Radiographic image cloud film multi-time phase comparison system based on patient unique identifier

Abstract

The invention discloses a radiographic image cloud film multi-time phase comparison system based on a patient unique identifier, and relates to the technical field of medical image information, and the system comprises a patient unique identifier analysis module, a cloud film dynamic archive, an image intelligent prefetching module, a multi-time phase comparison engine, and a dynamic visual interface. Through cooperative work of the patient unique identifier analysis module and the cloud film dynamic archive library, the organization mode of image data is changed, the patient unique identifier serves as an unchangeable core index key and is automatically collected and associated to the unique cloud film dynamic archive library under the identifier, and a structured multi-temporal image data set is formed. The image history of the patient is converted into complete, traceable and machine-readable data from a scattered and temporary file set, and when a doctor needs to retrieve the image history, the doctor only needs to position the patient through the identifier, and then the whole historical image of the patient can be accessed.

Description

Technical Field

[0001] This invention relates to the field of medical imaging information technology, specifically to a multi-time comparison system for radiographic cloud films based on a patient's unique identifier. Background Technology

[0002] Traditional radiological imaging data is typically stored locally by hospital, creating data silos. When patients visit different hospitals or have follow-up examinations at the same hospital at different times, doctors find it difficult to quickly and completely access and compare the patient's historical imaging data. Existing cloud-based film technology is mostly limited to browsing and sharing the results of a single examination, lacking structured integration and intelligent comparison tools for long-term patient imaging data.

[0003] Currently, when doctors perform multi-time comparisons, they usually need to manually retrieve images from different time points and compare them in multiple independent viewing windows based on memory and experience. This process is cumbersome, inefficient, and prone to missing subtle but crucial radiological changes. In addition, images acquired under different equipment and parameters lack effective normalization processing, which further increases the difficulty and subjectivity of the comparison.

[0004] Therefore, there is an urgent need in this field for a system that can automatically integrate patient-centered imaging data throughout their entire life cycle and provide intelligent, visualized, multi-time comparison capabilities. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of existing technologies and provide a multi-temporal comparison system for radiological images based on a patient's unique identifier. This system, through the collaborative work of a patient unique identifier resolution module and a dynamic archive of cloud films, changes the way image data is organized. The patient's unique identifier serves as an immutable core index key, automatically collected and associated with a unique dynamic archive of cloud films under that identifier, forming a structured multi-temporal image dataset. This transforms the patient's image history from a scattered, temporary collection of files into a complete, traceable, and machine-readable data set. When doctors need to access the data, they only need to locate the patient by identifier to access their full historical images. This lays a solid and unified data foundation for any subsequent diagnostic and treatment activities that require historical data support.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a multi-time comparison system for radiological images based on a patient's unique identifier, the system comprising: a patient's unique identifier resolution module, a dynamic archive of cloud films, an intelligent image prefetching module, a multi-time comparison engine, and a dynamic visualization interface;

[0007] The patient unique identifier resolution module is used to receive the patient identity information input by the user, verify the identity validity by connecting with the cloud film dynamic archive, and generate a globally unique patient unique identifier (PUI).

[0008] The cloud-based dynamic film archive is used to store the patient's medical records at different examination times, using the PUI as the index key. Multi-temporal image dataset in structured storage format ,in, Indicates the first The images and related data from the second examination;

[0009] The intelligent image pre-fetching module is used to pre-fetch images based on current diagnostic scenario parameters, including the examination site. Image modality Clinical diagnostic keywords Prefetching algorithm from multi-temporal image datasets The top results with the highest relevance were selected from the list. A set of historical image data;

[0010] The multi-temporal comparison engine includes an image registration unit and a change detection unit, which are used to register and align pre-captured multi-temporal images and identify changes in lesion features.

[0011] The dynamic visualization interface is connected in communication with the multi-temporal comparison engine and is used to display multi-temporal images in a side-by-side, overlay, and dynamic switching manner, and to visualize the change detection results.

