DR image analysis method and device, medium and product

Abstract

The invention relates to the technical field of image analysis, in particular to a DR image analysis method and device, a medium and a product, and the method comprises the steps: carrying out the preprocessing based on a DR image, and obtaining a target DR image, and the preprocessing is used for improving the image quality of the DR image. In order to improve the accuracy and stability of image segmentation and DR image analysis and facilitate capturing details of complex texture features in a DR image, a self-adaptive threshold adjustment method is utilized to perform threshold dynamic calculation on a target DR image, target thresholds corresponding to a plurality of local image areas are determined, and the image segmentation and DR image analysis accuracy and stability are improved. The adaptive threshold adjustment method dynamically calculates the threshold according to the brightness distribution of different areas so as to adapt to the difference between different DR images. And performing feature extraction on the target local image region by using the target threshold, and determining image features corresponding to the target local image. And finally, DR image analysis is carried out based on the image features corresponding to each target local image, and DR image indexes are obtained.

Description

Technical Field

[0001] The present application relates to the technical field of image analysis, and in particular to a DR image analysis method, device, medium and product. Background Art

[0002] Digital radiography (DR) technology is a non-destructive testing technology that uses digital detectors instead of traditional films to process and display X-ray images through computers. It is widely used in the field of medical imaging and has the advantages of real-time imaging, high resolution, excellent image quality, and environmental protection. In the field of medical imaging, DR technology can acquire and display X-ray images in real time, significantly improving the speed and efficiency of detection, while providing high-resolution images, clearly showing subtle defects and structural details, and providing doctors with accurate diagnostic basis.

[0003] With the advancement of AI technology, deep learning-based image analysis has begun to be integrated into DR imaging systems, which has improved the accuracy and efficiency of lesion detection to a certain extent. Current DR image analysis usually relies on preset thresholds for image segmentation and feature extraction to identify abnormal areas in DR images. However, existing DR image analysis methods are easily affected by noise. In the case of complex pathology, the preset threshold method often has difficulty in accurately capturing complex features, resulting in inaccurate image segmentation results, which makes the accuracy of DR image analysis low.

[0004] Therefore, how to improve the accuracy of DR image analysis methods is an urgent problem to be solved by those skilled in the art. Summary of the invention

[0005] The purpose of this application is to provide a DR image analysis method, device, medium and product to solve at least one of the above technical problems.

[0006] The above invention objectives of the present application are achieved through the following technical solutions: In a first aspect, the present application provides a DR image analysis method, which adopts the following technical solution: A DR image analysis method, comprising: Acquire a DR image, and perform preprocessing based on the DR image to obtain a target DR image, wherein the preprocessing is used to improve the image quality of the DR image; Performing dynamic threshold calculation on the target DR image using an adaptive threshold adjustment method to determine target thresholds corresponding to a plurality of local image regions, wherein the plurality of local image regions are obtained by segmenting the target DR image; Using the target threshold to perform feature extraction on a target local image region, and determining image features corresponding to the target local image, wherein the target local image region is any local image region; DR image analysis is performed based on the image features corresponding to each of the target local images to obtain DR image indicators.

[0007] By adopting the above technical solution, preprocessing is performed based on the DR image to obtain a target DR image, wherein the preprocessing is used to improve the image quality of the DR image. In order to improve the accuracy and stability of image segmentation and DR image analysis, and to facilitate capturing the details of complex texture features in the DR image, an adaptive threshold adjustment method is used to dynamically calculate the threshold of the target DR image, and the target thresholds corresponding to multiple local image regions are determined. The adaptive threshold adjustment method dynamically calculates the threshold according to the brightness distribution of different regions, thereby adapting to the differences between different DR images. Then, the target threshold is used to extract features from the target local image region to determine the image features corresponding to the target local image. Finally, DR image analysis is performed based on the image features corresponding to each target local image to obtain DR image indicators. The DR image indicators determined after performing DR image analysis provide quantitative image feature information, which helps doctors understand the DR image content more intuitively and enhances the interpretability of the image.

[0008] In a preferred example, the present application may be further configured as follows: after the preprocessing is performed based on the DR image to obtain the target DR image, the following further includes: Performing image conversion based on the target DR image to obtain an energy distribution map corresponding to the target DR image; Extracting features based on the energy distribution diagram to determine energy features, wherein the energy features include: energy density features, energy center of gravity features, and energy effective value features; Accordingly, DR image analysis is performed based on the image features corresponding to each of the target local images to obtain DR image indicators, including: DR image analysis is performed based on the energy feature and the image feature corresponding to each of the target local images to obtain a DR image index.

[0009] In a preferred example, the present application may be further configured as follows: performing DR image analysis based on the energy feature and the image feature corresponding to each of the target local images to obtain DR image indicators includes: Using morphological analysis technology, perform morphological analysis on each of the target local images to determine anatomical structure indicators; Performing feature fusion based on the energy density feature, the energy center of gravity feature and the energy effective value feature in the energy feature to obtain an energy feature vector, and performing energy calculation based on the energy feature vector to determine an energy feature index; Acquire a pathology recognition model, perform pathology recognition and quantification on the texture features in each of the image features based on the pathology recognition model, and determine a pathology quantification index; The anatomical structure index, the energy characteristic index and the pathological quantitative index are integrated to obtain a DR image index.

