Hospital department image auxiliary analysis and diagnosis system based on big data learning

Abstract

The application discloses a hospital department image auxiliary analysis and diagnosis system based on big data learning and belongs to the technical field of medical image diagnosis. The system comprises a big data storage and management module, which is used for storing image data and clinical information of patients in a preset region, wherein the clinical information comprises the age, gender and medical history of the patients; a database is constructed according to the image data and the clinical information; and a data preprocessing module, which is responsive to the database, is used for preprocessing the image data in the database, including denoising, normalization and segmentation of the image. Through big data learning and deep learning algorithms, the application realizes automatic analysis of medical image data and generation of diagnosis suggestions, improves the efficiency and accuracy of diagnosis, optimizes data management and utilization, enhances clinical decision support and improves the utilization efficiency of medical resources.

Description

Technical Field

[0001] This invention relates to the field of medical image diagnostic technology, and in particular to an image-assisted analysis and diagnostic system for hospital departments based on big data learning. Background Technology

[0002] In recent years, medical image diagnosis and analysis based on big data has received increasing attention in the medical field. This provides doctors and patients with medical decision-making and personalized treatment plans. In modern medicine, medical images (such as X-rays, CT scans, MRI scans, ultrasound scans, etc.) are important bases for disease diagnosis.

[0003] For research on medical image analysis and diagnosis, application document CN202410486788.5 provides an adjustable medical auxiliary diagnostic system and method based on big data. The technical solution includes a database for expanding and supporting the diagnostic reasoning module, a knowledge base management module, and a cyclic timing module. The cyclic timing module is used to acquire the patient's physical condition information within a limited time period and uses this time period as the cycle for judging the effectiveness of the medical plan. The diagnostic adjustment module adopts the technical solution of this invention, in which the doctor performs cyclic tracking and verification of the diagnosis and treatment process after confirming the diagnosis and treatment plan, thus perfecting the auxiliary diagnosis and treatment system to the entire process of patient treatment.

[0004] Another technical application, CN202310023862.5, provides a diagnostic report generation system based on big data of medical images. This solution categorizes existing case data and related medical images into different symptom sets, creating index keywords to improve retrieval efficiency. It acquires patient medical examination images using ultrasound equipment and completes and delineates the target area based on prior knowledge of the consistency between the density of the target area and the density characteristics of lesions in surrounding adjacent areas. The solution inputs keywords indicating confirmed symptoms, matches the acquired medical examination images with existing case images for similarities, and finds the existing case images with the most similarities. Based on the existing case diagnosis results, it infers the diagnostic result of the acquired medical examination images.

[0005] However, existing diagnostic systems based on big data learning struggle to identify subtle abnormalities in lesions within images and to label the features of lesions. This results in deficiencies in segmenting and comparing lesions in images, leading to insufficient ability to identify inconspicuous abnormalities, misdiagnosis, and missed diagnosis, thus hindering the improvement of diagnostic accuracy. Summary of the Invention

[0006] In view of the problems existing in the field of medical imaging diagnostic technology, the present invention is proposed.

[0007] Therefore, one of the objectives of this invention is to provide a hospital department image-assisted analysis and diagnostic system based on big data learning. Through big data learning and deep learning algorithms, it realizes automated analysis of medical image data and generation of diagnostic suggestions, improves the efficiency and accuracy of diagnosis, optimizes data management and utilization, enhances clinical decision support, and improves the efficiency of medical resource utilization.

[0008] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0009] This invention provides an image-assisted analysis and diagnostic system for hospital departments based on big data learning, comprising:

[0010] The big data storage and management module is used to store the imaging data and clinical information of patients in a preset region. The clinical information includes the patient's age, gender, and medical history. The module also constructs a database based on the imaging data and clinical information.

