Chest image report online reasoning method and system integrated with image quality evaluation
By denoising, enhancing, and extracting features from chest imaging data, imaging reports are generated and optimized, solving the problems of low intelligence and inconsistent quality in imaging reports, and improving the reliability of diagnosis.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-24
- Publication Date
- 2026-03-27
AI Technical Summary
In existing technologies, image reporting has a low level of intelligence, and the quality of chest images varies, affecting the reliability of diagnosis.
By denoising and enhancing chest imaging data, image quality is assessed, suspicious lesion areas are identified, standard feature data is extracted, lesion inference is performed, a basic imaging report is generated, and lesion trend analysis and optimization are conducted.
It has improved the structuring and intelligence of image reports, reduced the inconsistency in image quality, and improved the reliability of diagnosis.
Smart Images
Figure CN121745307A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application belongs to the technical field of image reporting, and particularly relates to a chest image reporting online reasoning method and system integrated with image quality evaluation. BACKGROUND
[0002] Image reporting is a professional medical document that systematically analyzes and comprehensively judges image data obtained after medical imaging examination, and records and conveys examination results in the form of text. Its core role is to convert complex image information into understandable and applicable medical conclusions, providing important basis for clinical doctors to make disease diagnosis, treatment decision and prognosis evaluation. Generally, it includes patient basic information, examination method, image findings, imaging diagnosis or suggestion, and necessary further examination or follow-up suggestion.
[0003] In the prior art, image reporting is gradually evolving towards structure and intelligence. However, at present, it is still at a low level of intelligence, and can only perform simple image reporting reasoning. It is often necessary to combine the image department doctors to systematically analyze and comprehensively judge the obtained image data. In actual medical scenarios, chest images are often of uneven quality due to different acquisition equipment and insufficient patient cooperation, affecting the reliability of diagnosis. SUMMARY
[0004] The purpose of the embodiments of the present application is to provide a chest image reporting online reasoning method and system integrated with image quality evaluation, aiming to solve the problems proposed in the background.
[0005] To achieve the above-mentioned purpose, the embodiments of the present application provide the following technical solutions: The chest image reporting online reasoning method integrated with image quality evaluation, specifically includes the following steps: Obtain chest image data of a target patient, denoise and enhance the chest image data to generate enhanced image data, and perform image quality evaluation to determine whether it has reasoning value; When it has reasoning value, perform region recognition and image feature extraction on the enhanced image data, select suspicious lesion regions, and extract standard feature data; According to the standard feature data, perform lesion reasoning on the suspicious lesion regions to determine the lesion position, lesion type, lesion size and lesion shape in the suspicious lesion regions; According to the lesion position, the lesion type, the lesion size and the lesion shape, generate a basic image report; Obtain historical report data of the target patient, perform lesion trend analysis, and optimize the basic image report to generate and display an optimized image report.
[0006] As a further limitation of the technical scheme of the embodiment of the present application, the chest image data of the target patient is acquired, the chest image data is denoised and enhanced to generate enhanced image data, and image quality evaluation is performed to determine whether it has inference value, which specifically comprises the following steps: Acquiring chest image data of a target patient; Performing image denoising processing on the chest image data to obtain denoised image data; Performing enhancement processing on the denoised image data to obtain enhanced image data; According to a plurality of preset quality indicators and a preset value quality standard, the enhanced image data is identified and quality evaluated to determine whether it has inference value.
[0007] As a further limitation of the technical scheme of the embodiment of the present application, the region identification and image feature extraction of the enhanced image data, the selection of the suspicious lesion region, and the extraction of the standard feature data specifically comprise the following steps: Region identification is performed on the enhanced image data to select a suspicious lesion region; Performing multi-scale feature analysis on the suspicious lesion region to extract multi-scale feature data; Standardizing the multi-scale feature data to obtain standard feature data.
[0008] As a further limitation of the technical scheme of the embodiment of the present application, the lesion inference of the suspicious lesion region according to the standard feature data to determine the lesion position, lesion type, lesion scale and lesion shape in the suspicious lesion region specifically comprises the following steps: According to the standard feature data, lesion detection is performed on the suspicious lesion region to determine the lesion position in the suspicious lesion region; According to the standard feature data, lesion identification is performed on the lesion position to determine the lesion type; According to the standard feature data, scale identification is performed on the lesion position to obtain the lesion scale; According to the standard feature data, shape identification is performed on the lesion position to obtain the lesion shape.
[0009] As a further limitation of the technical scheme of the embodiment of the present application, the generation of a basic image report according to the lesion position, the lesion type, the lesion scale and the lesion shape specifically comprises the following steps: Acquiring a standardized report template; Based on the standardized report template, a plurality of report related information is obtained from the lesion position, the lesion type, the lesion scale and the lesion shape; The report-related information is processed to generate report filling information. The report filling information is automatically filled in the standardized report template to generate a basic image report.
[0010] As a further limitation of the technical solutions of the embodiments of the present application, the historical report data of the target patient is obtained, the lesion trend analysis is performed, and the basic image report is optimized to generate and display an optimized image report, which specifically includes the following steps: Access data access rights of the target patient; Based on the data access rights, the historical report data of the target patient is obtained; The basic image report and the historical report data are integrated to perform lesion trend analysis and obtain lesion trend information; According to the lesion trend information, the recommended treatment information is matched; According to the lesion trend information and the recommended treatment information, the basic image report is optimized to generate an optimized image report; The optimized image report is displayed.