[0010] In a preferred example, the present application may be further configured as follows: after performing DR image analysis based on the energy feature and the image feature corresponding to each of the target local images to obtain the DR image index, the configuration further includes: Performing multi-dimensional content expansion based on each of the DR image indicators to obtain expansion information corresponding to the DR image indicators, wherein the multi-dimensional content expansion is used to expand the content from the indicator meaning, calculation method and normal range; Based on the anatomical structure index, the energy characteristic index, the pathological quantification index and the extended information, data marking is performed in the DR image to obtain a comprehensive image interpretation.

[0011] In a preferred example, the present application may be further configured as follows: after performing multi-dimensional content expansion based on each indicator in the DR image indicators and obtaining the expansion information corresponding to the DR image indicators, the method further includes: Based on the normal range of the indicator in the extended information, abnormal indicator detection is performed on the DR image indicator to determine the abnormal indicator, wherein the abnormal indicator is indicator information of the DR image indicator that is not within the normal range of the indicator; Based on the abnormal index, the abnormal area is locked, the abnormal local image area and the corresponding abnormal image features are determined, and three-dimensional modeling is performed based on the abnormal image features to determine the abnormal structure three-dimensional model.

[0012] In a preferred example, the present application may be further configured as follows: the adaptive threshold adjustment method is used to dynamically calculate the threshold of the target DR image to determine the target thresholds corresponding to the multiple local image regions, including: The threshold calculation parameters of the adaptive threshold calculation are obtained, and based on the Gaussian weighting method in the adaptive threshold adjustment method and the threshold calculation parameters, the threshold dynamic calculation is performed on the target DR image to determine the target thresholds corresponding to the multiple local image areas respectively.

[0013] In a second aspect, the present application provides an electronic device, which adopts the following technical solution: at least one processor; Memory; At least one application, wherein the at least one application is stored in the memory and configured to be executed by at least one processor, and the at least one application is configured to: execute the above-mentioned DR image analysis method.

[0014] In a third aspect, the present application provides a computer-readable storage medium, which adopts the following technical solution: A computer-readable storage medium stores a computer program, which, when executed in a computer, causes the computer to execute the above-mentioned DR image analysis method.

[0015] In a fourth aspect, the present application provides a computer program product, which adopts the following technical solution: A computer program product comprises a computer program, wherein when the computer program is executed by a processor, the computer program implements the above-mentioned DR image analysis method.

[0016] In summary, the present application includes at least one of the following beneficial technical effects: Preprocessing is performed based on the DR image to obtain a target DR image, wherein the preprocessing is used to improve the image quality of the DR image. In order to improve the accuracy and stability of image segmentation and DR image analysis, and to facilitate capturing the details of complex texture features in DR images, an adaptive threshold adjustment method is used to dynamically calculate the threshold of the target DR image, and the target thresholds corresponding to multiple local image regions are determined. The adaptive threshold adjustment method dynamically calculates the threshold according to the brightness distribution of different regions, thereby adapting to the differences between different DR images. Then, the target threshold is used to extract features from the target local image region to determine the image features corresponding to the target local image. Finally, DR image analysis is performed based on the image features corresponding to each target local image to obtain DR image indicators. The DR image indicators determined after performing DR image analysis provide quantitative image feature information, which helps doctors understand the DR image content more intuitively and enhances the interpretability of the image.

[0017] Image conversion is performed based on the target DR image to obtain the energy distribution map corresponding to the target DR image. Then, feature extraction is performed based on the energy distribution map to determine the energy features. Furthermore, DR image analysis is performed based on the energy features and the image features corresponding to each target local image to obtain DR image indicators. In energy space, noise usually appears as random fluctuations with low energy values, so the energy space is converted and more accurate energy features are extracted to improve the accuracy of DR image analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a flowchart of a DR image analysis method according to one embodiment of the present application; Figure 2It is a structural schematic diagram of a DR image analysis device according to one embodiment of the present application; Figure 3 It is a structural schematic diagram of an electronic device according to one embodiment of the present application. DETAILED DESCRIPTION

[0019] The following combination Figures 1 to 3 This application is described in further detail.

[0020] This specific embodiment is merely an explanation of the present application and is not a limitation of the present application. After reading this specification, a person skilled in the art may make non-creative modifications to the present embodiment as needed, but such modifications are protected by the patent law as long as they are within the scope of the present application.

[0021] In order to make the purpose, technical scheme and advantages of the embodiment of the present application clearer, the technical scheme in the embodiment of the present application will be clearly and completely described in conjunction with the drawings in the embodiment of the present application. Obviously, the described embodiment is a part of the embodiment of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present application. It should be noted that in the optional embodiments of the present application, the object information and other related data involved, when the embodiments in the present application are applied to specific products or technologies, need to obtain the permission or consent of the object, and the collection, use and processing of the relevant data need to comply with the relevant laws, regulations and standards of the relevant countries and regions. In other words, if the data related to the object is involved in the embodiment of the present application, it needs to be obtained through the authorization and consent of the object, the authorization and consent of the relevant departments, and in accordance with the relevant laws, regulations and standards of the country and region. If personal information is involved in the embodiment, the acquisition of all personal information needs to obtain the consent of the individual. If sensitive information is involved, the separate consent of the information subject needs to be obtained, and the embodiment also needs to be implemented with the authorization and consent of the object.

[0022] In addition, the term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article, unless otherwise specified, generally means that the associated objects before and after are in an "or" relationship.

[0023] The embodiments of the present application are further described in detail below in conjunction with the drawings in the specification.