[0012] Furthermore, the structured storage format of the cloud film dynamic archive includes a three-level directory:

[0013] First-level directory: named PUI, serving as the root directory for patient records;

[0014] Second-level directory: By inspection time Name and store complete data for a single check;

[0015] Third-level directory: Stores related data including radiological images, examination reports, image annotation information, and clinical diagnostic data;

[0016] All data supports multi-dimensional retrieval by inspection time and image modality.

[0017] Furthermore, the screening steps of the prefetching algorithm in the intelligent image prefetching module are as follows:

[0018] Extract the scene parameters currently being inspected ( ),in, Code the inspection site. For image modality coding, For diagnostic coding;

[0019] Retrieve the multi-temporal image dataset corresponding to the current patient's unique identifier (PUI) from the cloud film dynamic archive;

[0020] Calculate the relevance score (Score) between each inspection in the historical dataset and the current inspection. All historical checks are sorted in descending order based on their relevance scores.

[0021] Select relevance Historical data, The relevance threshold is set with a default value of 0.6, and the scores are ranked in descending order. Group.

[0022] Furthermore, the relevance ,in, for Corresponding examination sites The similarity score is calculated as follows: 1 for identical parts, 0.7 for similar parts, and 0 for irrelevant parts. for Corresponding image modalities Modal similarity is represented by 1 for identical modes, 0.5 for compatible modes, and 0 for incompatible modes. for Diagnostic keywords in the related report This represents the semantic similarity of keywords, with a value ranging from 0 to 1. These are the weighting coefficients, and ,default value .

[0023] Furthermore, in the multi-temporal comparison engine, the image registration unit is used to match the current image specified by the user. For spatial reference, historical images The floating examination image undergoes a three-dimensional non-rigid transformation to spatially register and align its anatomical structures with the current image. The specific steps are as follows:

[0024] From the current image to be registered and historical images Extract the feature point set, which is respectively and ,in, for The Middle Two-dimensional coordinates of a feature point for The Middle Two-dimensional coordinates of a feature point;

[0025] Calculate the matching degree of feature points, eliminate false matches using the RANSAC algorithm, and retain the set of valid matching point pairs. , This represents the number of feature points;

[0026] Calculate the spatial transformation matrix based on effective matching point pairs. ,Right now Where H is a 3×3 affine transformation matrix, the objective function is obtained by solving the least squares method: ;

[0027] Historical images By transforming the matrix Mapped to the current image The coordinate system is used to obtain the registered image. .

[0028] Furthermore, in the multi-time comparison engine, the change detection unit is used to perform quantitative analysis on a series of registered and aligned images, identify and calculate the changes in lesion features of the same region at different time points based on the registration, and the specific steps are as follows:

[0029] Segment the current image using the threshold segmentation method. The lesion area in and registered historical images The lesion area in ;

[0030] Calculate the average CT value of the two regions, based on The current imaging lesion area is calculated. Average CT value and historical imaging lesion areas average , where CT(v) is the HU value of voxel v, and N(V) is the total number of voxels contained in region V;

[0031] Calculate the density change rate ΔCT, the ,in, For the lesion area CT value, For the lesion area CT value;

[0032] Calculate the size and volume change rate of the lesion The lesion volume is determined by statistically analyzing the total number of voxels contained in the lesion area. With a single voxel volume The product of these terms yields the current imaging lesion area. volume Registered historical imaging lesion areas volume The volume change rate ,in, The actual spatial volume corresponding to a single voxel in the image. This indicates an increase in the size of the lesion. This indicates that the lesion has shrunk in size;

[0033] Output quantization results, including , , , , , , .

[0034] Furthermore, the dynamic visualization interface integrates:

[0035] The comparison sequence generation unit provides users with a timeline control, allowing users to select images at multiple time points from pre-fetched historical images to generate a multi-time comparison sequence for centralized comparison.

[0036] The lesion evolution unit is communicatively connected to the change detection unit in the multi-time comparison engine, and is used to visualize and render the calculated lesion feature change data in the form of a two-dimensional chart.