[0011] The data preprocessing module responds to the database and is used to preprocess the image data in the database, including denoising, normalization and segmentation of the images;

[0012] The image data acquisition module is used to acquire image data of preset patients and upload the image data and the clinical information of preset patients to the database for association with the image data;

[0013] The big data fusion learning module is used to perform feature recognition and / or learning on image data in the database using deep learning algorithms; the big data fusion learning module includes a feature extraction unit, an acquisition unit, a comparison unit, and an update unit;

[0014] The feature extraction unit is used to extract features from the image data in the database, including tissue structure, shape, size and texture of lesion areas;

[0015] The acquisition unit is used to retrieve the diagnostic results corresponding to the image data from the database;

[0016] The comparison unit is used to analyze and compare the features of image data acquired in future time periods with the features of image data in the database, and generate diagnostic suggestions for patients who match the acquired image data based on the analysis and comparison results; the diagnostic suggestions include the patient's disease type and lesion degree;

[0017] The update unit is used to periodically update the database, including by introducing an online learning mechanism that updates the database when it receives new image data and diagnostic results.

[0018] In a preferred embodiment of the present invention, the image data in the database is distinguished according to the diagnostic results, and the distinction method includes distinguishing by diagnostic category, by case, and / or by time sequence; wherein, the diagnostic category distinction is to distinguish the image data into different categories according to the diagnostic results; including distinguishing the image data into normal category, disease A category, and / or disease B category;

[0019] Case differentiation is performed based on the diagnosis results, distinguishing patients by case and constructing a dataset for each patient; temporal differentiation is performed based on the time sequence of image data acquisition, including using early-stage image data as the training set, mid-stage image data as the validation set, and late-stage image data as the test set.

[0020] In a preferred embodiment of the present invention, the number of different cases is counted in the database according to the case distinction, and the features of the image data of each case are obtained from the counted different cases, including extracting the shape features of lesions or organs from one or more cases in the database through contour detection and / or geometric analysis, and marking the extracted cases as reference cases; the shape features include area, perimeter, and roundness; and the variation pattern of shape features is analyzed in different cases based on the image data obtained at different times.

[0021] In a preferred embodiment of the present invention, the image data acquired at different times are divided into θ1, θ2, ..., θ n , where θ n This refers to the image data acquired at the nth time point. At least 3 to 5 feature points related to the lesion are selected from the image images corresponding to each segmented image data. The contour features of each feature point are obtained. If the contour of the feature point becomes clear, the system determines that the corresponding feature point is developing in a stable and / or improving trend; otherwise, no determination is made.

[0022] In a preferred embodiment of the present invention, the selected feature points are divided into early feature points, mid-term feature points, and late-term feature points. When new patient image data is collected in the future, the lesions in the image images corresponding to the image data are analyzed and compared with the early feature points. If the contour features of the lesions are the same as those of the early feature points, the system determines that the patient is the same as the corresponding case in the reference cases; otherwise, no determination is made.

[0023] In a preferred embodiment of the present invention, the lesion type and lesion features are labeled according to the initial feature points, intermediate feature points, and late feature points. The lesion features are labeled as blurred, irregular, missing, rough, swollen, displaced, lobulated, or umbilicated, with ring-like enhancement or halo sign, and serrated or ground-glass opacities. Based on the labeling of the lesion features, the lesion type is labeled accordingly. When new patient image data is acquired in a future period, the contour features of the lesion in the corresponding image image are analyzed, and the contour features are matched with the labeled lesion features. If the match is successful, the system determines that the patient has the lesion type corresponding to the labeled lesion features; if the match fails, no determination is made.

[0024] As a preferred embodiment of the present invention, the labeling of the corresponding feature points with lesion type and lesion features also includes labeling based on functional and physiological features, wherein the labeling of functional features includes labeling the image data with features related to organ function.

[0025] Annotation of physiological characteristics, including annotation of characteristics related to physiological state;

[0026] It also includes time-series-based annotation, which includes dynamic change annotation and time point annotation. Dynamic change annotation includes annotating the changes of lesions over time in image data.

[0027] Time point annotation, including annotation of the time points when image data was acquired;

[0028] It also includes annotation based on multimodal images, which includes multimodal fusion annotation. In multimodal fusion annotation, for image data containing multiple modalities, lesion features under different modalities are annotated and fusion analysis is performed; and according to each annotation, the lesion type is annotated accordingly.