[0024] The embodiment of the present application provides a DR image analysis method, which is executed by an electronic device, which can be a server or a terminal device, wherein the server can be an independent physical server, a server cluster or a distributed system composed of multiple physical servers, or a cloud server that provides cloud computing services. The terminal device can be a smart phone, a tablet computer, a laptop computer, a desktop computer, etc., but is not limited thereto. The terminal device and the server can be directly or indirectly connected via wired or wireless communication, which is not limited in the embodiment of the present application. Figure 1 As shown, the method includes step S101, step S102 and step S103, wherein: Step S101: Acquire a DR image, and perform preprocessing based on the DR image to obtain a target DR image, wherein the preprocessing is used to improve the image quality of the DR image.

[0025] For the embodiment of the present application, the DR image is a digital X-ray photography image, which is the product of a digital X-ray photography system. It is a digital matrix image converted by converting an analog video signal through computer digital processing. To obtain a DR image, the electronic device communicates with the DR imaging device wirelessly so that the electronic device can obtain a clear and complete DR image from the DR imaging device. The generation process of the DR image is as follows: In the field of medical imaging technology, the DR imaging device is controlled to emit X-rays to the target tissue part of the target object, and the X-rays will penetrate the target tissue part. Then, the DR imaging device receives the X-rays after passing through the target tissue part, and processes the received X-ray signals, including sampling, analog / digital conversion and other steps, and finally obtains a DR image of digital X-ray photography.

[0026] Then, preprocessing is performed based on the DR image to obtain the target DR image. The preprocessing operation is aimed at improving the image quality of the high DR image, and usually includes but is not limited to: denoising, contrast enhancement, distortion correction and other operations, among which image denoising is to apply a filter (for example, a Gaussian filter, a median filter, etc.) to remove noise in the image; contrast enhancement can adjust the contrast of the image by using techniques such as histogram equalization and adaptive contrast enhancement, which helps to improve the visibility of image details and make the lesion area clearer; distortion correction is used to correct geometric distortion, brightness unevenness, etc. in the image.

[0027] Step S102: dynamically calculating the threshold of the target DR image using an adaptive threshold adjustment method to determine target thresholds corresponding to a plurality of local image regions, wherein the plurality of local image regions are obtained by segmenting the target DR image; Step S103: extracting features of a target local image region using a target threshold value to determine image features corresponding to the target local image, wherein the target local image region is any local image region.

[0028] For the embodiments of the present application, the existing DR image analysis method uses a preset threshold to perform image segmentation and feature extraction. However, in the case of complex pathology, it is impossible to accurately distinguish areas of different brightness or contrast, which may lead to mis-segmentation of the target area and the background area, affecting the subsequent feature extraction and image analysis, so that the accuracy of DR image analysis is low.

[0029] In order to improve the accuracy and stability of image segmentation and DR image analysis methods, and to facilitate capturing the details of complex texture features in DR images, an adaptive threshold adjustment method is used to dynamically calculate the threshold of the target DR image, and determine the target thresholds corresponding to multiple local image areas, wherein the adaptive threshold adjustment method dynamically calculates the threshold according to the brightness distribution of different areas, thereby adapting to the differences between different DR images. The adaptive threshold method can calculate its local threshold according to the brightness distribution of different areas of the target DR image, and can cope with the brightness and contrast changes of different areas in the target DR image to improve the accuracy of image segmentation. There are many ways to dynamically calculate the threshold, which are no longer limited in the embodiments of the present application. In an achievable method, the threshold calculation parameters of the adaptive threshold calculation are obtained, and based on the Gaussian weighting method and the threshold calculation parameters in the adaptive threshold adjustment method, the threshold is dynamically calculated for the target DR image, and the target thresholds corresponding to multiple local image areas are determined, wherein the threshold calculation parameters specify the area size of each local image area.

[0030] Furthermore, the target threshold is an optimal threshold dynamically calculated based on the brightness and contrast of the local image area, and the target threshold can accurately distinguish the target area from the background area in the local image area. Therefore, the pixel value of the target local image area is compared with the corresponding target threshold. If the pixel value is greater than or equal to the target threshold, the pixel value is set to white (or a specific value), otherwise it is set to black (or another specific value), so that the target local image area is converted into a binary image. Then, the binary target local image area is feature extracted to determine the image features corresponding to the target local image, wherein the image features include but are not limited to: texture features, shape features, and statistical features. The shape features mainly describe the outline, size, shape and other information of the target in the target DR image, and the statistical features are used to describe the gray value distribution and statistical characteristics of the pixels in the target DR image.

[0031] Step S104: performing DR image analysis based on the image features corresponding to each target local image to obtain DR image indicators.

[0032] For the embodiment of the present application, DR image analysis is performed based on the image features corresponding to each target local image to obtain DR image indicators, wherein the DR image indicators include but are not limited to: anatomical structure indicators, energy feature indicators, pathological quantification indicators, etc. Of course, other indicators may also be included, and the embodiment of the present application will not be limited to this. There are many ways to analyze DR images, and users can select them according to actual needs. In one feasible way, morphological analysis technology is used to perform morphological analysis on the shape features in the image features to determine the anatomical structure indicators; based on the energy density features, energy center of gravity features and energy effective value features in the energy features, feature fusion is performed to obtain energy feature vectors, and abnormality detection is performed based on the energy feature vectors to determine energy feature indicators; a pathological recognition model is obtained, and the texture features in each image feature are pathologically recognized and quantified based on the pathological recognition model to determine the pathological quantification indicators; the anatomical structure indicators, energy feature indicators and pathological quantification indicators are combined to obtain DR image indicators. The DR image indicators determined after performing DR image analysis provide quantitative image feature information, which can be used as a reference for clinical decision-making, helping doctors to understand the DR image content more intuitively and enhance the interpretability of the image.

