Endocrine hormone multi-index parallel detection system and method based on AI pattern recognition
Patent Information
- Application Number
- CN202610788680.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-06-03
- Publication Date
- 2026-09-22
AI Technical Summary
[0005]鉴于此，本发明提出了一种基于AI模式识别的内分泌激素多指标并行检测系统及方法，旨在解决现有技术中存在多指标无法同步获取和图像特征利用不充分的问题
[0015]与现有技术相比，本发明的有益效果在于：本发明通过设置样本承载模块、激素反应模块、图像采集模块、特征提取模块以及识别模块，能够将待检测生物样本进行多通道并行检测，使不同内分泌激素在对应检测区域内同步发生反应，并通过同步采集各激素反应图像，实现多指标激素检测数据的统一获取；同时，通过提取颜色特征、灰度分布特征、纹理特征以及反应区域轮廓特征，并构建对应的激素图像特征矩阵，能够更加全面地表征不同激素反应区域的图像变化信息，提高不同激素之间的区分能力；进一步的，通过基于样本数据库进行特征匹配，可实现对激素指标的自动识别与检测，从而降低人工判读误差，提高内分泌激素检测的准确性、稳定性以及多指标并行检测效率。
Smart Images
Figure CN122800280A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of hormone detection technology, and more specifically, to a parallel detection system and method for multiple indicators of endocrine hormones based on AI pattern recognition. Background Technology
[0002] Endocrine hormone level testing has significant applications in clinical diagnosis, health assessment, and screening for related diseases. Common testing items include thyroid hormones, sex hormones, and metabolism-related hormones. Currently, most testing methods still focus on single-indicator detection, such as chemiluminescent immunoassay, enzyme-linked immunosorbent assay (ELISA), or colloidal gold immunochromatography. These methods typically perform separate reactions and result interpretations for each individual hormone, making the testing process relatively independent. In scenarios involving multiple indicators, multiple tests or step-by-step operations are often required, thus increasing the complexity and time cost of the testing process.
[0003] In addition, during the process of obtaining test results, some methods rely on manual interpretation or single-channel image acquisition to analyze the reaction results. This may be affected by factors such as fluctuations in imaging conditions, differences in operation time, and differences in human experience, which means that the interpretation stability under weak reaction or critical concentration sample conditions needs to be further improved.
[0004] Therefore, it is necessary to provide a parallel detection system and method for multiple indicators of endocrine hormones based on AI pattern recognition to solve the problems of the inability to acquire multiple indicators simultaneously and the insufficient utilization of image features in the existing technology. Summary of the Invention
[0005] In view of this, the present invention proposes a parallel detection system and method for multiple indicators of endocrine hormones based on AI pattern recognition, aiming to solve the problems of the inability to acquire multiple indicators simultaneously and the insufficient utilization of image features in the prior art.
[0006] On the one hand, this invention proposes a parallel detection system for multiple indicators of endocrine hormones based on AI pattern recognition, comprising: The sample carrying module is used to carry the biological sample to be tested and to partition the biological sample to be tested into multiple channels; The hormone reaction module is used to react the biological sample to be tested with the corresponding hormone detection reagent to form the corresponding hormone reaction image. The image acquisition module is used to simultaneously acquire images of various hormone responses to obtain images of hormone detection for several indicators. The feature extraction module is used to extract features from each hormone detection image to obtain feature parameters and construct the corresponding hormone image feature matrix; wherein, the feature parameters include color features, grayscale distribution features, texture features and reaction area contour features; The identification module is used to match the feature parameters based on the sample database to obtain hormone index detection results.
[0007] Furthermore, the sample carrying module, used to carry the biological sample to be tested and to perform multi-channel partitioning of the biological sample to be tested, includes: The biological sample to be tested is divided into multiple channels so that different channels correspond to different endocrine hormone detection items. The system includes an isolation structure between each partition and an identification coding area within each partition to establish a matching relationship between the corresponding hormone detection area and the detection image.
[0008] Furthermore, the image acquisition module is used to simultaneously acquire images of various hormone responses, and when obtaining images of several hormone indicators, it includes: Acquire hormone detection images of several indicators within the reaction time, and simultaneously acquire hormone detection images of each indicator at the same time. Each collected hormone detection image is timestamped.