[0029] In a preferred embodiment of the present invention, the proportion of different labels is calculated among all the labels based on the labeling of lesion features. Based on the calculation results, the proportions of different labels are divided into high proportion, medium proportion and low proportion in a stepwise manner. When new patient image data is collected in the future, the image data is matched with lesion features according to the corresponding division.

[0030] In a preferred embodiment of the present invention, if the system determines that the patient is the same case as the corresponding case in the reference cases, then the system provides an examination period for the patient's image data. The examination period is provided based on the acquisition time of the image data corresponding to the intermediate feature point. The system calculates the time difference between the acquisition time of the image data corresponding to the intermediate feature point and the acquisition time of the image data corresponding to the initial feature point. During the time difference, the system acquires image data for the patient at least 1 to 2 times. If the lesion in the image image corresponding to the acquired image data changes towards the contour features corresponding to the intermediate feature point, the system maintains the original determination; otherwise, it does not maintain it.

[0031] Beneficial effects:

[0032] 1. Through big data learning, the system can extract features from image data and generate diagnostic suggestions, thereby improving diagnostic efficiency;

[0033] 2. This invention not only analyzes common features of images such as contour, size, and shape, but also combines multi-dimensional information such as functional and physiological characteristics, time series changes, and multimodal images to more comprehensively assess lesions. This comprehensive analysis method helps to improve the accuracy of diagnosis and reduce the possibility of misdiagnosis and missed diagnosis.

[0034] 3. This invention distinguishes image data based on patient medical records and constructs a dataset for the corresponding patient. This patient-centered data management approach enables the system to perform personalized analysis for each patient's specific situation, providing support for personalized medicine.

[0035] 4. This invention annotates lesion features, including contour features, functional features, physiological features, etc. When new patient image data is input, the system will match the lesion features with the annotated features to quickly provide diagnostic suggestions. This big data-based learning and matching mechanism provides strong support for clinical decision-making.

[0036] 5. This invention can perform annotation and fusion analysis of multimodal images, which allows doctors to refer to information from multiple image modalities simultaneously, thereby making a more comprehensive assessment of the condition. This multimodal analysis method not only improves the accuracy of diagnosis, but also reduces the reliance on single-modal images and improves the efficiency of medical resource utilization. Attached Figure Description

[0037] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0038] in:

[0039] Figure 1 This is a schematic diagram of the modular structure of the hospital department image-assisted analysis and diagnosis system based on big data learning, according to an embodiment of the present invention.

[0040] Figure 2 This is a schematic diagram of the process structure of an embodiment of the present invention;

[0041] The diagram is labeled as follows: 110 - Big Data Storage and Management Module; 120 - Data Preprocessing Module; 130 - Image Data Acquisition Module; 140 - Big Data Fusion Learning Module; 1401 - Feature Extraction Unit; 1402 - Acquisition Unit; 1403 - Comparison Unit; 1404 - Update Unit. Detailed Implementation

[0042] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the described embodiments of the present invention are within the scope of protection of the present invention.

[0043] Existing diagnostic systems based on big data learning struggle to identify subtle abnormalities in lesions within images and fail to properly label these features. This results in limitations in lesion segmentation and comparison, leading to insufficient ability to identify subtle anomalies and causing misdiagnosis and missed diagnosis, thus hindering diagnostic accuracy. To address this, this invention proposes a hospital department image-assisted analysis and diagnostic system based on big data learning. Through big data learning and deep learning algorithms, it automates the analysis of medical image data and generates diagnostic suggestions, improving diagnostic efficiency and accuracy, optimizing data management and utilization, enhancing clinical decision support, and improving the efficiency of medical resource utilization.

[0044] The present solution will be further described in detail below through embodiments and in conjunction with the accompanying drawings.

[0045] Reference Figures 1 to 2 As one embodiment of the present invention, this embodiment provides an image-assisted analysis and diagnosis system for hospital departments based on big data learning, comprising:

[0046] The big data storage and management module 110 is used to store the image data and clinical information of patients in a preset region. The clinical information includes the patient's age, gender, and medical history. It also constructs a database based on the image data and clinical information.