[0033] It can be seen that in the embodiment of the present application, preprocessing is performed based on the DR image to obtain a target DR image, wherein the preprocessing is used to improve the image quality of the DR image. In order to improve the accuracy and stability of image segmentation and DR image analysis, and to facilitate capturing the details of complex texture features in DR images, an adaptive threshold adjustment method is used to dynamically calculate the threshold of the target DR image, and the target thresholds corresponding to multiple local image regions are determined. The adaptive threshold adjustment method dynamically calculates the threshold according to the brightness distribution of different regions, thereby adapting to the differences between different DR images. Then, the target threshold is used to extract features from the target local image region to determine the image features corresponding to the target local image. Finally, DR image analysis is performed based on the image features corresponding to each target local image to obtain DR image indicators. The DR image indicators determined after performing DR image analysis provide quantitative image feature information, which helps doctors understand the DR image content more intuitively and enhances the interpretability of the image.

[0034] Furthermore, in order to extract more accurate energy features and improve the accuracy of DR image analysis, in the embodiment of the present application, after preprocessing the DR image to obtain the target DR image, the following steps are further included: Perform image conversion based on the target DR image to obtain an energy distribution map corresponding to the target DR image; Extracting features based on the energy distribution diagram to determine energy features, wherein the energy features include: energy density features, energy center of gravity features, and energy effective value features; Accordingly, DR image analysis is performed based on the image features corresponding to each target local image to obtain DR image indicators, including: DR image analysis is performed based on the energy features and the image features corresponding to each target local image to obtain DR image indicators.

[0035] For the embodiments of the present application, in the energy space, noise usually appears as random fluctuations with low energy values. By setting a reasonable energy threshold, it is convenient to effectively filter out the noise and extract more accurate energy features, thereby improving the accuracy of DR image analysis.

[0036] Therefore, an orthogonal transformation is performed based on the target DR image, such as Fourier transform, wavelet transform, etc., to convert the image from the spatial domain to the energy domain. After the orthogonal transformation, the energy distribution map corresponding to the target DR image can be obtained. The energy distribution map can clearly show the energy difference in different regions, which helps to more accurately identify the lesion area. Then, feature extraction is performed based on the energy distribution map to determine the energy features. The energy features include: energy density features, energy center of gravity features, and energy effective value features. Among them, the energy density feature is used to characterize the energy density of different regions or pixels in the DR image. The area with higher energy density may correspond to the tissue structure intensive or lesion area; the energy center of gravity feature is used to characterize the main energy concentration area in the DR image. The energy center of gravity corresponds to the important tissue structure or lesion area. By tracking the change of the energy center of gravity, the development of the lesion area can be monitored; the energy effective value feature is used to characterize the overall level or intensity of energy in the DR image, reflecting the overall brightness and contrast information of the DR image. By comparing the energy effective value features of different DR images or the same DR image at different time points, the quality change of the DR image and the development trend of the lesion can be evaluated. Accordingly, DR image analysis is performed based on energy features and image features corresponding to each target local image to obtain DR image indicators. DR image analysis is performed from two dimensions: energy features and image features, which improves the accuracy and comprehensiveness of DR image analysis.

[0037] It can be seen that in the embodiment of the present application, image conversion is performed based on the target DR image to obtain an energy distribution map corresponding to the target DR image. Then, feature extraction is performed based on the energy distribution map to determine the energy feature. Furthermore, DR image analysis is performed based on the energy feature and the image feature corresponding to each target local image to obtain the DR image index. In the energy space, noise usually appears as random fluctuations with low energy values, so the energy space is converted and more accurate energy features are extracted to improve the accuracy of DR image analysis.

[0038] Furthermore, in order to consider the correlation and complementarity between different energy features and improve the accuracy and reliability of energy detection, in the embodiment of the present application, DR image analysis is performed based on the energy features and the image features corresponding to each target local image to obtain DR image indicators, including: Using morphological analysis technology, morphological analysis is performed on each target local image to determine the anatomical structure indicators; Based on the energy density feature, energy center of gravity feature and energy effective value feature in the energy feature, feature fusion is performed to obtain an energy feature vector, and energy calculation is performed based on the energy feature vector to determine the energy feature index; Acquire a pathology recognition model, perform pathology recognition and quantification on the texture features in each image feature based on the pathology recognition model, and determine a pathology quantification index; The DR image index is obtained by integrating the anatomical structure index, energy characteristic index and pathological quantitative index.

[0039] For the embodiment of the present application, a morphological operation is performed on each target local image, and the morphological operation includes but is not limited to: corrosion operation, expansion operation, opening operation and closing operation. Then, a boundary extraction is performed based on the result after the morphological operation to extract the boundary of the anatomical structure. The boundary extraction is usually determined by subtracting the corrosion result from the original target DR image after the corrosion operation. At the same time, a skeleton extraction is performed based on the result after the morphological operation to extract the skeleton of the anatomical structure, which is the center line or main supporting structure of the shape in the target DR image. At the same time, the connected area in the target DR image is analyzed to determine the key indicators such as the number, size, and position of the anatomical structure. Furthermore, the boundary, skeleton, and connected area of ​​the structure are used for directional measurement to determine the anatomical structure indicators, wherein the directional measurement is used to quantitatively evaluate the anatomical structure, and the anatomical structure indicators are not limited to: boundary length, skeleton length, number of skeleton branches, skeleton diameter, number of connected areas, area of ​​connected areas, and connectivity strength.