[0011] The chest image report online reasoning system integrated with image quality evaluation includes an image quality evaluation unit, an image feature extraction unit, a regional lesion reasoning unit, a basic report generation unit, and an image report optimization unit, wherein: The image quality evaluation unit is used to obtain chest image data of a target patient, denoises and enhances the chest image data to generate enhanced image data, and performs image quality evaluation to determine whether it has reasoning value; The image feature extraction unit is used to perform regional recognition and image feature extraction on the enhanced image data when it has reasoning value, select a suspicious lesion region, and extract standard feature data; The regional lesion reasoning unit is used to perform lesion reasoning on the suspicious lesion region according to the standard feature data to determine the lesion position, lesion type, lesion size, and lesion shape in the suspicious lesion region; The basic report generation unit is used to generate a basic image report according to the lesion position, lesion type, lesion size, and lesion shape; The image report optimization unit is used to obtain historical report data of the target patient, perform lesion trend analysis, and optimize the basic image report to generate and display an optimized image report.
[0012] As a further limitation of the technical solutions of the embodiments of the present application, the image quality evaluation unit specifically includes: An image data acquisition module is used to obtain chest image data of a target patient; an image denoising processing module, configured to perform image denoising processing on the chest image data to obtain denoised image data; an enhancement processing module, configured to perform enhancement processing on the denoised image data to obtain enhanced image data; a quality assessment module, configured to perform identification and quality assessment on the enhanced image data according to a plurality of preset quality indexes and a preset value quality standard, and determine whether the enhanced image data has inference value.
[0013] As a further limitation of the technical scheme of the embodiment of the present application, the basic report generation unit specifically comprises: a template acquisition module, configured to acquire a standardized report template; a report-related information acquisition module, configured to acquire a plurality of report-related information from the lesion position, the lesion type, the lesion scale and the lesion shape based on the standardized report template; a report filling information generation module, configured to process the plurality of report-related information to generate a plurality of report filling information; an automatic filling processing module, configured to perform automatic filling processing on the plurality of report filling information in the standardized report template to generate a basic image report.
[0014] As a further limitation of the technical scheme of the embodiment of the present application, the image report optimization unit specifically comprises: a permission acquisition module, configured to acquire data access permission of a target patient; a historical data acquisition module, configured to acquire historical report data of the target patient based on the data access permission; a lesion trend analysis module, configured to comprehensively analyze the basic image report and the historical report data to obtain lesion trend information; a treatment information matching module, configured to match suggested treatment information according to the lesion trend information; a report optimization module, configured to optimize the basic image report according to the lesion trend information and the suggested treatment information to generate an optimized image report; a report display module, configured to display the optimized image report.
[0015] Compared with the prior art, the present application has the following advantages: The embodiment of the present application can perform image quality evaluation on the chest image data after denoising and enhancement, judge whether it has reasoning value, extract standard feature data, perform lesion reasoning, generate a basic image report, and perform lesion trend analysis to optimize the basic image report. After denoising and enhancement on the chest image data, image quality evaluation is performed to judge whether it has reasoning value. When it has reasoning value, region recognition and image feature extraction are performed. Then, lesion reasoning is performed to generate a basic image report and perform lesion trend analysis to optimize the basic image report. The structured and intelligent level of image report reasoning is effectively improved. Through image quality evaluation and judgment screening, the situation of uneven image quality is reduced, and the reliability of diagnosis is improved. BRIEF DESCRIPTION OF DRAWINGS
[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application.
[0017] Figure 1 A flowchart of the method provided by the embodiment of the present application is shown.
[0018] Figure 2 A flowchart of image quality evaluation in the method provided by the embodiment of the present application is shown.
[0019] Figure 3 A flowchart of region recognition and image feature extraction in the method provided by the embodiment of the present application is shown.
[0020] Figure 4 A flowchart of lesion reasoning on the suspicious lesion region in the method provided by the embodiment of the present application is shown.
[0021] Figure 5 A flowchart of generating a basic image report in the method provided by the embodiment of the present application is shown.
[0022] Figure 6 A flowchart of generating an optimized image report in the method provided by the embodiment of the present application is shown.
[0023] Figure 7 An application architecture diagram of the system provided by the embodiment of the present application is shown.
[0024] Figure 8 A structure block diagram of the image quality evaluation unit in the system provided by the embodiment of the present application is shown.
[0025] Figure 9 A structure block diagram of the basic report generation unit in the system provided by the embodiment of the present application is shown.
[0026] Figure 10 The diagram shows the structural block diagram of the image report optimization unit in the system provided by an embodiment of the present invention. Detailed Implementation
[0027] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.
[0028] Understandably, current imaging reports are gradually evolving towards structured and intelligent technologies. However, they are still at a relatively low level of intelligence, only capable of simple inferences from the images. They often require radiologists to conduct systematic analysis and comprehensive judgment of the acquired imaging data. Furthermore, in actual medical scenarios, chest images often exhibit inconsistent quality due to differences in acquisition equipment and insufficient patient cooperation, affecting the reliability of diagnosis.