[0047] In this embodiment, a complete database is constructed, providing a rich data foundation for subsequent analysis and diagnosis;

[0048] The data preprocessing module 120 responds to the database and is used to preprocess the image data in the database, including denoising, normalization and segmentation of the images;

[0049] In this embodiment, preprocessing the image data in the database can improve the quality and analyzability of the images, thus enhancing the accuracy and reliability of subsequent analysis.

[0050] The image data acquisition module 130 is used to acquire image data of a preset patient and upload the image data and the clinical information of the preset patient to the database for association with the image data;

[0051] In this embodiment, image data of a preset patient is collected. The collection method includes connecting to the hospital's imaging equipment (such as X-ray machine, CT scanner, MRI equipment, etc.) to collect the patient's image data.

[0052] The big data fusion learning module 140 is used to perform feature recognition and / or learning on image data in the database using deep learning algorithms; the big data fusion learning module 140 includes a feature extraction unit 1401, an acquisition unit 1402, a comparison unit 1403 and an update unit 1404;

[0053] In this embodiment, the deep learning algorithm includes a deep learning algorithm based on a convolutional neural network;

[0054] The feature extraction unit 1401 is used to extract features from the image data in the database, including tissue structure, shape, size and texture of lesion areas;

[0055] The acquisition unit 1402 is used to retrieve the diagnostic results corresponding to the image data from the database;

[0056] The comparison unit 1403 is used to analyze and compare the features of image data acquired in a future time period with the features of image data in the database, and generate diagnostic suggestions for patients who match the acquired image data based on the analysis and comparison results; the diagnostic suggestions include the patient's disease type and lesion degree;

[0057] The update unit 1404 is used to update the database periodically, including introducing an online learning mechanism, which updates the database when it receives new image data and diagnostic results;

[0058] In this embodiment, the online learning mechanism refers to updating its own parameters or strategies immediately each time new data is received, without needing to regenerate the database, in order to adapt to the continuous generation of data;

[0059] The big data fusion learning module uses deep learning algorithms to automatically extract features from image data, enabling automated feature recognition and learning, and generating diagnostic suggestions, thereby improving the efficiency and accuracy of diagnosis.

[0060] The update unit introduces an online learning mechanism, which can update the information in the database in real time, enabling the system to continuously learn and optimize, adapt to new image data and diagnostic results, and maintain the advanced nature of diagnosis.

[0061] Specifically, in this embodiment, the image data in the database is differentiated according to the diagnostic results. This differentiation can be done by diagnostic category, by case, and / or by time sequence.

[0062] Diagnostic category differentiation involves classifying imaging data into different categories based on the diagnostic results; this includes classifying imaging data into normal category, disease category A, and / or disease category B.

[0063] In this embodiment, it is convenient to analyze and study image data of different disease types separately;

[0064] Case differentiation is performed by classifying patients based on their diagnostic results and constructing a dataset for each patient. In this embodiment, this ensures that all image data of the same patient are assigned to the same dataset, preventing the same patient's image data from appearing in other datasets at the same time, and keeping all image data of the same case together for easy longitudinal analysis.

[0065] The time sequence is distinguished according to the time sequence of image data acquisition, including using early-stage image data of patients as the training set, mid-stage image data as the validation set, and late-stage image data as the test set.

[0066] In this embodiment, the time sequence is closer to that in actual clinical applications, which allows for better assessment of patients based on new imaging data and facilitates analysis of changes in the patient's disease over time.

[0067] The system can differentiate image data according to diagnostic results, such as diagnostic category, case, or time sequence, which facilitates analysis and research on different types of data and meets different application scenarios and needs.

[0068] Furthermore, by differentiating by diagnostic categories, imaging data of different diseases can be classified and analyzed, which helps to gain a deeper understanding of the characteristics and patterns of each disease; case differentiation enables patient-centered analysis, providing support for personalized medicine; and time sequence differentiation facilitates the study of dynamic changes in lesions and the evaluation of treatment effects.