[0040] Then, based on the energy density feature, energy center of gravity feature and energy effective value feature in the energy feature, feature fusion is performed to obtain an energy feature vector. The feature fusion operation comprehensively considers the correlation and complementarity between different energy features, and improves the accuracy and reliability of energy detection. For feature fusion, the energy density feature, energy center of gravity feature and energy effective value feature can be combined in a serial fusion or parallel fusion manner to form an energy feature vector containing multiple eigenvalues, and the fused energy feature vector is normalized or standardized to eliminate the dimension differences and numerical range differences between different features. Then, energy calculation is performed based on the energy feature vector to determine the energy feature index. The energy calculation is used to quantify the energy feature. The energy feature index includes but is not limited to: energy density, energy center of gravity offset, and energy effective value. Among them, energy density refers to the energy value per unit area or unit volume, which can reflect the activity or functional state of the structure; energy center of gravity offset refers to the offset of the energy center of gravity relative to the geometric center of the anatomical structure, which can be used to evaluate whether the energy distribution inside the anatomical structure is uniform; energy effective value refers to the root mean square of all eigenvalues ​​in the energy feature vector, which is used to quantify the overall energy level of the anatomical structure, and can reflect the stability and functional characteristics of the structure.

[0041] The electronic device pre-stores a pathology recognition model, which is obtained by training the model using a large number of annotated image data sets so that the model can learn the mapping relationship between texture features and pathology features. Then, the texture features in each image feature are pathologically recognized based on the pathology recognition model to determine the pathology type, and the texture features are quantitatively calculated to determine the pathology quantification index, where the pathology quantification index includes but is not limited to: pathology feature score, pathology feature probability and pathology feature confidence. The pathology feature score can reflect the obviousness, severity or malignancy of the pathology feature; the pathology feature probability refers to the possibility or probability of the existence of the pathology feature; the pathology feature confidence indicates the degree of certainty or reliability of the prediction result of the pathology recognition model. The pathology quantification index provides an objective basis for doctors, which helps doctors to judge the patient's condition and prognosis more accurately. Finally, the anatomical structure index, energy feature index and pathology quantification index are combined to obtain the DR image index.

[0042] It can be seen that in the embodiment of the present application, morphological analysis technology is used to perform morphological analysis on each target local image to determine the anatomical structure index. Then, feature fusion is performed based on the energy density feature, energy center of gravity feature and energy effective value feature in the energy feature to obtain an energy feature vector, and energy calculation is performed based on the energy feature vector to determine the energy feature index. The feature fusion operation comprehensively considers the correlation and complementarity between different energy features, and improves the accuracy and reliability of energy detection. Furthermore, based on the pathological recognition model, the texture features in each image feature are pathologically identified and quantified to determine the pathological quantification index. Finally, the anatomical structure index, energy feature index and pathological quantification index are combined to obtain the DR image index.

[0043] Furthermore, in order to obtain more comprehensive patient information and accurately evaluate the patient's DR image, in the embodiment of the present application, after performing DR image analysis based on the energy feature and the image feature corresponding to each target local image and obtaining the DR image index, the following is further included: Based on each indicator in the DR image indicators, multi-dimensional content expansion is performed to obtain the expansion information corresponding to the DR image indicators, wherein the multi-dimensional content expansion is used to expand the content from the indicator meaning, calculation method and normal range; Based on anatomical structure indicators, energy characteristic indicators, pathological quantitative indicators and extended information, data marking is performed in DR images to obtain comprehensive image interpretation.

[0044] For the embodiments of the present application, in order to facilitate doctors to more clearly understand the meaning represented by each DR image indicator and more accurately evaluate the patient's DR image, multi-dimensional content expansion is performed based on each indicator in the DR image indicators to obtain the expanded information corresponding to the DR image indicators. The dimensions of multi-dimensional content expansion include but are not limited to: indicator meaning, calculation method and normal range. For the meaning of the indicator, by explaining the meaning of each DR image indicator, it helps doctors better understand the meaning represented by each indicator after DR image analysis; for the calculation method, for each DR image indicator, its calculation method or measurement steps are given to enhance the credibility and calculation transparency of the DR image indicator; for the normal range, for each DR image indicator, its normal range or reference value is given, which helps doctors quickly determine whether the DR image indicator is within the normal range, so as to promptly discover abnormalities or potential problems.

[0045] Then, based on the anatomical structure indicators, energy characteristic indicators, pathological quantitative indicators and extended information, data is marked in the DR image to obtain a comprehensive image interpretation. For data marking, by marking the anatomical structure indicators, energy characteristic indicators, pathological quantitative indicators and extended information at the corresponding positions on the DR image, the relevant indicators of each part in the DR image can be clearly known, and more comprehensive patient information can be obtained, which helps doctors make more scientific medical decisions.

[0046] It can be seen that in the embodiment of the present application, in order to facilitate doctors to more clearly understand the meaning of each DR image indicator and more accurately evaluate the patient's DR image, multi-dimensional content expansion is performed based on each indicator in the DR image indicator to obtain the expanded information corresponding to the DR image indicator. Then, based on the anatomical structure indicators, energy characteristic indicators, pathological quantitative indicators and expanded information, data labeling is performed in the DR image to obtain a comprehensive image interpretation. Performing data labeling makes it possible to clearly know the relevant indicators of each part in the DR image, and to obtain more comprehensive patient information, thereby helping doctors make more scientific medical decisions.