[0029] To address the aforementioned issues, this invention acquires chest imaging data from a target patient, denoises and enhances the data to generate enhanced image data, and performs image quality assessment to determine its inferential value. If inferential value is found, region identification and image feature extraction are performed on the enhanced image data to select suspicious lesion areas and extract standard feature data. Based on the standard feature data, lesion inference is performed on the suspicious lesion areas to determine the lesion location, type, scale, and shape. A basic imaging report is generated based on the lesion location, type, scale, and shape. Historical report data of the target patient is acquired, lesion trend analysis is performed, and the basic imaging report is optimized to generate and display an optimized imaging report. This method, which denoises and enhances chest imaging data, performs image quality assessment to determine its inferential value, and then performs lesion inference to generate a basic imaging report, analyzes lesion trends, and optimizes the basic imaging report, effectively improves the structured and intelligent level of imaging report inference. Furthermore, the image quality assessment and screening reduce inconsistencies in image quality, improving diagnostic reliability.
[0030] Figure 1 A flowchart of the method provided by an embodiment of the present invention is shown.
[0031] Specifically, the online inference method for chest image reports integrating image quality assessment includes the following steps: Step S101: Obtain chest image data of the target patient, denoise and enhance the chest image data to generate enhanced image data, and perform image quality assessment to determine whether it has inference value.
[0032] Specifically, Figure 2 A flowchart illustrating image quality assessment is shown in the method provided by an embodiment of the present invention.
[0033] In a preferred embodiment of the present invention, the steps of acquiring chest image data of the target patient, denoising and enhancing the chest image data to generate enhanced image data, and evaluating image quality to determine whether it has inference value specifically include the following steps: Step S1011: Obtain chest imaging data of the target patient; Step S1012: Perform image denoising processing on the chest image data to obtain denoised image data; Step S1013: Enhance the denoised image data to obtain enhanced image data; Step S1014: According to multiple preset quality indicators and preset value quality standards, the enhanced image data is identified and its quality is evaluated to determine whether it has inference value.
[0034] Specifically, the enhanced image data is identified and its quality assessed according to multiple preset quality indicators and preset value quality standards to determine whether it has inference value. The specific steps are as follows: The structural similarity index is obtained based on the preset quality indicators. Using the structural similarity index, the structural similarity of the enhanced image data is evaluated by calculating the similarity between the enhanced image data and the reference image in the preset value quality standard, and the structural similarity result is obtained. The peak signal-to-noise ratio (PSNR) is obtained based on the preset quality indicators; the PSNR of the enhanced image data is evaluated based on the structural similarity results and the PSNR to obtain the PSNR evaluation results. A first threshold is determined based on the structural similarity index and a preset value quality standard; the structural similarity assessment result is compared with the first threshold to obtain the structural similarity compliance result; A second threshold is determined based on the peak signal-to-noise ratio and a preset value quality standard; the signal-to-noise ratio evaluation result is compared with the second threshold to obtain the signal-to-noise ratio compliance result. Based on the structural similarity and signal-to-noise ratio compliance results, the chest images are comprehensively evaluated using preset quality indicators and preset value quality standards to obtain a comprehensive quality assessment result. The criteria for determining the inference value qualification are determined based on preset value quality standards and preset quality indicators. By matching the comprehensive quality judgment results with the inference value qualification criteria, it is determined whether the enhanced image data meets the preset inference value requirements, and a preliminary inference value judgment result is obtained. The preliminary inference value judgment results are verified by using the structural similarity and signal-to-noise ratio compliance results to obtain the final inference value judgment results; based on the final inference value judgment results, it is determined whether the enhanced image data has inference value.
[0035] Furthermore, this invention transforms subjective and ambiguous image quality judgments into objective, quantitative, and executable automated decisions through a multi-indicator, multi-stage, and verifiable process, laying a solid and reliable data foundation for all subsequent advanced image analysis functions (such as feature extraction and lesion inference).
[0036] Furthermore, the online inference method for chest image reports with integrated image quality assessment also includes the following steps: Step S102: When the enhanced image data has inference value, perform region identification and image feature extraction on the enhanced image data, select suspicious lesion areas, and extract standard feature data.
[0037] In this embodiment of the invention, when it is determined that the enhanced image data has inference value, region identification is performed on the enhanced image data, suspicious lesion areas are selected, and then the suspicious lesion areas are analyzed for multi-scale features such as texture features and morphological features to extract multi-scale feature data. After that, the multi-scale feature data is standardized to obtain standard feature data.
[0038] Specifically, Figure 3 A flowchart illustrating region identification and image feature extraction in the method provided by an embodiment of the present invention is shown.
[0039] In a preferred embodiment of the present invention, the step of performing region identification and image feature extraction on enhanced image data, selecting suspicious lesion areas, and extracting standard feature data specifically includes the following steps: Step S1021: Perform region identification on the enhanced image data and select suspicious lesion areas; Step S1022: Perform multi-scale feature analysis on the suspected lesion area and extract multi-scale feature data; Step S1023: Standardize the multi-scale feature data to obtain standard feature data.
