providing device as described in claim 1, characterized in that, The auxiliary diagnostic report also includes: A target state reference image that is matched with the sample analysis image but has an anomaly level different from the target reference sample image; wherein the anomaly level includes severe, moderate and mild. The anomaly level of the target reference sample image is one of severe, moderate, and mild, and the anomaly level of the target state reference image includes the other two of severe, moderate, and mild.

4. The auxiliary diagnostic information providing device as described in claim 1, characterized in that, The auxiliary diagnostic information providing device further includes: a reference image determination module, configured to determine a matching image set in the sample analysis image library based on the category of the first abnormal feature information; determine a target reference sample image whose similarity meets a preset condition based on the similarity between the sample analysis image and a reference image in the matching image set; or, configured to perform a search in the sample analysis image library based on similarity based on the sample analysis image, determine a matching image set based on the category of the first abnormal feature information, perform a second search in the matching image set based on similarity, and determine a target reference sample image whose similarity meets a preset condition based on the results of the two searches.

5. The auxiliary diagnostic information providing device as described in claim 1, characterized in that, The auxiliary diagnostic report also includes a target status reference diagram; The reference image determination module is further configured to determine, based on the pattern category to which the target reference sample image belongs, the target state reference image in the sample analysis image library that has the same pattern category as the target reference sample image but a different anomaly level than the target reference sample image; Wherein, at least one of the target reference sample image and the target state reference image carries a positioning mark, and the positioning mark carried by at least one of the target reference sample image and the target state reference image is used to highlight the location of the diagnostic basis for obtaining the first abnormal feature information.

6. The auxiliary diagnostic information providing device as described in claim 1, characterized in that, The auxiliary diagnostic report also includes: The sample analysis image carries a positioning marker, which is used to indicate the location of the first abnormal feature information in the sample analysis image; The auxiliary diagnostic information providing device is also used to add corresponding positioning marks in the matching target reference sample image according to the position of the positioning marks in the sample analysis image.

7. An auxiliary diagnostic information providing device, characterized in that, include: The acquisition module is used to acquire sample analysis images obtained from the detection and analysis of biological samples of the object to be tested; The first anomaly identification module is used to obtain first anomaly feature information characterizing the abnormal features in the sample analysis image based on the sample analysis image; The output module is used to output an auxiliary diagnostic report based on the sample analysis image and the first abnormal feature information. The auxiliary diagnostic report includes: Information on auxiliary diagnostic categories; The sample analysis image carries a positioning marker, which is used to indicate the location of the first abnormal feature information in the sample analysis image; The auxiliary diagnostic information providing device generates the location markers for the sample analysis image using any of the following methods: Gradient backpropagation analysis is performed based on the classification labels in the first abnormal feature information to obtain a heatmap, and the highlighted areas in the heatmap are identified. A minimum bounding rectangle is constructed based on the contours of the highlighted areas to form the location markers for the sample analysis image; or The sample analysis image is identified using a target detection neural network. Based on the identified abnormal features in the sample analysis image, corresponding classification labels and sample analysis images carrying location markers are output; or An image segmentation neural network is used to detect whether there are abnormal features in the sample analysis image, and the target detection results of the local image corresponding to the abnormal features are obtained. Based on the target detection results of the local image, the position of the local marker in the sample analysis image is determined; or The localization markers of the sample analysis image are formed based on at least one of the target detection algorithm, peak finding algorithm, and valley finding algorithm of region of interest.

8. The auxiliary diagnostic information providing device as described in claim 7, characterized in that, The first anomaly detection module includes: The abnormal feature classification module determines the classification label of the abnormal features in the sample analysis image through an image classification neural network model, thereby obtaining the first abnormal feature information; The auxiliary diagnostic information providing device further includes: The activation map forming module is used to perform gradient backpropagation analysis based on the classification labels in the first abnormal feature information to obtain a heat map, and the heat map is magnified and superimposed on the sample analysis image to obtain the activation map; A positioning marker module is used to form positioning markers on the sample analysis image based on the highlighted areas of the activation map.

9. A blood analysis system, characterized in that, include: A sampling component for collecting and distributing biological samples, specifically blood samples, from the object to be tested. A reaction assembly for processing the biological sample to form a test solution; A driving component is used to drive the liquid path between the sampling component and the reaction component; The detection component is used to classify and count the blood cells contained in the test solution to obtain sample detection data containing sample detection parameters and / or sample analysis images. The auxiliary diagnostic information providing device as described in any one of claims 1 to 8 is used to output an auxiliary diagnostic report based on the sample detection data.

Images