[0047] Furthermore, in order to improve the accuracy of abnormal index detection and provide intuitive and accurate abnormal structure information, in the embodiment of the present application, after multi-dimensional content expansion is performed based on each indicator in the DR image indicator and the expanded information corresponding to the DR image indicator is obtained, it also includes: Based on the normal range of the index in the extended information, abnormal index detection is performed on the DR image index to determine the abnormal index, wherein the abnormal index is the index information of the DR image index that is not within the normal range of the index; The abnormal area is locked based on the abnormal index, the abnormal local image area and the corresponding abnormal image features are determined, and three-dimensional modeling is performed based on the abnormal image features to determine the three-dimensional model of the abnormal structure.

[0048] For the embodiment of the present application, the extended information gives the normal range of indicators corresponding to each DR image indicator. Therefore, based on the comparison between each specific indicator in the DR image indicator and the set normal range of indicators, if the indicator is not in the normal range of indicators, the indicator is determined to be an abnormal indicator; otherwise, the indicator is determined to be a normal indicator. Then, the abnormal area is locked based on the abnormal indicator to determine the abnormal local image area and the corresponding abnormal image features. Automated abnormal indicator detection and abnormal area locking not only improves the accuracy of abnormal indicator detection, but also helps doctors quickly locate the lesion.

[0049] Furthermore, the abnormal image features corresponding to the abnormal local image area represent the shape, size, density, texture and other characteristics of the abnormal part. Therefore, three-dimensional modeling is performed based on the abnormal image features to determine the three-dimensional model of the abnormal structure. The three-dimensional model of the abnormal structure provides intuitive and accurate abnormal structure information. The three-dimensional model of the abnormal structure can be used as an intuitive teaching tool to help doctors explain the condition, surgical plan and expected results to patients. Of course, it can also help doctors more accurately judge the nature, severity and possible causes of the abnormality.

[0050] It can be seen that in the embodiment of the present application, based on the normal range of the indicators in the extended information, the DR image indicators are detected for abnormal indicators to determine the abnormal indicators. Automated abnormal indicator detection and abnormal area locking not only improves the accuracy of abnormal indicator detection, but also helps doctors quickly locate the lesion. Then, based on the abnormal indicators, the abnormal area is locked to determine the abnormal local image area and the corresponding abnormal image features, and three-dimensional modeling is performed based on the abnormal image features to determine the abnormal structure three-dimensional model. The abnormal structure three-dimensional model provides intuitive and accurate abnormal structure information.

[0051] Furthermore, in order to improve the accuracy and stability of image segmentation and image analysis, and to facilitate capturing the details of complex texture features in DR images, in an embodiment of the present application, an adaptive threshold adjustment method is used to dynamically calculate the threshold of the target DR image, and the target thresholds corresponding to multiple local image areas are determined, including: The threshold calculation parameters of the adaptive threshold calculation are obtained, and based on the Gaussian weighting method and the threshold calculation parameters in the adaptive threshold adjustment method, the threshold dynamic calculation is performed on the target DR image to determine the target thresholds corresponding to the multiple local image regions.

[0052] For the embodiment of the present application, the threshold calculation parameter of the adaptive threshold calculation is obtained, and the threshold calculation parameter is used to specify the size of the local image area so that the target DR image can be divided into multiple small local areas, wherein the threshold calculation parameter can be pre-set and stored in the electronic device, and of course, it can also be determined according to the actual situation, that is, the user inputs the threshold calculation parameter value through the display interface of the electronic device. The local image area is divided based on the threshold calculation parameter, that is, the entire DR image is divided into multiple overlapping or non-overlapping local image areas. Then, the Gaussian weighting method in the adaptive threshold adjustment method is selected for dynamic threshold calculation. The Gaussian weighting method has the advantages of reasonable weight distribution, strong noise suppression ability, good smooth transition, high adaptability and flexibility, so that the calculated target threshold is more in line with the current situation of each local image area. In the process of dynamic threshold calculation: a Gaussian weighting function is applied to each local image area to calculate the weighted average value or weighted histogram of the area, and then, according to the weighted average value or weighted histogram, the target threshold of each local image area is dynamically calculated, that is, the optimal threshold is automatically determined according to the weighted data. The adaptive threshold adjustment method dynamically calculates the threshold according to the brightness distribution of different regions, thereby adapting to the differences between different DR images, thereby improving the accuracy and stability of image segmentation and image analysis, and facilitating the capture of details of complex texture features in DR images.

[0053] It can be seen that in the embodiment of the present application, the threshold calculation parameters of the adaptive threshold calculation are obtained, and based on the Gaussian weighting method and the threshold calculation parameters in the adaptive threshold adjustment method, the threshold is dynamically calculated for the target DR image to determine the target thresholds corresponding to the multiple local image regions. The adaptive threshold adjustment method dynamically calculates the threshold according to the brightness distribution of different regions, thereby adapting to the differences between different DR images, thereby improving the accuracy and stability of image segmentation and image analysis, and facilitating the capture of the details of complex texture features in the DR image.

[0054] The above-mentioned embodiment introduces a DR image analysis method from the perspective of method flow, and the following embodiment introduces a DR image analysis device from the perspective of a virtual module or a virtual unit. Please refer to the following embodiment for details.

[0055] The present application embodiment provides a DR image analysis device, such as Figure 2 As shown, the DR image analysis device may specifically include: A preprocessing module 210 is used to acquire a DR image and perform preprocessing based on the DR image to obtain a target DR image, wherein the preprocessing is used to improve the image quality of the DR image; A threshold dynamic calculation module 220 is used to perform threshold dynamic calculation on the target DR image using an adaptive threshold adjustment method to determine target thresholds corresponding to a plurality of local image regions, wherein the plurality of local image regions are obtained by segmenting the target DR image; A feature extraction module 230 is used to extract features of a target local image region using the target threshold value to determine image features corresponding to the target local image, wherein the target local image region is any local image region; The DR image analysis module 240 is used to perform DR image analysis based on the image features corresponding to each of the target local images to obtain DR image indicators.