[0040] Specifically, multi-scale feature analysis is performed on the suspected lesion area to extract multi-scale feature data. The specific steps are as follows: By utilizing the boundary information of the suspicious lesion area, the specific spatial location of the suspicious lesion area in the entire chest cavity image is determined through coordinate mapping. At the same time, the ensemble parameters of the suspicious lesion area contour are measured to obtain the spatial features of the lesion area; wherein, the spatial features of the lesion area include spatial location and ensemble size. Spatial localization in the spatial features of the lesion area is used to determine the extent of the lesion area. Gray values of pixels are sampled within the lesion area, and gray-level distribution statistics are calculated using the gray values of the pixels to obtain the gray-level statistical features of the lesion area. Among them, the gray-level distribution statistics include mean, variance, and distribution skewness. The lesion area is divided into windows, and the spatial distribution pattern of pixel gray values in each window is calculated using the gray-scale statistical features of the lesion area to obtain the texture quantization features of the lesion area. Based on the spatial features and texture quantization features of the lesion area, the lesion boundary is observed at different preset magnifications to evaluate the gradient change intensity and directional consistency of edge pixels and obtain multi-scale edge quantization features. By utilizing multi-scale edge quantization features, the uniformity of gray-level distribution within the lesion region is detected to obtain the quantization features of the internal structure of the lesion region. By utilizing the spatial features of the lesion region, the gray-scale statistical features of the lesion region, the texture quantization features of the lesion region, the multi-scale edge quantization features, and the internal structure quantization features of the lesion region, a structured dataset with the lesion region as the sole descriptor is constructed to obtain multi-scale feature data.
[0041] Furthermore, this invention transforms complex medical image regions into multi-dimensional, structured, and quantifiable digital objects, which not only improves the accuracy and reliability of subsequent reasoning but also makes it possible for precision medicine across time (such as efficacy assessment and disease progression monitoring).
[0042] Furthermore, the online inference method for chest image reports with integrated image quality assessment also includes the following steps: Step S103: Based on the standard feature data, perform lesion reasoning on the suspected lesion area to determine the lesion location, lesion type, lesion size, and lesion shape in the suspected lesion area.
[0043] In this embodiment of the invention, based on standard feature data, lesion detection is performed on the suspicious lesion area to determine the lesion location in the suspicious lesion area, and lesion identification is performed on the lesion location to determine the lesion type. At the same time, scale identification is performed on the lesion location to obtain the lesion scale, and shape identification is performed on the lesion location to obtain the lesion shape, thereby realizing the acquisition of lesion location, lesion type, lesion scale, and lesion shape.
[0044] Specifically, Figure 4 The flowchart illustrating lesion reasoning for suspected lesion areas in the method provided by an embodiment of the present invention is shown.
[0045] In a preferred embodiment of the present invention, the step of performing lesion reasoning on the suspected lesion area based on the standard feature data to determine the lesion location, lesion type, lesion size, and lesion shape in the suspected lesion area specifically includes the following steps: Step S1031: Based on the standard feature data, perform lesion detection on the suspected lesion area to determine the location of the lesion in the suspected lesion area; Step S1032: Based on the standard feature data, identify the lesion location and determine the lesion type; Step S1033: Based on the standard feature data, perform scale identification on the lesion location to obtain the lesion scale; Step S1034: Based on the standard feature data, perform shape recognition on the lesion location to obtain the lesion shape.
[0046] Furthermore, the online inference method for chest image reports with integrated image quality assessment also includes the following steps: Step S104: Generate a basic image report based on the lesion location, lesion type, lesion size, and lesion shape.
[0047] In this embodiment of the invention, a standardized report template is obtained. Based on the standardized report template, multiple report-related information is extracted from the lesion location, lesion type, lesion scale, and lesion shape. Then, key filtering is performed on the multiple report-related information to generate multiple report entry information. Finally, the multiple report entry information is automatically filled into the standardized report template to generate a basic imaging report.
[0048] Specifically, Figure 5 A flowchart illustrating the generation of a basic image report in the method provided by an embodiment of the present invention is shown.
[0049] In a preferred embodiment of the present invention, generating a basic imaging report based on the lesion location, lesion type, lesion size, and lesion shape specifically includes the following steps: Step S1041: Obtain the standardized report template; Step S1042: Based on the standardized report template, obtain multiple report-related information from the lesion location, the lesion type, the lesion size, and the lesion shape; Step S1043: Process the information related to the multiple reports to generate multiple report entry information; Step S1044: In the standardized report template, the multiple report input information are automatically filled in to generate a basic image report.
[0050] Furthermore, the online inference method for chest image reports with integrated image quality assessment also includes the following steps: Step S105: Obtain historical report data of the target patient, perform lesion trend analysis, optimize the basic imaging report, and generate and display the optimized imaging report.
[0051] In this embodiment of the invention, under the condition of obtaining data access permissions for the target patient, the historical report data of the target patient is obtained. By comprehensively processing the basic imaging report and the historical report data, the lesion trend of the target patient is analyzed and identified to obtain lesion trend information. Based on the lesion trend information, corresponding suggested treatment information is matched. Then, based on the lesion trend information and the suggested treatment information, the basic imaging report is optimized to generate an optimized imaging report, and then the optimized imaging report is displayed.
[0052] Specifically, Figure 6 A flowchart illustrating the method for generating an optimized image report provided in an embodiment of the present invention is shown.