[0069] In this embodiment, based on case differentiation, the number of different cases is counted in the database, and the features of the image data of each case are obtained from the counted different cases. This includes extracting the shape features of lesions or organs from one or more cases in the database through contour detection and / or geometric analysis, and marking the extracted cases as reference cases. The shape features include area, perimeter, and roundness. In different cases, the variation pattern of the shape features is analyzed based on the image data obtained at different times.

[0070] In this embodiment, shape features are of great significance for determining the benign or malignant nature of tumors;

[0071] It also includes methods such as Gray-Level Co-occurrence Matrix (GLCM) and Local Binary Pattern (LBP) to extract texture features from image data; these features can reflect subtle structural changes in tissues and are suitable for lesion detection and tissue classification.

[0072] It also includes calculating statistical features of image data such as mean intensity, standard deviation, and skewness to reflect tissue density and homogeneity. It also includes using pre-trained CNN models (such as ResNet, VGG, etc.) to extract high-level semantic features of images. These features can capture complex image patterns and are suitable for disease classification and diagnosis.

[0073] It also includes feature extraction through self-supervised learning methods (such as contrastive learning), which do not require a large amount of labeled data and can better adapt to feature changes in different cases;

[0074] It also includes using principal component analysis (PCA) to obtain features. PCA reduces the dimensionality of the high-dimensional feature space to a low-dimensional space while retaining the main variation information. PCA can help remove feature redundancy and improve analysis efficiency.

[0075] Furthermore, the system can count the number of different cases and extract the shape features of each case's image data, such as area, perimeter, and roundness, providing detailed feature information for subsequent analysis.

[0076] By analyzing the shape feature changes of image data acquired at different times, we can better understand the development trend of lesions and provide important references for clinical diagnosis and treatment.

[0077] Based on the above, the image data acquired at different times are divided into θ1, θ2, ..., θ n , where θ nThis refers to the image data acquired at the nth time point. At least 3 to 5 feature points related to the lesion are selected from the image images corresponding to each segmented image data. The contour features of each feature point are obtained. If the contour of the feature point becomes clear, the system determines that the corresponding feature point is developing in a stable and / or improving trend; otherwise, no determination is made.

[0078] In this embodiment, a clearer outline usually indicates that the lesion is stabilizing or improving. For example, after treatment of an inflammatory lesion, the boundary between the lesion and the surrounding normal tissue becomes clearer.

[0079] Conversely, if the outline is missing or becomes rough, it usually indicates that the lesion is aggressively destroying the surrounding tissues; for example, when a malignant tumor infiltrates the surrounding tissues, it can cause the interface between the normal tissue and the lesion to disappear.

[0080] The system selects feature points from image data at different times and analyzes the changes in their contour features. This can determine whether the lesion is stable or improving, providing doctors with intuitive information on the trend of lesion development and helping to adjust the treatment plan in a timely manner.

[0081] Meanwhile, quantitative analysis of the contour features of characteristic points makes the assessment of lesions more objective and accurate, reducing the subjectivity of human judgment.

[0082] Furthermore, the selected feature points are divided into early-stage feature points, mid-stage feature points, and late-stage feature points. When new patient image data is collected in the future, the lesions in the image images corresponding to the image data are analyzed and compared with the early-stage feature points. If the contour features of the lesions are the same as those of the early-stage feature points, the system determines that the patient is the same as the corresponding case in the reference cases; otherwise, no determination is made.

[0083] In this embodiment, doctors are provided with diagnostic references based on historical cases, which improves the accuracy and reliability of diagnosis. At the same time, by comparing and matching existing case data, the advantages of big data are fully utilized, enabling the system to learn from and draw on a large number of historical cases, providing strong support for the diagnosis of new cases.

[0084] This embodiment emphasizes that, based on initial, intermediate, and late-stage feature points, corresponding feature points are labeled with lesion types and lesion characteristics. Specifically, based on contour features, lesion characteristics are labeled as blurred contours, irregular contours, missing or rough contours, swollen or displaced contours, lobulated or umbilicated contours, annular enhancement or halo sign, and serrated or ground-glass opacities. Based on these lesion characteristic labels, the lesion type is correspondingly labeled. When new patient image data is acquired in future time periods, the contour features of the lesions in the corresponding image images are analyzed, and the contour features are matched with the labeled lesion features. If the match is successful, the system determines that the patient has the lesion type corresponding to the labeled lesion features; if the match fails, no determination is made.