[0009] Furthermore, the image acquisition module is used to simultaneously acquire images of various hormone responses, and after obtaining several indicator hormone detection images, it includes: The hormone detection images were subjected to noise filtering, brightness compensation, and color normalization. Based on the identified coding area, the hormone detection images of each indicator are processed for region localization and image alignment.
[0010] Furthermore, the feature extraction module, when extracting features from each hormone detection image to obtain feature parameters and constructing the corresponding hormone image feature matrix, includes: The mean and variance of the color components of the reaction region in the hormone detection image of each processed indicator were extracted as color features. The grayscale distribution characteristics are obtained by statistically analyzing the pixel distribution probability of each grayscale level within the reaction area. Texture features are extracted using the gray-level co-occurrence matrix; The reaction region is segmented based on the edge detection algorithm, and its perimeter, area, and roundness are calculated to obtain the contour features; The feature parameters are serialized and concatenated in a preset order to generate a feature vector corresponding to a single detection index. The feature vectors under all timestamps of the same detection indicator are serialized and concatenated to obtain the corresponding hormone image feature matrix.
[0011] Furthermore, the parallel detection system for multiple indicators of endocrine hormones also includes: The database storage module is used to store the sample database; The sample database is used to store reaction images, feature parameters, and historical test results of different endocrine hormones at different times.
[0012] Furthermore, when the recognition module matches the feature parameters based on the sample database, it includes: The feature vector marked by each time stamp in the same hormone image feature matrix is matched with the sample database, and the similarity is calculated; wherein, the reaction time corresponding to each time stamp is consistent with the reaction time in the historical detection results matched in the sample database; Historical detection results with similarity greater than a preset threshold are selected as candidate detection results; Statistical analysis of candidate detection results under different timestamps in the same hormone image feature matrix; Hormone indicator test results are determined based on the number of times the same historical test result appears in the candidate test results.
[0013] Furthermore, when determining the hormone indicator test result based on the number of occurrences of the same historical test result in the candidate test results, it includes: The candidate test results are sorted from highest to lowest frequency, and the candidate test result with the highest frequency is selected as the final test result for the corresponding hormone indicator. If there are multiple candidate detection results with the same number of occurrences and all being the highest, and the number of occurrences is not 1, then calculate the average similarity between the feature vectors corresponding to the multiple candidate detection results and the feature vector to be detected, and select the candidate detection result with the highest average similarity as the final detection result.
[0014] Furthermore, when determining the hormone index test result based on the number of occurrences of the same historical test result in the candidate test results, it also includes: If the number of candidate test results is zero, then the hormone indicator test result is determined to be undetectable; If the number of candidate test results is greater than zero, and each historical test result in the candidate results appears once, then the hormone index test result is determined to be undetectable. When the hormone indicator test result is not detected, manual testing is triggered, and the manual test result is entered. The hormone image feature matrix of the biological sample to be tested and the manual test result are marked as new sample data and stored in the sample database.
[0015] Compared with existing technologies, the beneficial effects of this invention are as follows: By setting up a sample carrying module, a hormone response module, an image acquisition module, a feature extraction module, and a recognition module, this invention can perform multi-channel parallel detection of biological samples to be tested, enabling different endocrine hormones to react synchronously in their corresponding detection areas. Furthermore, by synchronously acquiring images of each hormone response, unified acquisition of multi-index hormone detection data is achieved. Simultaneously, by extracting color features, grayscale distribution features, texture features, and response area contour features, and constructing a corresponding hormone image feature matrix, it can more comprehensively characterize the image change information of different hormone response areas, improving the ability to distinguish between different hormones. Moreover, by performing feature matching based on a sample database, automatic identification and detection of hormone indicators can be achieved, thereby reducing human interpretation errors and improving the accuracy, stability, and efficiency of multi-index parallel detection in endocrine hormone testing.
[0016] On the other hand, this application also provides a parallel detection method for multiple indicators of endocrine hormones based on AI pattern recognition, including: It carries the biological sample to be tested and divides the biological sample to be tested into multiple channels; The biological samples to be tested are reacted with the corresponding hormone detection reagents to form corresponding hormone reaction images; Simultaneously acquire images of various hormone responses to obtain images of hormone detection for several indicators; Feature extraction is performed on the hormone detection images for each indicator to obtain feature parameters, and a corresponding hormone image feature matrix is constructed; wherein, the feature parameters include color features, grayscale distribution features, texture features, and reaction area contour features; The feature parameters are matched based on the sample database to obtain hormone index detection results.