[0037] Furthermore, the dynamic visualization interface also includes a permission management and auditing unit. The permission management and auditing unit is communicatively connected to the patient unique identifier resolution module and the cloud film dynamic archive. It has a preset visitor permission management function, which allows visitors to view different levels of directory data in the cloud film dynamic archive based on their permissions, and records all access behaviors, including visitor ID, access time, and operation content, and generates an audit log.

[0038] Compared with existing technologies, this multi-time comparison system for radiographic images based on patient-unique identifiers has the following advantages:

[0039] I. This invention, through the collaborative work of the patient unique identifier resolution module and the cloud film dynamic archive, changes the way image data is organized. The patient unique identifier, as an immutable core index key, is automatically collected and associated with the unique cloud film dynamic archive under that identifier, forming a structured multi-temporal image dataset. This transforms the patient's image history from a scattered, temporary collection of files into a complete, traceable, and machine-readable data set. When doctors need to access the data, they only need to locate the patient through the identifier to access their full historical images. This lays a solid and unified data foundation for any subsequent diagnostic and treatment activities that require historical data support.

[0040] Second, this invention eliminates spatial deviations in images caused by differences in body position and equipment through an image registration unit. Then, based on the registration results, a change detection unit objectively presents the density and size changes of lesions through quantitative indicators such as average CT values ​​and volume change rate. This process does not require manual intervention, avoids subjective judgment errors, and transforms image differences into calculable numerical indicators, upgrading the judgment of lesions from qualitative description to quantitative analysis, thus providing an objective basis for diagnosis and treatment decisions.

[0041] Other advantages, objectives and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination or study, or may be learned from the practice of the invention. Attached Figure Description

[0042] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0043] Figure 1 The flowchart is for the operation of a multi-time comparison system for radiological images based on a patient's unique identifier;

[0044] Figure 2 This is a diagram showing the module composition of a multi-time comparison system for radiographic cloud films based on a patient's unique identifier. Detailed Implementation

[0045] The technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0046] Example 1

[0047] This embodiment discloses the specific implementation process of a multi-time comparison system for radiographic image cloud films based on a patient's unique identifier, such as... Figure 2As shown, the system comprises a patient unique identifier resolution module, a cloud-based dynamic film archive, an intelligent image prefetching module, a multi-temporal comparison engine, and a dynamic visualization interface. The patient unique identifier resolution module generates a globally unique patient identifier, using the PUI as the core index. The cloud-based dynamic film archive structurally stores multi-temporal image data throughout the patient's lifecycle. The intelligent image prefetching module filters highly relevant historical images. The multi-temporal comparison engine completes image registration and quantitative analysis of lesion changes. Finally, the dynamic visualization interface provides a diverse display of multi-temporal images and presents the results of changes. The entire process eliminates the need for manual retrieval and comparison, effectively solving the problems of low efficiency, strong subjectivity, and scattered data in traditional image comparison, providing objective and accurate image analysis support for clinical diagnosis and treatment.

[0048] (a) Patient Unique Identifier Resolution Module

[0049] When a user needs to perform multi-time-to-multi-contrast analysis of a patient's radiological images, an authentication request is initiated through the system's Patient Unique Identifier (PUI) resolution module. The user inputs the patient's identity information, which may include one or more common identifying information such as the patient's name, ID number, or medical card number. After receiving this identity information, the PUI resolution module immediately establishes a communication connection with the cloud-based dynamic film archive, comparing the input identity information with the patient identity data stored in the archive to verify the validity of the patient's identity. If the authentication is successful, the module will generate a globally unique Patient Unique Identifier (PUI) based on preset coding rules. This PUI serves as the system's index key, possessing uniqueness and immutability. All subsequent image data and report data related to this patient will be stored and retrieved based on this PUI. If the authentication fails, the system will provide the user with a verification failure message, prompting the user to verify the identity information and re-enter it.