[0085] In this embodiment, the lesion type is labeled according to the labeling of lesion features. This includes labeling the lesion type as tumor, inflammation, fibrosis, etc., based on the labeling of lesion features. For example, blurred outline usually indicates an inflammatory reaction or edema around the lesion. In lung imaging, inflammatory exudation can cause the boundary between the lesion and the surrounding normal tissue to become blurred. This is common in infectious diseases (such as pneumonia), early neoplastic lesions, or inflammatory lesions. This change may indicate that the lesion is in the active or progressive stage and requires further observation and treatment.

[0086] Irregular outlines may indicate aggressive growth or fibrosis of the lesion. For example, in chronic inflammatory lung diseases, fibrosis around the bronchi and blood vessels can result in a serrated outline. This is common in malignant tumors (such as lung cancer), chronic inflammatory diseases (such as pulmonary fibrosis), or certain infectious diseases (such as tuberculosis). Such changes may indicate a higher degree of malignancy or a more complex condition, requiring further pathological examination and treatment.

[0087] Bulking or displacement of the lesion outline usually indicates compressive growth of the lesion on surrounding tissues. For example, benign tumors or neoplastic lesions can compress surrounding tissues, causing bulging or displacement of the lesion outline. This is common in benign tumors (such as pulmonary hamartomas), certain inflammatory lesions, or cystic lesions. This change may indicate that the lesion is growing slowly, but it is still necessary to follow up regularly to observe its changes.

[0088] At the same time, the labeled data is classified and stored according to the lesion type to facilitate subsequent query and analysis;

[0089] The system provides detailed annotations of lesion features, including various conditions such as blurred, irregular, missing, or rough outlines, and assigns corresponding annotations to lesion types based on the annotations, enriching the information in the image data and providing more detailed evidence for subsequent diagnosis and research.

[0090] When new patient image data is input, the system can match lesion features with labeled features and quickly provide diagnostic suggestions, improving the efficiency and accuracy of diagnosis and reducing the workload of doctors.

[0091] Furthermore, the corresponding feature points are labeled with lesion types and lesion characteristics, including labeling based on functional and physiological characteristics. Among them, the labeling of functional characteristics includes labeling the image data with features related to organ function.

[0092] In this embodiment, for example, cardiac images can be labeled with myocardial contractile function, heart valve movement, etc.; the labeling of physiological features includes labeling features related to physiological state.

[0093] In this embodiment, for example, the metabolic activity of the lesion area can be marked in PET images;

[0094] It also includes time-series-based annotation, which includes dynamic change annotation and time point annotation. Dynamic change annotation includes annotating the changes of lesions over time in image data.

[0095] In this embodiment, the changes of the lesion over time can be marked, such as the enlargement, shrinkage, and morphological changes of the lesion. For example, in the follow-up of cancer patients, the changes in the volume of the tumor can be marked.

[0096] Time point annotation, including annotation of the time points when image data was acquired;

[0097] In this embodiment, it is convenient to analyze the rate of lesion progression;

[0098] It also includes annotation based on multimodal images, which includes multimodal fusion annotation. In multimodal fusion annotation, for image data containing multiple modalities, lesion features under different modalities are annotated and fusion analysis is performed; and according to each annotation, the lesion type is annotated accordingly.

[0099] In this embodiment, multiple modalities of image data are used, including CT, MRI, PET, etc.; for example, in PET / CT images, anatomical structures in CT images and metabolic information in PET images are labeled.

[0100] In addition to contour features, the system also adds annotations based on functional and physiological features, as well as annotations based on time series and multimodal images, making the annotation information more comprehensive and richer, and able to reflect the characteristics and changes of lesions from multiple perspectives.