[0017] It is understood that the AI pattern recognition-based parallel detection system and method for multiple indicators of endocrine hormones provided in this application have the same beneficial effects, which will not be elaborated here. Attached Figure Description
[0018] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings: Figure 1 A functional block diagram of a parallel detection system for multiple indicators of endocrine hormones based on AI pattern recognition provided in an embodiment of the present invention; Figure 2 The flowchart illustrates a parallel detection method for multiple indicators of endocrine hormones based on AI pattern recognition, as provided in an embodiment of the present invention. Detailed Implementation
[0019] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the disclosure to those skilled in the art. It should be noted that, unless otherwise specified, embodiments and features in the embodiments of the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.
[0020] In some embodiments of this application, see Figure 1 As shown, this embodiment provides a parallel detection system for multiple indicators of endocrine hormones based on AI pattern recognition, including: The sample carrying module is used to carry the biological sample to be tested and to partition the biological sample to be tested into multiple channels; The hormone reaction module is used to react the biological sample to be tested with the corresponding hormone detection reagent to form the corresponding hormone reaction image. The image acquisition module is used to simultaneously acquire images of various hormone responses to obtain images of hormone detection for several indicators. The feature extraction module is used to extract features from each hormone detection image to obtain feature parameters and construct the corresponding hormone image feature matrix; wherein, the feature parameters include color features, grayscale distribution features, texture features and reaction area contour features; The identification module is used to match the feature parameters based on the sample database to obtain hormone index detection results.
[0021] It is understood that this invention, by setting up a sample carrying module, a hormone response module, an image acquisition module, a feature extraction module, and a recognition module, can perform multi-channel parallel detection of biological samples to be tested, enabling different endocrine hormones to react synchronously in their corresponding detection areas. By simultaneously acquiring images of each hormone response, unified acquisition of multi-indicator hormone detection data is achieved. Simultaneously, by extracting color features, grayscale distribution features, texture features, and reaction area contour features, and constructing a corresponding hormone image feature matrix, the image change information of different hormone response areas can be more comprehensively characterized, improving the ability to distinguish between different hormones. Furthermore, by performing feature matching based on a sample database, automatic identification and detection of hormone indicators can be achieved, thereby reducing human interpretation errors and improving the accuracy, stability, and efficiency of multi-indicator parallel detection of endocrine hormones.
[0022] In some specific embodiments of this application, the sample carrying module is used to carry the biological sample to be tested, and when performing multi-channel partitioning on the biological sample to be tested, it includes: The biological sample to be tested is divided into multiple channels so that different channels correspond to different endocrine hormone detection items. The system includes an isolation structure between each partition and an identification coding area within each partition to establish a matching relationship between the corresponding hormone detection area and the detection image.
[0023] Understandably, by dividing the biological sample to be tested into multiple channels, with each zone corresponding to a different endocrine hormone detection item, it is possible to achieve simultaneous detection of multiple hormone indicators and improve overall detection efficiency. At the same time, by setting up isolation structures between each zone, cross-interference caused by reagent diffusion, liquid penetration, or reaction signal crosstalk between different detection areas can be reduced, thereby improving the independence and accuracy of each hormone detection result.
[0024] Furthermore, by setting up identification coding areas within each partition and establishing a matching relationship between hormone detection areas and detection images, automatic region positioning and corresponding recognition can be achieved in subsequent image acquisition processes, reducing the probability of image matching errors and improving the stability and reliability of multi-index image recognition and detection result analysis.
[0025] In some specific embodiments of this application, the image acquisition module is used to simultaneously acquire images of various hormone responses to obtain images of several indicator hormones, including: Acquire hormone detection images of several indicators within the reaction time, and simultaneously acquire hormone detection images of each indicator at the same time. Each collected hormone detection image is timestamped.