[0053] In a preferred embodiment of the present invention, the steps of acquiring historical report data of the target patient, performing lesion trend analysis, optimizing the basic imaging report, and generating and displaying an optimized imaging report specifically include the following steps: Step S1051: Obtain data access permissions for the target patient; Step S1052: Based on the data access permissions, obtain the historical report data of the target patient; Step S1053: Combine the basic imaging report and the historical report data to perform lesion trend analysis and obtain lesion trend information; Step S1054: Match suggested treatment information based on the lesion trend information; Step S1055: Based on the lesion trend information and the suggested treatment information, optimize the basic imaging report to generate an optimized imaging report; Step S1056: Display the optimized image report.
[0054] Specifically, by combining the basic imaging report and the historical report data, lesion trend analysis is performed to obtain lesion trend information. The specific steps are as follows: The basic imaging report is used to obtain current lesion feature data and obtain a structured current lesion feature set; the structured current lesion feature set includes the current lesion location, current lesion type, current lesion scale, and current lesion shape; The matching range is set based on the current lesion location; historical lesion records are retrieved from historical report data using the matching range; historical lesion type, historical lesion scale, and historical lesion shape are extracted from the historical lesion records; historical lesion type, historical lesion scale, and historical lesion shape are integrated to obtain matching historical lesion feature data; Based on the current lesion feature set and the matched historical lesion feature data, the analysis is conducted on three aspects: consistency of lesion type, data changes in lesion scale, and morphological differences in lesion shape, in order to obtain the comparison results of lesion changes. By comparing the changes in lesion scale and the evolution of lesion shape at different time points in the lesion change results, the dynamic change pattern of the lesion can be determined. Timestamp information is obtained from historical report data; using the dynamic change pattern of lesions, the speed and direction of lesion scale changes are calculated based on the timestamp information, and a description of lesion development trend is generated; The differences in lesion scale and shape are obtained from the comparison results of lesion changes. Based on the description of lesion development trend, the differences in lesion scale and shape are transformed into a unified and comparable numerical measure to obtain a quantitative indicator of lesion trend. By combining quantitative indicators of lesion trends with descriptions of lesion development trends, the dynamic change patterns of lesions are combined with the quantitative magnitude of lesions to form a judgment result on the evolution of lesions over time and obtain lesion trend information.
[0055] Furthermore, this invention transforms raw image data into insights into disease evolution with profound clinical significance through rigorous lesion matching, multi-dimensional change analysis, and dynamic velocity quantification.
[0056] Specifically, by combining quantitative indicators of lesion trends with descriptions of lesion development trends, the dynamic change patterns of lesions are integrated with the quantitative magnitude of lesion changes to form a judgment on the evolution of lesions over time and obtain lesion trend information. The specific steps are as follows: Based on the differences in lesion scale values and lesion shape in the lesion trend quantification indicators, extract the quantification data of lesion scale values and shape to obtain the lesion quantification change amplitude value. Based on the speed and direction of lesion scale changes in the lesion development trend description, the rate of change of lesion scale over time and the lesion evolution trend are extracted to obtain lesion change rate features and lesion evolution trend features; based on the lesion change rate features and lesion evolution trend features, the lesion change direction and speed are generated. By utilizing the changing trends of lesion scale and the evolutionary patterns of lesion shape in the dynamic change patterns of lesions, the evolutionary patterns of lesions can be identified in terms of lesion type, scale, and shape, so as to obtain the evolutionary patterns of lesions. The overall trend of lesion change over time is calculated by analyzing the dynamic change pattern of the lesion. The quantitative change amplitude value of the lesion is combined with the change direction and speed of the lesion to obtain the overall trend of lesion change. By utilizing the patterns of lesion evolution and the overall trend of lesion changes, the development of lesions over time is evaluated in terms of three aspects: consistency of lesion type, data changes in lesion scale, and morphological differences in lesion shape, so as to obtain the lesion stability assessment results. By integrating the consistency between lesion stability assessment results and lesion type, a judgment result on the evolution of lesions over time is generated; the judgment result on the evolution of lesions over time is used to obtain lesion trend information.
[0057] Furthermore, this invention integrates the multi-source, heterogeneous data generated in the preceding steps into a "pathological trend information" that combines quantitative accuracy and qualitative insight and can directly serve clinical practice through a multi-level decision-making logic.
[0058] Furthermore, Figure 7 An application architecture diagram of the system provided in an embodiment of the present invention is shown.
[0059] In another preferred embodiment of the present invention, the online inference system for chest image reports integrating image quality assessment includes: The image quality assessment unit 101 is used to acquire chest image data of the target patient, denoise and enhance the chest image data to generate enhanced image data, and perform image quality assessment to determine whether it has inference value.
[0060] In this embodiment of the invention, the image quality assessment unit 101 acquires chest image data of the target patient, and then performs image denoising processing on the chest image data using Gaussian filtering, median filtering or a deep learning denoising model to obtain denoised image data. Then, the U-Net model is used to enhance the denoised image data to obtain enhanced image data. Subsequently, based on preset value quality standards, the enhanced image data is identified, quality assessed and compared according to multiple preset quality indicators such as structural similarity index and peak signal-to-noise ratio to determine whether it has inference value.
[0061] Specifically, Figure 8 A structural block diagram of the image quality assessment unit 101 in the system provided in an embodiment of the present invention is shown.