[0101] Multi-dimensional annotation information provides more data for comprehensive analysis of lesions, which helps doctors to understand the condition more comprehensively and make more accurate diagnostic and treatment decisions.

[0102] Based on the above, according to the annotation of lesion features, the proportion of different annotations in all annotations is calculated. Based on the calculation results, the proportions of different annotations are divided into high, medium, and low proportions in a stepwise manner. When new patient image data is collected in the future, the image data is matched with lesion features according to the corresponding division. In this embodiment, the system divides the image data into steps according to the proportion of different annotations and matches lesion features according to the corresponding division during matching. This can more reasonably assess the importance of different features and improve the reliability of diagnostic suggestions. At the same time, by considering the proportion of different annotations, the system can match lesion features more accurately, avoid misjudgment due to accidental similarity of certain features, and improve the accuracy of diagnosis.

[0103] Furthermore, if the system determines that the patient is the same case as the corresponding case in the reference cases, then the system will assign an examination period for the patient's image data. The given examination period includes being based on the acquisition time of the image data corresponding to the intermediate feature point, calculating the time difference between the acquisition time of the image data corresponding to the intermediate feature point and the acquisition time of the image data corresponding to the initial feature point, and acquiring image data for the patient at least 1 to 2 times within the time difference. If the lesion in the image image corresponding to the acquired image data changes towards the contour features corresponding to the intermediate feature point, then the system will maintain the original determination; otherwise, it will not maintain it.

[0104] In this embodiment, after determining that the patient and the reference case are the same case, the system will give the image data inspection cycle according to the acquisition time of the image data corresponding to the mid-term feature points, and perform dynamic monitoring within the cycle. If the lesion changes in the expected direction, the original judgment will be maintained; otherwise, it will not be maintained. This dynamic monitoring mechanism can detect changes in the condition in a timely manner, ensuring the accuracy and timeliness of the diagnosis.

[0105] Furthermore, by acquiring imaging data and analyzing lesion changes during the examination period, the system can evaluate the treatment effect, provide a basis for doctors to adjust the treatment plan, and help improve the treatment effect and patient prognosis.

[0106] In summary, this invention, through big data learning and deep learning algorithms, achieves automated analysis and diagnostic suggestion generation of medical image data, improving diagnostic efficiency and accuracy, optimizing data management and utilization, enhancing clinical decision support, and improving the efficiency of medical resource utilization.

[0107] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A hospital department image-assisted analysis and diagnosis system based on big data learning, characterized in that, include: The big data storage and management module is used to store the image data and clinical information of patients in a preset area. The clinical information includes the patient's age, gender, and medical history. A database is constructed based on the imaging data and clinical information; A data preprocessing module, which responds to the database, is used to preprocess the image data in the database, including denoising, normalizing and segmenting the images; The image data acquisition module is used to acquire image data of a preset patient and upload the image data and the clinical information of the preset patient to the database for association with the image data; The big data fusion learning module is used to perform feature recognition and / or learning on the image data in the database using deep learning algorithms; the big data fusion learning module includes a feature extraction unit, an acquisition unit, a comparison unit, and an update unit; The feature extraction unit is used to extract features from the image data in the database, including tissue structure, shape, size and texture of lesion areas; The acquisition unit is used to acquire diagnostic results corresponding to the image data from the database; The comparison unit is used to analyze and compare the features of image data acquired in future time periods with the features of image data in the database, and generate diagnostic suggestions for patients matching the acquired image data based on the analysis and comparison results; the diagnostic suggestions include the patient's disease type and lesion degree; The updating unit is used to periodically update the database, including introducing an online learning mechanism, which updates the database when it receives new image data and diagnostic results; Based on the diagnostic results, the image data in the database are differentiated, including by diagnostic category, by case, and / or by time sequence; wherein, The diagnostic category distinction involves classifying the image data into different categories based on the diagnostic results; including classifying the image data into a normal category, disease A category, and / or disease B category. The case differentiation is based on the diagnostic results, distinguishing cases by patient case and constructing a dataset for the corresponding patients; The time sequence distinction is based on the time sequence of image data acquisition, including using image data acquired in the early stage of the patient's life as the training set, image data acquired in the middle stage as the validation set, and image data acquired in the late stage as the test set. Based on the case distinctions, the number of different cases is counted in the database, and the features of the image data of each case are obtained from the counted different cases. This includes extracting shape features of lesions or organs from one or more cases in the database through contour detection and / or geometric analysis, and marking the extracted cases as reference cases. The shape features include area, perimeter, and roundness. In different cases, the variation pattern of shape features is analyzed based on the image data acquired at different times. Image data acquired at different times are divided into , ,..., ,in, Indicated at the For each time period, at least 3 to 5 feature points related to the lesion are selected from the image images corresponding to each segmented image data. The contour features of each feature point are obtained. If the contour of the feature point becomes clear, the system determines that the corresponding feature point is developing in a stable and / or improving trend; otherwise, no determination is made.