[0026] Understandably, by synchronously acquiring hormone detection images of various indicators at the same time and adding timestamps to each image, it is possible to ensure that the response data of all detection indicators remain synchronized in the time dimension. This facilitates subsequent unified analysis of the characteristic changes of different hormones in the same reaction process, avoids feature extraction errors caused by time differences, and ensures the consistency of data for all detection indicators. At the same time, the timestamps also provide a time-dimensional index for subsequent time-segmented feature matching, facilitating feature comparison step by step according to the reaction process, and further improving the accuracy of matching and recognition.
[0027] In some specific embodiments of this application, the image acquisition module is used to simultaneously acquire images of various hormone responses, and after obtaining images of several hormone indicators, it includes: The hormone detection images were subjected to noise filtering, brightness compensation, and color normalization. Based on the identified coding area, the hormone detection images of each indicator are processed for region localization and image alignment.
[0028] Understandably, noise filtering, brightness compensation, and color normalization can eliminate image interference caused by equipment noise and lighting fluctuations during the acquisition process, unify the image color benchmark of different detection areas, and avoid the impact of environmental factors on the accuracy of subsequent feature extraction. Furthermore, region localization and image alignment based on the identifier coding area can quickly locate the reaction area corresponding to each indicator, correct image offsets that may occur during the acquisition process, and ensure that subsequent feature extraction comes from the reaction area of the corresponding hormone, further improving the accuracy of feature extraction and avoiding feature interference from irrelevant areas.
[0029] In some specific embodiments of this application, when the feature extraction module is used to extract features from each indicator hormone detection image to obtain feature parameters and construct the corresponding hormone image feature matrix, it includes: The mean and variance of the color components of the reaction region in the hormone detection image of each processed indicator were extracted as color features. The grayscale distribution characteristics are obtained by statistically analyzing the pixel distribution probability of each grayscale level within the reaction area. Texture features are extracted using the gray-level co-occurrence matrix; The reaction region is segmented based on the edge detection algorithm, and its perimeter, area, and roundness are calculated to obtain the contour features; The feature parameters are serialized and concatenated in a preset order to generate a feature vector corresponding to a single detection index. The feature vectors under all timestamps of the same detection indicator are serialized and concatenated to obtain the corresponding hormone image feature matrix.
[0030] Understandably, multi-dimensional feature extraction can fully capture the image changes of the reaction area during the hormone response process from multiple levels such as color, grayscale distribution, texture and contour. Compared with single-dimensional feature extraction, it can more fully explore the differences in the response process of different concentrations and types of hormones. After the multi-dimensional features are serialized and spliced to construct a feature matrix, it is also easier to quickly complete feature matching and comparison in the subsequent AI pattern recognition process, thereby improving the overall recognition efficiency and the accuracy of the detection results.
[0031] In some specific embodiments of this application, the parallel detection system for multiple indicators of endocrine hormones further includes: The database storage module is used to store the sample database; The sample database is used to store reaction images, feature parameters, and historical test results of different endocrine hormones at different times.
[0032] Understandably, by establishing a dedicated sample database to store historical detection data, sufficient matching samples can be provided for the AI pattern recognition process. As the number of detections increases, the sample data stored in the database will continue to be enriched, thereby gradually improving the accuracy of subsequent feature matching.
[0033] In some specific embodiments of this application, when the recognition module is used to match the feature parameters based on a sample database, it includes: The feature vector marked by each time stamp in the same hormone image feature matrix is matched with the sample database, and the similarity is calculated; wherein, the reaction time corresponding to each time stamp is consistent with the reaction time in the historical detection results matched in the sample database; Historical detection results with similarity greater than a preset threshold are selected as candidate detection results; Statistical analysis of candidate detection results under different timestamps in the same hormone image feature matrix; Hormone indicator test results are determined based on the number of times the same historical test result appears in the candidate test results.
[0034] Understandably, by performing feature matching at different timestamps and counting the occurrences of candidate results, we can make full use of the feature matching results at different reaction times during the hormonal response to make a comprehensive judgment. This avoids the interference of feature fluctuations or random matching errors at a single time point on the final detection results, and further improves the stability and accuracy of the final detection results. Compared with relying solely on the matching results of a single time point, the comprehensive matching statistics of multiple time points can better reflect the true hormonal response.