[0062] In a preferred embodiment of the present invention, the image quality assessment unit 101 specifically includes: The image data acquisition module 1011 is used to acquire chest image data of the target patient. The image denoising processing module 1012 is used to perform image denoising processing on the chest image data to obtain denoised image data. Enhancement processing module 1013 is used to enhance the denoised image data to obtain enhanced image data; The quality assessment module 1014 is used to identify and assess the quality of the enhanced image data according to multiple preset quality indicators and preset value quality standards, and to determine whether it has inference value.
[0063] Furthermore, the integrated image quality assessment online reasoning system for chest image reports also includes: The image feature extraction unit 102 is used to perform region identification and image feature extraction on enhanced image data when it has inference value, select suspicious lesion areas, and extract standard feature data.
[0064] In this embodiment of the invention, when it is determined that the image feature extraction unit 102 performs region identification on the enhanced image data, selects the suspicious lesion area, and then performs multi-scale feature analysis on the suspicious lesion area, such as texture features and morphological features, to extract multi-scale feature data. After that, the multi-scale feature data is standardized to obtain standard feature data.
[0065] The regional lesion reasoning unit 103 is used to perform lesion reasoning on the suspected lesion area based on the standard feature data, and to determine the lesion location, lesion type, lesion scale and lesion shape in the suspected lesion area.
[0066] In this embodiment of the invention, the regional lesion reasoning unit 103 performs lesion detection on the suspicious lesion area according to standard feature data, determines the lesion location in the suspicious lesion area, identifies the lesion location, determines the lesion type, and simultaneously performs scale identification on the lesion location to obtain the lesion scale and shape identification on the lesion location to obtain the lesion shape, thereby realizing the acquisition of lesion location, lesion type, lesion scale and lesion shape.
[0067] The basic report generation unit 104 is used to generate a basic image report based on the lesion location, the lesion type, the lesion size, and the lesion shape.
[0068] In this embodiment of the invention, the basic report generation unit 104 obtains a standardized report template, extracts multiple report-related information from lesion location, lesion type, lesion scale and lesion shape based on the standardized report template, performs key screening on the multiple report-related information, generates multiple report filling information, and then automatically fills in the multiple report filling information in the standardized report template to generate a basic image report.
[0069] Specifically, Figure 9 The diagram shows the structure of the basic report generation unit 104 in the system provided by an embodiment of the present invention.
[0070] In a preferred embodiment provided by the present invention, the basic report generation unit 104 specifically includes: Template acquisition module 1041 is used to acquire standardized report templates; The report-related information acquisition module 1042 is used to acquire multiple report-related information from the lesion location, the lesion type, the lesion size, and the lesion shape based on the standardized report template; The report entry information generation module 1043 is used to process multiple report-related information and generate multiple report entry information. The automatic data entry processing module 1044 is used to automatically enter multiple report entry information into the standardized report template to generate a basic image report.
[0071] Furthermore, the integrated image quality assessment online reasoning system for chest image reports also includes: The image report optimization unit 105 is used to acquire historical report data of the target patient, perform lesion trend analysis, optimize the basic image report, and generate and display the optimized image report.
[0072] In this embodiment of the invention, under the condition of obtaining data access permissions for the target patient, the image report optimization unit 105 obtains the historical report data of the target patient, analyzes and identifies the lesion trend of the target patient by comprehensively processing the basic image report and the historical report data, obtains lesion trend information, matches the corresponding suggested treatment information according to the lesion trend information, optimizes the basic image report according to the lesion trend information and the suggested treatment information, generates an optimized image report, and then displays the optimized image report.
[0073] Specifically, Figure 10 A structural block diagram of the image report optimization unit 105 in the system provided in an embodiment of the present invention is shown.
[0074] In a preferred embodiment provided by the present invention, the image report optimization unit 105 specifically includes: The permission acquisition module 1051 is used to acquire data access permissions for the target patient; The historical data acquisition module 1052 is used to acquire historical report data of the target patient based on the data access permissions. The lesion trend analysis module 1053 is used to integrate the basic imaging report and the historical report data to perform lesion trend analysis and obtain lesion trend information; The treatment information matching module 1054 is used to match suggested treatment information based on the lesion trend information; The report optimization module 1055 is used to optimize the basic image report based on the lesion trend information and the suggested treatment information, and generate an optimized image report; The report display module 1056 is used to display the optimized image report.
[0075] It should be understood that although the steps in the flowcharts of the various embodiments of the present invention are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the various embodiments may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least a portion of the sub-steps or stages of other steps.
[0076] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.
[0077] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0078] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.
[0079] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
Claims
1. An online inference method for chest image reports integrating image quality assessment, characterized in that, The method specifically includes the following steps: Acquire chest imaging data of the target patient, denoise and enhance the chest imaging data to generate enhanced image data, and perform image quality assessment to determine whether it has inference value; When the data has inference value, region identification and image feature extraction are performed on the enhanced image data to select suspicious lesion areas and extract standard feature data. Based on the standard feature data, lesion inference is performed on the suspected lesion area to determine the lesion location, lesion type, lesion size, and lesion shape in the suspected lesion area; A basic image report is generated based on the lesion location, lesion type, lesion size, and lesion shape. The system acquires historical report data of the target patient, performs lesion trend analysis, optimizes the basic imaging report, and generates and displays an optimized imaging report.