[0050] (II) Cloud Film Dynamic Archive

[0051] The cloud-based dynamic film archive uses the generated PUI as the first-level directory name to create a dedicated root directory for the patient's records, achieving independent storage and isolation of patient data. Within the root directory, the system organizes the data according to the time of each of the patient's examinations. A two-level directory structure is established, with each level corresponding to a complete dataset from a single examination. Within each level, a third-level directory is created to store various related data from that examination, including radiological image files, examination report documents, doctor-annotated image annotations, and clinical diagnostic data. All stored data is organized in a structured format, supporting multi-dimensional retrieval by examination time and image modality, providing efficient data access support for subsequent image prefetching and comparative analysis. When a patient undergoes a new radiological examination, the system automatically stores the image data and related data generated from that examination in the patient's corresponding PUI root directory, following the three-level directory structure described above, with the current examination time as the second-level directory name, and updates the patient's multi-temporal image dataset. ,in, Indicates the first The images and related data from each examination ensure the complete collection of the patient's imaging data throughout the entire life cycle.

[0052] (III) Intelligent Image Prefetching Module

[0053] When a doctor initiates a multi-time comparison request for the current examination, the intelligent image prefetching module first extracts the parameters of the current diagnostic scene, including the code of the current examination site. Current image modality coding and current clinical diagnostic codes This forms the current set of scene parameters for inspection. Subsequently, based on the patient's unique identifier (PUI), the module retrieves the complete multi-temporal image dataset corresponding to that patient from the cloud film dynamic archive. For each set of historical image data in the dataset , The module will extract the corresponding historical inspection area based on the historical inspection number. Historical image modalities and diagnostic keywords in historical related reports The module calculates historical image data using a prefetching algorithm. Relevance to the current inspection The relevance is calculated using the following formula: ,in, The similarity between historically inspected areas and currently inspected areas. The modal similarity between historical image modalities and current image modalities. The semantic similarity score between historical and current diagnostic keywords is used, with all three values ​​ranging from 0 to 1. α, β, and γ are weighting coefficients, satisfying α + β + γ = 1. Their default values ​​are set to α = 0.4, β = 0.3, and γ = 0.3, respectively. Doctors can adjust these weighting coefficients in the system according to actual clinical needs. After calculation, the module sorts all historical examination data in descending order based on the relevance score, filtering out the most relevant keywords. Historical data, θ is the correlation threshold, with a default value of 0.6, and the top N sets of historical image data are selected in descending order of score as the prefetch results, providing core analysis data for the subsequent multi-temporal comparison engine.

[0054] (iv) Multi-time comparison engine

[0055] The multi-temporal contrast engine, including an image registration unit and a change detection unit, is used to register and align pre-captured multi-temporal images and identify changes in lesion features.

[0056] The image registration unit receives N sets of historical image data output by the intelligent image prefetching module and the currently examined image. With current images For spatial reference, each set of historical images A three-dimensional non-rigid transformation is performed on the floating examination images, k=1, 2, ..., N, to ensure that the anatomical structures of historical images and current images are spatially registered and aligned. Specifically: First, from the current image... and historical images Extract feature point sets from each and ,in , , for The Middle Two-dimensional coordinates of a feature point for The Middle The two-dimensional coordinates of feature points are given by m, where m is the number of feature points. Next, the feature point set is calculated. and The matching degree of each feature point is used to eliminate mismatched feature point pairs using the RANSAC algorithm, retaining a set of valid matching point pairs M={( , Then, based on the set of valid matching point pairs, the spatial transformation matrix H is calculated. This matrix is ​​a 3×3 affine transformation matrix that satisfies the following formula: The transformation matrix H is solved using the least squares method, and the objective function is: a, b, c, d, e, and f are elements of the transformation matrix H. Finally, the historical images are... The transformation matrix H obtained by solving is mapped to the current image. The coordinate system is used to obtain the registered image. Complete the spatial registration of a single set of historical images with the current image. Repeat the above steps to complete the registration process of all pre-fetched historical images with the current image.