2. The hospital department image-assisted analysis and diagnosis system based on big data learning as described in claim 1, characterized in that, The selected feature points are divided into early-stage feature points, mid-stage feature points, and late-stage feature points. When new patient image data is collected in the future, the lesions in the image images corresponding to the image data are analyzed and compared with the early-stage feature points. If the contour features of the lesions are the same as those of the early-stage feature points, the system determines that the patient is the same as the corresponding case in the reference cases; otherwise, no determination is made.

3. The hospital department image-assisted analysis and diagnosis system based on big data learning as described in claim 2, characterized in that, Based on the initial, intermediate, and late-stage feature points, the corresponding feature points are labeled with lesion types and lesion characteristics. Specifically, based on the contour features, the lesion characteristics are labeled as blurred contours, irregular contours, missing or rough contours, swollen or displaced contours, lobulated or umbilicated contours, annular enhancement or halo sign, and serrated or ground-glass opacities. Based on the labeling of these lesion characteristics, the lesion type is correspondingly labeled. When new patient image data is acquired in future time periods, the contour features of the lesions in the corresponding image images are analyzed, and the contour features are matched with the labeled lesion features. If the match is successful, the system determines that the patient has the lesion type corresponding to the labeled lesion features; if the match fails, no determination is made.

4. The hospital department image-assisted analysis and diagnosis system based on big data learning as described in claim 3, characterized in that, The corresponding feature points are labeled with lesion types and lesion characteristics, and also include labeling based on functional and physiological characteristics. Among them, the labeling of functional characteristics includes labeling the image data with features related to organ function. Annotation of physiological characteristics, including annotation of characteristics related to physiological state; It also includes time-series-based annotation, which includes dynamic change annotation and time point annotation, wherein dynamic change annotation includes annotating the changes of lesions in image data over time; Time point annotation, including annotation of the time points when image data was acquired; It also includes annotation based on multimodal images, which includes multimodal fusion annotation. In multimodal fusion annotation, for image data containing multiple modalities, lesion features under different modalities are annotated and fusion analysis is performed; and according to each annotation, the lesion type is annotated accordingly.

5. The hospital department image-assisted analysis and diagnosis system based on big data learning as described in any one of claims 3 to 4, characterized in that, Based on the annotation of lesion features, the proportion of different annotations is calculated in all annotations. Based on the calculation results, the proportions of different annotations are divided into high proportion, medium proportion and low proportion. When new patient image data is collected in the future, the image data is matched with lesion features according to the corresponding division.

6. The hospital department image-assisted analysis and diagnosis system based on big data learning as described in claim 2, characterized in that, If the system determines that the patient is the same case as the corresponding case in the reference cases, then the system assigns an examination period for the patient's image data. The assignment of the examination period includes being based on the acquisition time of the image data corresponding to the intermediate feature point, calculating the time difference between the acquisition time of the image data corresponding to the intermediate feature point and the acquisition time of the image data corresponding to the initial feature point, and acquiring image data for the patient at least 1 to 2 times within the time difference. If the lesion in the image image corresponding to the acquired image data changes towards the contour features corresponding to the intermediate feature point, then the system maintains the original determination; otherwise, it does not maintain it.

Images