[0035] In some specific embodiments of this application, determining the hormone index detection result based on the number of occurrences of the same historical detection result in the candidate detection results includes: The candidate test results are sorted from highest to lowest frequency, and the candidate test result with the highest frequency is selected as the final test result for the corresponding hormone indicator. If there are multiple candidate detection results with the same number of occurrences and all being the highest, and the number of occurrences is not 1, then calculate the average similarity between the feature vectors corresponding to the multiple candidate detection results and the feature vector to be detected, and select the candidate detection result with the highest average similarity as the final detection result.
[0036] Understandably, the judgment logic of sorting by frequency of occurrence and then filtering by average similarity can make full use of the statistical characteristics of matching results at multiple time points to eliminate random interference, and can also effectively solve the problem of judging multiple candidate results with the same number of votes, ensuring that the final detection results can be unique and accurate, further improving the robustness of the automatic recognition judgment process and avoiding errors in recognition results due to defects in the judgment logic.
[0037] In some specific embodiments of this application, when determining the hormone index detection result based on the number of occurrences of the same historical detection result in the candidate detection results, it further includes: If the number of candidate test results is zero, then the hormone indicator test result is determined to be undetectable; If the number of candidate test results is greater than zero, and each historical test result in the candidate results appears once, then the hormone index test result is determined to be undetectable. When the hormone indicator test result is not detected, manual testing is triggered, and the manual test result is entered. The hormone image feature matrix of the biological sample to be tested and the manual test result are marked as new sample data and stored in the sample database.
[0038] Understandably, by triggering manual retesting and adding undetected samples to the database, we can address special hormone types or concentrations not yet included in the sample database, avoiding system errors when there are no matching results. We can also dynamically expand and update the sample database, continuously optimize its coverage, and thus continuously improve the accuracy of subsequent detection and identification by the system, achieving adaptive iterative optimization of the system.
[0039] In one specific embodiment of this application, the simultaneous detection of four endocrine hormones—thyroid-stimulating hormone (TSH), luteinizing hormone (LH), estradiol (E2), and progesterone (P4)—is taken as an example.
[0040] First, the serum sample to be tested is added to the sample carrier module. The sample carrier module uses its internal microfluidic structure to partition the serum sample into four independent detection zones, each corresponding to a different hormone detection test. The zones are physically isolated from each other using a hydrophobic isolation structure to reduce cross-interference caused by liquid diffusion. Simultaneously, an identification coding area is set in each detection zone. This identification coding area includes a QR code or positioning markers to establish a spatial correspondence between the detection area and subsequently acquired images.
[0041] Subsequently, the hormone reaction module adds the corresponding hormone detection reagents to each detection zone, causing a colorimetric reaction between the serum sample and the reagents to form the corresponding hormone reaction area. The image acquisition module uses a CMOS image sensor in conjunction with a constant color temperature ring light source to synchronously image and acquire the hormone reaction areas of each detection zone, obtaining several hormone detection images, and adding a timestamp to the image acquired at each moment; wherein, the image acquisition time nodes are multiple sampling moments preset in the reaction process, preferably at the 5th minute, 10th minute, and 15th minute after the start of the reaction, but not limited to these.
[0042] In the image processing stage, the image acquisition module first performs noise filtering, brightness compensation, and color normalization on the acquired hormone detection images to eliminate the effects of changes in ambient light and equipment noise. Then, it automatically locates the detection area based on the identification coding area of each partition and performs spatial alignment processing on images acquired at different time points to ensure a consistent spatial reference relationship between different times and different partitions.
[0043] Furthermore, the feature extraction module performs multi-dimensional feature extraction on each processed hormone detection image, including: extracting the mean and variance of the color components of the reaction area as color features, statistically analyzing the gray-level distribution probability as gray-level distribution features, extracting texture features through the gray-level co-occurrence matrix, and obtaining the perimeter, area, and roundness of the reaction area as contour features based on the edge detection algorithm.
[0044] Subsequently, the multidimensional feature parameters extracted at the same time point are sequentially spliced in a preset order to form the feature vector of the corresponding single detection index; and the feature vectors of the same hormone index at different time points are spliced in chronological order to form the hormone image feature matrix corresponding to the hormone, which is used to characterize its overall feature evolution process as the reaction time changes.