2. The online inference method for chest image reports with integrated image quality assessment according to claim 1, characterized in that, The process of acquiring chest imaging data of the target patient, denoising and enhancing the chest imaging data to generate enhanced image data, and then assessing the image quality to determine its inference value specifically includes the following steps: Acquire chest imaging data of the target patient; The chest image data is subjected to image denoising processing to obtain denoised image data; The denoised image data is enhanced to obtain enhanced image data; The enhanced image data is identified and its quality is evaluated according to multiple preset quality indicators and preset value quality standards to determine whether it has inference value.
3. The online inference method for chest image reports with integrated image quality assessment according to claim 2, characterized in that, The enhanced image data is identified and its quality assessed according to multiple preset quality indicators and preset value quality standards to determine whether it has inference value. The specific steps are as follows: The structural similarity index is obtained based on the preset quality indicators. Using the structural similarity index, the structural similarity of the enhanced image data is evaluated by calculating the similarity between the enhanced image data and the reference image in the preset value quality standard, and the structural similarity result is obtained. The peak signal-to-noise ratio is obtained based on preset quality indicators; Based on the structural similarity results and peak signal-to-noise ratio (PSNR), the enhanced image data is evaluated for PSNR to obtain the SNR evaluation results. The first threshold is determined based on the structural similarity index and the preset value quality standard; The structural similarity assessment results are compared with the first threshold to obtain the structural similarity compliance result; A second threshold is determined based on the peak signal-to-noise ratio and a preset value quality standard; the signal-to-noise ratio evaluation result is compared with the second threshold to obtain the signal-to-noise ratio compliance result. Based on the structural similarity and signal-to-noise ratio compliance results, the chest images are comprehensively evaluated using preset quality indicators and preset value quality standards to obtain a comprehensive quality assessment result. Determine the qualification conditions for inference value based on preset value quality standards and preset quality indicators; By matching the overall quality assessment results with the inference value qualification criteria, it is determined whether the enhanced image data meets the preset inference value requirements, and a preliminary inference value assessment result is obtained: The preliminary inference value judgment results are verified by using the structural similarity and signal-to-noise ratio compliance results to obtain the final inference value judgment results; The final inference value judgment result determines whether the enhanced image data has inference value.
4. The online inference method for chest image reports with integrated image quality assessment according to claim 3, characterized in that, The process of region identification and image feature extraction from enhanced image data, selecting suspicious lesion areas, and extracting standard feature data specifically includes the following steps: Region identification is performed on enhanced image data to select areas of suspected lesions; Multi-scale feature analysis was performed on the suspected lesion area to extract multi-scale feature data; The multi-scale feature data is standardized to obtain standard feature data.
5. The online inference method for chest image reports with integrated image quality assessment according to claim 4, characterized in that, Multi-scale feature analysis was performed on the suspected lesion area to extract multi-scale feature data. The specific steps are as follows: By utilizing the boundary information of the suspicious lesion area, the specific spatial location of the suspicious lesion area in the entire chest cavity image is determined through coordinate mapping. At the same time, the ensemble parameters of the suspicious lesion area contour are measured to obtain the spatial features of the lesion area; wherein, the spatial features of the lesion area include spatial location and ensemble size. Spatial localization in the spatial features of the lesion area is used to determine the extent of the lesion area. Gray values of pixels are sampled within the lesion area, and gray-level distribution statistics are calculated using the gray values of the pixels to obtain the gray-level statistical features of the lesion area. Among them, the gray-level distribution statistics include mean, variance, and distribution skewness. The lesion area is divided into windows, and the spatial distribution pattern of pixel gray values in each window is calculated using the gray-scale statistical features of the lesion area to obtain the texture quantization features of the lesion area. Based on the spatial features and texture quantization features of the lesion area, the lesion boundary is observed at different preset magnifications to evaluate the gradient change intensity and directional consistency of edge pixels and obtain multi-scale edge quantization features. By utilizing multi-scale edge quantization features, the uniformity of gray-level distribution within the lesion region is detected to obtain the quantization features of the internal structure of the lesion region. By utilizing the spatial features of the lesion region, the gray-scale statistical features of the lesion region, the texture quantization features of the lesion region, the multi-scale edge quantization features, and the internal structure quantization features of the lesion region, a structured dataset with the lesion region as the sole descriptor is constructed to obtain multi-scale feature data.
6. The online inference method for chest image reports with integrated image quality assessment according to claim 5, characterized in that, The step of performing lesion inference on the suspected lesion area based on the standard feature data to determine the lesion location, lesion type, lesion size, and lesion shape in the suspected lesion area specifically includes the following steps: Based on the standard feature data, lesion detection is performed on the suspected lesion area to determine the location of the lesion in the suspected lesion area; Based on the standard feature data, the lesion location is identified to determine the lesion type; Based on the standard feature data, the lesion location is scaled to obtain the lesion scale; Based on the standard feature data, shape recognition is performed on the lesion location to obtain the lesion shape.