[0056] For the embodiment of the present application, preprocessing is performed based on the DR image to obtain a target DR image, wherein the preprocessing is used to improve the image quality of the DR image. In order to improve the accuracy and stability of image segmentation and DR image analysis, and to facilitate capturing the details of complex texture features in the DR image, an adaptive threshold adjustment method is used to dynamically calculate the threshold of the target DR image, and the target thresholds corresponding to multiple local image regions are determined. The adaptive threshold adjustment method dynamically calculates the threshold according to the brightness distribution of different regions, thereby adapting to the differences between different DR images. Then, the target threshold is used to extract features from the target local image region to determine the image features corresponding to the target local image. Finally, DR image analysis is performed based on the image features corresponding to each target local image to obtain DR image indicators. The DR image indicators determined after performing DR image analysis provide quantitative image feature information, which helps doctors understand the DR image content more intuitively and enhances the interpretability of the image.

[0057] In a possible implementation of the embodiment of the present application, the DR image analysis device further includes: An energy feature extraction module, used for performing image conversion based on the target DR image to obtain an energy distribution map corresponding to the target DR image; Extracting features based on the energy distribution diagram to determine energy features, wherein the energy features include: energy density features, energy center of gravity features, and energy effective value features; Accordingly, when the DR image analysis module 240 performs DR image analysis based on the image features corresponding to each of the target partial images to obtain the DR image index, it is used to: DR image analysis is performed based on the energy feature and the image feature corresponding to each of the target local images to obtain a DR image index.

[0058] In a possible implementation of the embodiment of the present application, when the DR image analysis module 240 performs DR image analysis based on the energy feature and the image feature corresponding to each of the target local images to obtain the DR image index, it is used to: Using morphological analysis technology, perform morphological analysis on each of the target local images to determine anatomical structure indicators; Performing feature fusion based on the energy density feature, the energy center of gravity feature and the energy effective value feature in the energy feature to obtain an energy feature vector, and performing energy calculation based on the energy feature vector to determine an energy feature index; Acquire a pathology recognition model, perform pathology recognition and quantification on the texture features in each of the image features based on the pathology recognition model, and determine a pathology quantification index; The anatomical structure index, the energy characteristic index and the pathological quantitative index are integrated to obtain a DR image index.

[0059] In a possible implementation of the embodiment of the present application, the DR image analysis device further includes: A multi-dimensional content expansion module, used to perform multi-dimensional content expansion based on each indicator of the DR image indicators to obtain expansion information corresponding to the DR image indicators, wherein the multi-dimensional content expansion is used to expand the content from the indicator meaning, calculation method and normal range; Based on the anatomical structure index, the energy characteristic index, the pathological quantification index and the extended information, data marking is performed in the DR image to obtain a comprehensive image interpretation.

[0060] In a possible implementation of the embodiment of the present application, the DR image analysis device further includes: An abnormal index detection model, used to perform abnormal index detection on the DR image index based on the normal range of the index in the extended information, and determine the abnormal index, wherein the abnormal index is the index information of the DR image index that is not within the normal range of the index; Based on the abnormal index, the abnormal area is locked, the abnormal local image area and the corresponding abnormal image features are determined, and three-dimensional modeling is performed based on the abnormal image features to determine the abnormal structure three-dimensional model.

[0061] In a possible implementation of the embodiment of the present application, when the threshold dynamic calculation module 220 performs threshold dynamic calculation on the target DR image using the adaptive threshold adjustment method to determine the target thresholds corresponding to the multiple local image regions, the threshold dynamic calculation module 220 is used to: The threshold calculation parameters of the adaptive threshold calculation are obtained, and based on the Gaussian weighting method in the adaptive threshold adjustment method and the threshold calculation parameters, the threshold dynamic calculation is performed on the target DR image to determine the target thresholds corresponding to the multiple local image areas respectively.

[0062] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the DR image analysis device described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0063] An electronic device is provided in an embodiment of the present application, such as Figure 3 As shown, Figure 3 The electronic device 300 shown includes: a processor 301 and a memory 303. The processor 301 and the memory 303 are connected, such as through a bus 302. Optionally, the electronic device 300 may also include a transceiver 304. It should be noted that in actual applications, the transceiver 304 is not limited to one, and the structure of the electronic device 300 does not constitute a limitation on the embodiments of the present application.

[0064] The processor 301 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array) or other programmable logic devices, transistor logic devices, hardware components or any combination thereof. It may implement or execute various exemplary logic blocks, modules and circuits described in conjunction with the disclosure of this application. The processor 301 may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0065] The bus 302 may include a path to transmit information between the above components. The bus 302 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus. The bus 302 may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 3 Only one thick line is used in the diagram, but it does not mean that there is only one bus or only one type of bus.

[0066] The memory 303 may be a ROM (Read Only Memory) or other types of static storage devices that can store static information and instructions, a RAM (Random Access Memory) or other types of dynamic storage devices that can store information and instructions, or an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory) or other optical disk storage, optical disk storage (including compressed optical disk, laser disk, optical disk, digital versatile disk, Blu-ray disk, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto.

[0067] The memory 303 is used to store the application code for executing the solution of the present application, and the execution is controlled by the processor 301. The processor 301 is used to execute the application code stored in the memory 303 to implement the contents shown in the above method embodiment.