[0045] Furthermore, the recognition module calls the sample database in the database storage module, performs similarity matching between the feature vector at each time point and the historical detection samples with the corresponding reaction time in the sample database, and selects historical detection results with similarity greater than a preset threshold as candidate detection results.
[0046] Furthermore, statistical analysis is performed on the candidate detection results of the same hormone at different time points to determine the frequency of occurrence of each candidate detection result, and the candidate results are ranked based on the frequency of occurrence. The candidate detection result with the highest frequency of occurrence is selected as the preliminary detection result of the hormone indicator. When multiple candidate detection results have the same highest frequency of occurrence, the average similarity between the feature vector corresponding to each candidate result and the feature vector to be detected is further calculated, and the candidate result with the highest average similarity is selected as the final detection result.
[0047] Finally, when the number of candidate test results is zero, or the frequency of each candidate result is 1, the hormone indicator test result is determined to be undetectable. At this time, the system triggers a manual review process, where the testing personnel conduct a retest using standard testing methods. The manually confirmed test results and the corresponding hormone image feature matrix are then stored as new sample data in the sample database to achieve continuous expansion and iterative optimization of the sample database.
[0048] On the other hand, see Figure 2 As shown, this application also provides a parallel detection method for multiple indicators of endocrine hormones based on AI pattern recognition, applied to the aforementioned parallel detection system for multiple indicators of endocrine hormones based on AI pattern recognition, including the following steps: S100: Carry the biological sample to be tested and partition the biological sample to be tested into multiple channels; S200: React the biological sample to be tested with the corresponding hormone detection reagent to form the corresponding hormone reaction image; S300: Simultaneously acquire images of various hormone responses to obtain images of hormone detection for several indicators; S400. Extract features from each hormone detection image to obtain feature parameters and construct the corresponding hormone image feature matrix; wherein, the feature parameters include color features, grayscale distribution features, texture features and reaction area contour features. S500: Match the feature parameters based on the sample database to obtain hormone index detection results.
[0049] Understandably, this invention achieves parallel and synchronous detection of multiple endocrine hormones through multi-channel partitioning, avoiding the cumbersome process of traditional single-index multi-stage detection and effectively improving detection efficiency. At the same time, by constructing a feature matrix through multi-dimensional image feature extraction and combining it with AI mode matching to complete result judgment, it can fully explore the effective information in the reaction image, reduce errors caused by manual operation and interpretation, improve the accuracy and stability of multi-index detection, and meet the needs of rapid batch testing in clinical settings.
[0050] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program goods. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program goods embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0051] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program goods according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0052] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0053] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0054] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.
Claims
1. A parallel detection system for multiple indicators of endocrine hormones based on AI pattern recognition, characterized in that, include: The sample carrying module is used to carry the biological sample to be tested and to partition the biological sample to be tested into multiple channels; The hormone reaction module is used to react the biological sample to be tested with the corresponding hormone detection reagent to form the corresponding hormone reaction image. The image acquisition module is used to simultaneously acquire images of various hormone responses to obtain images of hormone detection for several indicators. The feature extraction module is used to extract features from each hormone detection image to obtain feature parameters and construct the corresponding hormone image feature matrix; wherein, the feature parameters include color features, grayscale distribution features, texture features and reaction area contour features; The identification module is used to match the feature parameters based on the sample database to obtain hormone index detection results.
2. The parallel detection system for multiple indicators of endocrine hormones based on AI pattern recognition according to claim 1, characterized in that, The sample carrying module is used to carry the biological sample to be tested, and when performing multi-channel partitioning of the biological sample to be tested, it includes: The biological sample to be tested is divided into multiple channels so that different channels correspond to different endocrine hormone detection items. The system includes an isolation structure between each partition and an identification coding area within each partition to establish a matching relationship between the corresponding hormone detection area and the detection image.
3. The parallel detection system for multiple indicators of endocrine hormones based on AI pattern recognition according to claim 2, characterized in that, The image acquisition module is used to simultaneously acquire images of various hormone responses. When obtaining images of several hormone indicators, it includes: Acquire hormone detection images of several indicators within the reaction time, and simultaneously acquire hormone detection images of each indicator at the same time. Each collected hormone detection image is timestamped.