[0057] The change detection unit receives the registered historical image output by the image registration unit. and current images The process involves quantitative analysis of a series of registered and aligned images to identify and calculate the characteristic changes of the same lesion region at different time points. Specifically, firstly, a threshold segmentation method is used to analyze the current images. The affected areas were segmented and processed. Simultaneously, the registered historical images Perform the same threshold segmentation process to segment out the corresponding lesion areas. Next, the average CT value of the two lesion areas is calculated. The formula for calculating the average CT value is: Where CT(v) is the HU value of voxel v, and N(V) is the total number of voxels contained in region V. The current imaging lesion region is calculated using this formula. Average CT value and historical imaging lesion areas average Then, the density change rate of the lesion area was calculated based on the average CT value. This formula can quantitatively reflect the changes in lesion density at different examination time points. It calculates the rate of change in lesion size and volume. The lesion volume is determined by statistically analyzing the total number of voxels contained in the lesion area. With a single voxel volume The product of these terms yields the current imaging lesion area. volume Registered historical imaging lesion areas volume The volume change rate ,in, The actual spatial volume corresponding to a single voxel in the image. This indicates an increase in the size of the lesion. This indicates that the lesion volume has decreased. Finally, the change detection unit outputs quantitative analysis results, including the average CT value of the current lesion area. Average CT value of historical lesion area Average CT value Density change rate and volume change rate These key indicators provide quantitative data support for clinical diagnosis.

[0058] (v) Dynamic Visual Interface

[0059] The dynamic visualization interface maintains real-time communication with the multi-temporal comparison engine, receiving image registration results and quantitative indicators of change detection, providing doctors with diverse display and interactive functions. In terms of image display, the interface supports three modes: side-by-side, overlay, and dynamic switching. Side-by-side mode displays the current image alongside one or more selected historical registered images in the same section, facilitating intuitive comparison by doctors. Overlay mode displays the current image and historical images with varying transparency, highlighting changes in the lesion's location. Dynamic switching mode allows for quick switching between images at different time points via clicks or swipes, visually presenting the dynamic changes in the lesion. The integrated comparison sequence generation unit provides users with a timeline control, allowing doctors to view all examination time points for the patient on the timeline. Doctors can also select multiple time points from pre-fetched historical images, and the system automatically generates a centralized multi-temporal comparison sequence, supporting coherent comparative analysis of images from multiple time points. The lesion evolution unit communicates with the change detection unit, visualizing the quantitative results output by the change detection unit as two-dimensional charts. It can generate various chart types, such as line charts and bar charts, to intuitively display the changing trends of lesion characteristics over time, helping doctors quickly grasp the development of lesions. In addition, the interface includes a permission management and auditing unit, which communicates with the patient unique identifier resolution module and the cloud film dynamic archive. The system presets different access permission levels, such as doctor permissions, nurse permissions, and administrator permissions. Visitors with different permissions can only view data in the corresponding level directory of the cloud film dynamic archive, ensuring the security of patient privacy data. Simultaneously, this unit records all access behaviors, including visitor ID, access time, and operation content, generating detailed audit logs for subsequent traceability and management.

[0060] In summary, this embodiment fully presents the operational flow of a multi-time comparison system for radiological images based on a patient's unique identifier, verifying the system's feasibility and practicality. With the patient's unique identifier at its core, the system achieves structured integration and unified management of the patient's full-cycle radiological image data, solving the problem of scattered traditional image data. The intelligent image prefetching module accurately filters highly relevant historical images, improving the efficiency of comparative analysis. The multi-time comparison engine eliminates spatial bias through image registration and achieves objective analysis of lesion characteristics using quantitative formulas, avoiding subjective errors inherent in manual comparison. The dynamic visualization interface provides diverse display methods and interactive functions, allowing doctors to intuitively and efficiently obtain information on lesion changes. The entire system provides strong technical support for clinical diagnosis and treatment decisions.

[0061] Example 2

[0062] Based on Example 1, such as Figure 1 As shown, this embodiment provides an operation procedure for multi-time comparison of radiographic cloud films based on a patient's unique identifier. The specific steps of this procedure are as follows:

[0063] (1) Initiate a comparison request: The user enters the patient's identity information and initiates a multi-time comparison request.