[0068] The electronic devices include, but are not limited to, mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), and fixed terminals such as digital TVs, desktop computers, etc. It can also be a server, etc. Figure 3 The electronic device shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0069] An embodiment of the present application provides a computer-readable storage medium, on which a computer program is stored. When the computer-readable storage medium is run on a computer, the computer can execute the corresponding content in the aforementioned method embodiment.

[0070] The embodiment of the present application provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, the method in any of the above embodiments is implemented. Compared with the related art, the embodiment of the present application performs preprocessing based on the DR image to obtain a target DR image, wherein the preprocessing is used to improve the image quality of the DR image. In order to improve the accuracy and stability of image segmentation and DR image analysis, and to facilitate capturing the details of complex texture features in the DR image, an adaptive threshold adjustment method is used to dynamically calculate the threshold of the target DR image, and the target thresholds corresponding to multiple local image regions are determined. The adaptive threshold adjustment method dynamically calculates the threshold according to the brightness distribution of different regions, thereby adapting to the differences between different DR images. Then, the target threshold is used to extract features of the target local image region to determine the image features corresponding to the target local image. Finally, DR image analysis is performed based on the image features corresponding to each target local image to obtain DR image indicators. The DR image indicators determined after performing DR image analysis provide quantitative image feature information, which helps doctors to understand the DR image content more intuitively and enhance the interpretability of the image.

[0071] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0072] The above description is only a partial implementation method of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A DR image analysis method, characterized in that: include: Acquire a DR image, and perform preprocessing based on the DR image to obtain a target DR image, wherein the preprocessing is used to improve the image quality of the DR image; Performing dynamic threshold calculation on the target DR image using an adaptive threshold adjustment method to determine target thresholds corresponding to a plurality of local image regions, wherein the plurality of local image regions are obtained by segmenting the target DR image; Using the target threshold to perform feature extraction on a target local image region, and determining image features corresponding to the target local image, wherein the target local image region is any local image region; DR image analysis is performed based on the image features corresponding to each of the target local images to obtain DR image indicators.

2. The DR image analysis method according to claim 1, characterized in that: After the preprocessing is performed based on the DR image to obtain the target DR image, the method further includes: Performing image conversion based on the target DR image to obtain an energy distribution map corresponding to the target DR image; Extracting features based on the energy distribution diagram to determine energy features, wherein the energy features include: energy density features, energy center of gravity features, and energy effective value features; Accordingly, DR image analysis is performed based on the image features corresponding to each of the target local images to obtain DR image indicators, including: DR image analysis is performed based on the energy feature and the image feature corresponding to each of the target local images to obtain a DR image index.

3. The DR image analysis method according to claim 2, characterized in that: The performing DR image analysis based on the energy feature and the image feature corresponding to each of the target local images to obtain the DR image index includes: Using morphological analysis technology, perform morphological analysis on each of the target local images to determine anatomical structure indicators; Performing feature fusion based on the energy density feature, the energy center of gravity feature and the energy effective value feature in the energy feature to obtain an energy feature vector, and performing energy calculation based on the energy feature vector to determine an energy feature index; Acquire a pathology recognition model, perform pathology recognition and quantification on the texture features in each of the image features based on the pathology recognition model, and determine a pathology quantification index; The anatomical structure index, the energy characteristic index and the pathological quantitative index are integrated to obtain a DR image index.

4. The DR image analysis method according to claim 3, characterized in that: After performing DR image analysis based on the energy feature and the image feature corresponding to each of the target local images to obtain the DR image index, the method further includes: Performing multi-dimensional content expansion based on each of the DR image indicators to obtain expansion information corresponding to the DR image indicators, wherein the multi-dimensional content expansion is used to expand the content from the indicator meaning, calculation method and normal range; Based on the anatomical structure index, the energy characteristic index, the pathological quantification index and the extended information, data marking is performed in the DR image to obtain a comprehensive image interpretation.

5. The DR image analysis method according to claim 4, characterized in that: After performing multi-dimensional content expansion based on each indicator of the DR image indicators to obtain expansion information corresponding to the DR image indicators, the method further includes: Based on the normal range of the indicator in the extended information, abnormal indicator detection is performed on the DR image indicator to determine the abnormal indicator, wherein the abnormal indicator is indicator information of the DR image indicator that is not within the normal range of the indicator; Based on the abnormal index, the abnormal area is locked, the abnormal local image area and the corresponding abnormal image features are determined, and three-dimensional modeling is performed based on the abnormal image features to determine the abnormal structure three-dimensional model.

6. The DR image analysis method according to claim 1, characterized in that: The method of dynamically calculating the threshold of the target DR image using the adaptive threshold adjustment method to determine the target thresholds corresponding to the multiple local image regions respectively includes: The threshold calculation parameters of the adaptive threshold calculation are obtained, and based on the Gaussian weighting method in the adaptive threshold adjustment method and the threshold calculation parameters, the threshold dynamic calculation is performed on the target DR image to determine the target thresholds corresponding to the multiple local image areas respectively.

7. An electronic device, characterized in that: include: at least one processor; Memory; At least one application, wherein the at least one application is stored in a memory and configured to be executed by at least one processor, and the at least one application is configured to: execute the DR image analysis method according to any one of claims 1 to 6.

8. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed in a computer, the computer is caused to execute the DR image analysis method according to any one of claims 1 to 6.

9. A computer program product, characterized in that The invention comprises a computer program, wherein the computer program is used by a processor to execute the DR image analysis method according to any one of claims 1 to 6.

Images