7. The online inference method for chest image reports with integrated image quality assessment according to claim 6, characterized in that, The process of generating a basic imaging report based on the lesion location, lesion type, lesion size, and lesion shape specifically includes the following steps: Obtain a standardized report template; Based on the standardized report template, multiple report-related information are obtained from the lesion location, the lesion type, the lesion size, and the lesion shape; The information related to multiple reports is processed to generate multiple report entry information; In the standardized report template, multiple report input information are automatically filled in to generate a basic image report.
8. The online inference method for chest image reports with integrated image quality assessment according to claim 7, characterized in that, The process of acquiring historical report data of the target patient, performing lesion trend analysis, optimizing the basic imaging report, and generating and displaying the optimized imaging report specifically includes the following steps: Obtain access to the target patient's data; Based on the aforementioned data access permissions, historical report data of the target patient can be obtained; By combining the basic imaging report and the historical report data, lesion trend analysis is performed to obtain lesion trend information; Based on the lesion trend information, suggested treatment information is matched; Based on the lesion trend information and the recommended treatment information, the basic imaging report is optimized to generate an optimized imaging report; The optimized image report is then presented.
9. The online inference method for chest image reports with integrated image quality assessment according to claim 8, characterized in that, By combining the basic imaging report and the historical report data, lesion trend analysis is performed to obtain lesion trend information. The specific steps are as follows: The basic imaging report is used to obtain current lesion feature data and obtain a structured current lesion feature set; the structured current lesion feature set includes the current lesion location, current lesion type, current lesion scale, and current lesion shape; The matching range is set based on the current lesion location; historical lesion records are retrieved from historical report data using the matching range; historical lesion type, historical lesion scale, and historical lesion shape are extracted from the historical lesion records; historical lesion type, historical lesion scale, and historical lesion shape are integrated to obtain matching historical lesion feature data; Based on the current lesion feature set and the matched historical lesion feature data, the analysis is conducted on three aspects: consistency of lesion type, data changes in lesion scale, and morphological differences in lesion shape, in order to obtain the comparison results of lesion changes. By comparing the changes in lesion scale and the evolution of lesion shape at different time points in the lesion change results, the dynamic change pattern of the lesion can be determined. Timestamp information is obtained from historical report data; using the dynamic change pattern of lesions, the speed and direction of lesion scale changes are calculated based on the timestamp information, and a description of lesion development trend is generated; The differences in lesion scale and shape are obtained from the comparison results of lesion changes. Based on the description of lesion development trend, the differences in lesion scale and shape are transformed into a unified and comparable numerical measure to obtain a quantitative indicator of lesion trend. By combining quantitative indicators of lesion trends with descriptions of lesion development trends, the dynamic change patterns of lesions are combined with the quantitative magnitude of lesions to form a judgment result on the evolution of lesions over time and obtain lesion trend information.
10. The online inference method for chest image reports with integrated image quality assessment according to claim 8, characterized in that, By combining quantitative indicators of lesion trends with descriptions of lesion development trends, the dynamic change patterns of lesions are integrated with the quantitative magnitude of lesion changes to form a judgment on the evolution of lesions over time and obtain lesion trend information. The specific steps are as follows: Based on the differences in lesion scale values and lesion shape in the lesion trend quantification indicators, extract the quantification data of lesion scale values and shape to obtain the lesion quantification change amplitude value. Based on the speed and direction of lesion scale changes in the lesion development trend description, the rate of change of lesion scale over time and the lesion evolution trend are extracted to obtain lesion change rate characteristics and lesion evolution trend characteristics. The direction and speed of lesion change are generated based on the characteristics of lesion change rate and lesion evolution trend. By utilizing the changing trends of lesion scale and the evolutionary patterns of lesion shape in the dynamic change patterns of lesions, the evolutionary patterns of lesions can be identified in terms of lesion type, scale, and shape, so as to obtain the evolutionary patterns of lesions. The overall trend of lesion change over time is calculated by analyzing the dynamic change pattern of the lesion. The quantitative change amplitude value of the lesion is combined with the change direction and speed of the lesion to obtain the overall trend of lesion change. By utilizing the patterns of lesion evolution and the overall trend of lesion changes, the development of lesions over time is evaluated in terms of three aspects: consistency of lesion type, data changes in lesion scale, and morphological differences in lesion shape, so as to obtain the lesion stability assessment results. By integrating the consistency between lesion stability assessment results and lesion type, a judgment result on the evolution of lesions over time is generated; the judgment result on the evolution of lesions over time is used to obtain lesion trend information.
11. An online inference system for chest image reports integrating image quality assessment, characterized in that, The system employs the online inference method for chest image reports with integrated image quality assessment as described in any one of claims 1 to 10 above, and the system comprises: The image quality assessment unit is used to acquire chest image data of the target patient, denoise and enhance the chest image data to generate enhanced image data, and perform image quality assessment to determine whether it has inference value. The image feature extraction unit is used to perform region identification and image feature extraction on enhanced image data when it has inference value, select suspicious lesion areas, and extract standard feature data. The regional lesion reasoning unit is used to perform lesion reasoning on the suspected lesion region based on the standard feature data, and to determine the lesion location, lesion type, lesion scale and lesion shape in the suspected lesion region; The basic report generation unit is used to generate a basic image report based on the lesion location, the lesion type, the lesion size, and the lesion shape. The image report optimization unit is used to acquire historical report data of the target patient, perform lesion trend analysis, optimize the basic image report, and generate and display the optimized image report.