4. The parallel detection system for multiple indicators of endocrine hormones based on AI pattern recognition according to claim 3, characterized in that, The image acquisition module is used to simultaneously acquire images of various hormone responses, and after obtaining images of several hormone indicators, it includes: The hormone detection images were subjected to noise filtering, brightness compensation, and color normalization. Based on the identified coding area, the hormone detection images of each indicator are processed for region localization and image alignment.
5. The parallel detection system for multiple indicators of endocrine hormones based on AI pattern recognition according to claim 4, characterized in that, The feature extraction module is used to extract features from each hormone detection image to obtain feature parameters and construct the corresponding hormone image feature matrix. This includes: The mean and variance of the color components of the reaction region in the hormone detection image of each processed indicator were extracted as color features. The grayscale distribution characteristics are obtained by statistically analyzing the pixel distribution probability of each grayscale level within the reaction area. Texture features are extracted using the gray-level co-occurrence matrix; The reaction region is segmented based on the edge detection algorithm, and its perimeter, area, and roundness are calculated to obtain the contour features; The feature parameters are serialized and concatenated in a preset order to generate a feature vector corresponding to a single detection index. The feature vectors under all timestamps of the same detection indicator are serialized and concatenated to obtain the corresponding hormone image feature matrix.
6. The parallel detection system for multiple indicators of endocrine hormones based on AI pattern recognition according to claim 5, wherein the parallel detection system for multiple indicators of endocrine hormones further comprises: The database storage module is used to store the sample database; The sample database is used to store reaction images, feature parameters, and historical test results of different endocrine hormones at different times.
7. The parallel detection system for multiple indicators of endocrine hormones based on AI pattern recognition according to claim 6, characterized in that, When the recognition module is used to match the feature parameters based on the sample database, it includes: The feature vector marked by each time stamp in the same hormone image feature matrix is matched with the sample database, and the similarity is calculated; wherein, the reaction time corresponding to each time stamp is consistent with the reaction time in the historical detection results matched in the sample database; Historical detection results with similarity greater than a preset threshold are selected as candidate detection results; Statistical analysis of candidate detection results under different timestamps in the same hormone image feature matrix; Hormone indicator test results are determined based on the number of times the same historical test result appears in the candidate test results.
8. The parallel detection system for multiple indicators of endocrine hormones based on AI pattern recognition according to claim 7, characterized in that, When determining the hormone index test result based on the number of occurrences of the same historical test result in the candidate test results, the following is included: The candidate test results are sorted from highest to lowest frequency, and the candidate test result with the highest frequency is selected as the final test result for the corresponding hormone indicator. If there are multiple candidate detection results with the same number of occurrences and all being the highest, and the number of occurrences is not 1, then calculate the average similarity between the feature vectors corresponding to the multiple candidate detection results and the feature vector to be detected, and select the candidate detection result with the highest average similarity as the final detection result.
9. The parallel detection system for multiple indicators of endocrine hormones based on AI pattern recognition according to claim 7, characterized in that, When determining the hormone index test result based on the number of occurrences of the same historical test result in the candidate test results, it also includes: If the number of candidate test results is zero, then the hormone indicator test result is determined to be undetectable; If the number of candidate test results is greater than zero, and each historical test result in the candidate results appears once, then the hormone index test result is determined to be undetectable. When the hormone indicator test result is not detected, manual testing is triggered, and the manual test result is entered. The hormone image feature matrix of the biological sample to be tested and the manual test result are marked as new sample data and stored in the sample database.
10. A method for parallel detection of multiple indicators of endocrine hormones based on AI pattern recognition, applied to the parallel detection system for multiple indicators of endocrine hormones based on AI pattern recognition as described in any one of claims 1-9, characterized in that, include: It carries the biological sample to be tested and divides the biological sample to be tested into multiple channels; The biological samples to be tested are reacted with the corresponding hormone detection reagents to form corresponding hormone reaction images; Simultaneously acquire images of various hormone responses to obtain images of hormone detection for several indicators; Feature extraction is performed on the hormone detection images for each indicator to obtain feature parameters, and a corresponding hormone image feature matrix is constructed; wherein, the feature parameters include color features, grayscale distribution features, texture features, and reaction area contour features; The feature parameters are matched based on the sample database to obtain hormone index detection results.