[0064] (2) Authentication and PUI generation: The system verifies the validity of the patient's identity by connecting to the cloud film dynamic archive through the patient unique identifier resolution module, and generates a globally unique patient unique identifier (PUI).

[0065] (3) Retrieve patient image dataset: Using PUI as the index, retrieve the multi-temporal image dataset corresponding to all examination times of the patient from the cloud film dynamic archive.

[0066] (4) Extract parameters of the current diagnostic scenario: Obtain key information of the current examination, including examination site, imaging modality, and clinical diagnostic keywords.

[0067] (5) Screening highly relevant historical images: Based on the current scene parameters, the top N sets of historical image data with the highest relevance are screened from the patient's multi-temporal image dataset.

[0068] (6) Image registration unit registration and alignment:

[0069] The currently examined image is used as the spatial reference reference image.

[0070] Feature point sets for registration are extracted from the current reference image and each selected set of historical images.

[0071] Calculate the feature point matching degree, remove mismatched feature points, and retain valid matching point pairs.

[0072] Based on effective matching point pairs, the spatial transformation matrix for image alignment is solved.

[0073] By using a spatial transformation matrix, historical images are mapped to the coordinate system of the current reference image, thus completing the spatial alignment of the images.

[0074] (7) Change detection unit for identifying lesion feature changes:

[0075] Threshold segmentation was used to segment the corresponding lesion areas from the current image and the registered historical image, respectively.

[0076] Calculate the average CT value and density change rate of the two lesion regions, and other quantitative characteristic indicators.

[0077] (8) Generate comparison sequence: Users can use the timeline control to select multiple time point images to be compared and generate a multi-time comparison sequence.

[0078] (9) Image visualization display: Display multi-temporal images and registration results in side-by-side, overlay or dynamic switching modes.

[0079] (10) Visualization of lesion changes: The quantitative indicators are rendered in the form of two-dimensional charts to show the trend of lesion characteristics over time.

[0080] (11) Record access and operation: The access control and audit unit records the visitor ID, access time, operation content, and generates audit logs.

[0081] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A multi-time comparison system for radiological image cloud films based on a patient's unique identifier, characterized in that, The system consists of: a patient unique identifier resolution module, a cloud-based dynamic film archive, an intelligent image prefetching module, a multi-time comparison engine, and a dynamic visualization interface; The patient unique identifier resolution module is used to receive the patient identity information input by the user, verify the identity validity by connecting with the cloud film dynamic archive, and generate a globally unique patient unique identifier (PUI). The cloud-based dynamic film archive is used to store the patient's medical records at different examination times, using the PUI as the index key. Multi-temporal image dataset in structured storage format ,in, Indicates the first The images and related data from the second examination; The intelligent image pre-fetching module is used to pre-fetch images based on current diagnostic scenario parameters, including the examination site. Image modality Clinical diagnostic keywords Prefetching algorithm from multi-temporal image datasets The top results with the highest relevance were selected from the list. A set of historical image data; The multi-temporal comparison engine includes an image registration unit and a change detection unit, which are used to register and align pre-captured multi-temporal images and identify changes in lesion features. The dynamic visualization interface is connected in communication with the multi-temporal comparison engine and is used to display multi-temporal images in a side-by-side, overlay, and dynamic switching manner, and to visualize the change detection results.

2. The multi-time comparison system for radiographic images based on a patient's unique identifier according to claim 1, characterized in that, The structured storage format of the cloud film dynamic archive includes a three-level directory: First-level directory: named PUI, serving as the root directory for patient records; Second-level directory: By inspection time Name and store complete data for a single check; Third-level directory: Stores related data including radiological images, examination reports, image annotation information, and clinical diagnostic data; All data supports multi-dimensional retrieval by inspection time and image modality.

3. The multi-time comparison system for radiographic images based on a patient's unique identifier according to claim 1, characterized in that, The selection steps for the prefetching algorithm in the intelligent image prefetching module are as follows: Extract the scene parameters currently being inspected ( ),in, Code the inspection site. For image modality coding, For diagnostic coding; Retrieve the multi-temporal image dataset corresponding to the current patient's unique identifier (PUI) from the cloud film dynamic archive; Calculate the relevance score (Score) between each inspection in the historical dataset and the current inspection. All historical checks are sorted in descending order based on their relevance scores. Select relevance Historical data, The relevance threshold is set with a default value of 0.6, and the scores are ranked in descending order. Group.

4. The multi-time comparison system for radiographic images based on a patient's unique identifier according to claim 3, characterized in that, The relevance ,in, for Corresponding examination sites For part similarity, for Corresponding image modalities For modal similarity, for Diagnostic keywords in the related report This represents the semantic similarity of keywords, with a value ranging from 0 to 1. These are the weighting coefficients, and ,default value .

5. The multi-time comparison system for radiographic images based on a patient's unique identifier according to claim 1, characterized in that, In the multi-time comparison engine, the image registration unit is used to match the current image specified by the user. For spatial reference, historical images The floating examination image undergoes a three-dimensional non-rigid transformation to spatially register and align its anatomical structures with the current image. The specific steps are as follows: From the current image to be registered and historical images Extract the feature point set, which is respectively and ,in, for The Middle Two-dimensional coordinates of a feature point for The Middle Two-dimensional coordinates of feature points, k=1,2,…,N; Calculate the matching degree of feature points, eliminate false matches using the RANSAC algorithm, and retain the set of valid matching point pairs. , This represents the number of feature points; Calculate the spatial transformation matrix based on effective matching point pairs. ,Right now Where H is a 3×3 affine transformation matrix, and the matrix parameters (a, b, c, d, e, f) are solved using the least squares method. The objective function is: ; Historical images By transforming the matrix Mapped to the current image The coordinate system is used to obtain the registered image. .

6. The multi-time comparison system for radiographic cloud films based on a patient's unique identifier according to claim 1, characterized in that, In the multi-time comparison engine, the change detection unit is used to perform quantitative analysis on a series of images that have been registered and aligned, identify and calculate the changes in lesion features of the same region at different time points based on the registration, and the specific steps are as follows: Segment the current image using the threshold segmentation method. The lesion area in and registered historical images The lesion area in ; Calculate the average CT value of the two regions, based on The current imaging lesion area is calculated. Average CT value and historical imaging lesion areas average , where CT(v) is the HU value of voxel v, and N(V) is the total number of voxels contained in region V; Calculate the density change rate ΔCT, the ,in, For the lesion area CT value, For the lesion area CT value; Calculate the size and volume change rate of the lesion The lesion volume is determined by statistically analyzing the total number of voxels contained in the lesion area. With a single voxel volume The product of these terms yields the current imaging lesion area. volume Registered historical imaging lesion areas volume The volume change rate ,in, The actual spatial volume corresponding to a single voxel in the image; Output quantization results, including , , , , , , .

7. The multi-time comparison system for radiographic cloud films based on a patient's unique identifier according to claim 1, characterized in that, The dynamic visualization interface integrates: The comparison sequence generation unit provides users with a timeline control, allowing users to select images at multiple time points from pre-fetched historical images to generate a multi-time comparison sequence for centralized comparison. The lesion evolution unit is communicatively connected to the change detection unit in the multi-time comparison engine, and is used to visualize and render the calculated lesion feature change data in the form of a two-dimensional chart.

8. The multi-time comparison system for radiographic images based on a patient's unique identifier according to claim 1, characterized in that, The dynamic visualization interface also includes a permission management and auditing unit. The permission management and auditing unit is communicatively connected to the patient unique identifier resolution module and the cloud film dynamic archive. It is preset with visitor permission management, which allows the user to view different levels of directory data in the cloud film dynamic archive based on visitor permissions, and records all access behaviors, including visitor ID, access time, and operation content, and generates audit logs.

Images