Image quality intelligent detection and classification method based on ultrasonic image

Abstract

The invention relates to the technical field of image recognition, in particular to an intelligent image quality detection and classification method based on an ultrasonic image. The method comprises the following steps: acquiring standard ultrasonic image data; performing ultrasonic image visual feature distribution analysis based on the standard ultrasonic image data to generate ultrasonic image visual feature distribution data; performing ultrasonic image effective information quantitative intelligent analysis based on the ultrasonic image visual feature distribution data to generate ultrasonic image effective information quantitative data; performing effective image screening on the standard ultrasonic image data according to the ultrasonic image effective information quantitative data to obtain effective ultrasonic image data; performing image quality intelligent analysis on the effective ultrasonic image data to generate quality identification ultrasonic image data; and performing ultrasonic image intelligent classification operation based on the quality identification ultrasonic image data. Intelligent ultrasonic image quality detection and classification are realized by analyzing the effective image information and the image blurring degree of the ultrasonic image.

Description

Technical Field

[0001] The present invention relates to the technical field of image recognition, and in particular to an intelligent detection and classification method of image quality based on ultrasonic images. Background Art

[0002] Ultrasound imaging technology is widely used in medical diagnosis. Its non-invasiveness and real-time nature make it an important tool for diagnosing and monitoring diseases. However, differences in image quality affect the accuracy of later diagnostic results. The quality of ultrasound images is affected by many factors, including equipment performance, operating skills, and the physiological state of the subject being tested. Ultrasound images of different quality may lead to differences in diagnostic accuracy and reliability. Especially in more complex cases, through automated detection and classification, low-quality images can be quickly identified and excluded, the reliability and consistency of later diagnosis can be improved, and the workflow can be optimized, human errors can be reduced, and the quality and efficiency of medical services can be effectively improved. However, the traditional image quality detection and classification method based on ultrasound images is to identify and classify ultrasound imaging quality through human subjectivity. There is no unified standard for ultrasound imaging quality detection and classification, and ultrasound imaging quality assessment faces problems in accuracy, efficiency, and generalization ability. Summary of the invention

[0003] Based on this, the present invention provides an intelligent detection and classification method for image quality based on ultrasound images to solve at least one of the above technical problems.

[0004] To achieve the above purpose, an intelligent detection and classification method of image quality based on ultrasound images includes the following steps:

[0005] Step S1: custom-adjusting the ultrasound device according to a preset ultrasound device multi-source adjustment parameter group to obtain an adjusted ultrasound device, and performing an ultrasound image acquisition operation on a patient through the adjusted ultrasound device; and performing ultrasound image acquisition based on the ultrasound image acquisition operation to obtain ultrasound image data;

[0006] Step S2: performing image standardization processing on the ultrasonic image data to generate standard ultrasonic image data; based on the standard ultrasonic image data; performing ultrasonic image visual feature distribution analysis to generate ultrasonic image visual feature distribution data;

[0007] Step S3: performing local grid division processing on the ultrasonic image visual feature distribution data to generate grid visual feature distribution data; performing ultrasonic image visual feature information entropy analysis based on the grid visual feature distribution data to generate ultrasonic image visual feature information entropy data; performing ultrasonic image effective information quantization intelligent analysis based on the ultrasonic image visual feature information entropy data to generate ultrasonic image effective information quantization data;

[0008] Step S4: dividing the standard ultrasonic image data into valid and abnormal ultrasonic image data according to the ultrasonic image effective information quantization data to obtain valid ultrasonic image data or abnormal ultrasonic image data; when abnormal ultrasonic image data is identified, transmitting the abnormal ultrasonic image data to the terminal to perform ultrasonic image acquisition abnormality feedback operation;

[0009] Step S5: extract ultrasonic image contour features based on the effective ultrasonic image data to generate ultrasonic image contour feature data; perform ultrasonic image contour blurriness analysis on the ultrasonic image contour feature data to generate ultrasonic image contour blurriness data; perform image quality intelligent identification processing on the effective ultrasonic image data based on the ultrasonic image contour blurriness data to generate quality-identified ultrasonic image data; and perform ultrasonic image intelligent classification based on the quality-identified ultrasonic image data.

[0010] The present invention customizes the ultrasonic device according to the preset multi-source adjustment parameter group, ensures that the images collected from different ultrasonic devices are different in parameter settings, so as to obtain ultrasonic images of different qualities. The ultrasonic device after adjustment is used to collect ultrasonic images of the patient, and ultrasonic image data with different acquisition parameters can be obtained. The ultrasonic device is precisely adjusted by using the preset ultrasonic device multi-source adjustment parameter group to obtain ultrasonic images of different qualities. These parameters involve ultrasonic frequency, gain, depth setting, etc., and provide diversified data samples for subsequent image quality evaluation, so as to more comprehensively analyze and compare the performance of images under different adjustment parameters. By performing image standardization processing on the collected ultrasonic image data, consistent standard ultrasonic image data is generated, so as to ensure that the image differences caused by image settings are eliminated, so that the ultrasonic image quality evaluation is carried out under the same conditions, and then the ultrasonic image visual feature distribution analysis is performed based on these standardized data, so as to deeply understand the spatial distribution of different visual features in the image, and provide accurate and reliable basic data for subsequent quality evaluation and intelligent classification. By performing local grid division processing on the ultrasonic image visual feature distribution data, grid visual feature distribution data is generated, which can effectively capture the changes and distribution of local features in the image, and provide a more detailed and accurate data basis for subsequent analysis. Subsequently, the information entropy of the ultrasonic image visual features is analyzed according to the grid visual feature distribution data to generate the information entropy data of the ultrasonic image visual features, quantify the complexity and randomness of the information in the image, so as to determine whether the image has useful valuable information or no available information, and help further understand and evaluate the visual feature quality of the ultrasonic image. According to the quantified data of the effective information of the ultrasonic image, the standard ultrasonic image data is divided into effective and abnormal, and the high-quality effective ultrasonic image data and the abnormal ultrasonic image data with problems are accurately distinguished, such as the abnormal ultrasonic image that is blank or only black and white without any effective ultrasonic information. In particular, when the abnormal ultrasonic image data is identified, the data can be transmitted to the terminal in time for abnormal feedback operation, so as to promote equipment adjustment or operation improvement, and improve the overall quality and accuracy of ultrasonic image acquisition. By extracting the contour features of the effective ultrasonic image data, the key shape and contour information are extracted from the image. These features are very important for identifying and describing the structure and morphology in the image, and provide key data support for subsequent analysis. The blur degree analysis of the extracted contour feature data can quantify the clarity or blur degree of the contour in the image, which helps to evaluate the visual quality of the image and determine whether the image has sufficient clarity and accuracy. According to the contour blur degree data, the valid ultrasound image data is intelligently labeled, that is, the image quality is marked with a corresponding quality label, and the image is automatically classified into high quality, medium quality or low quality, providing accurate image grading for medical diagnosis or research.Based on the ultrasound image data after quality identification, intelligent classification operations are performed, including classifying or sorting the images according to their quality levels, and associating them with the multi-source adjustment parameters of the ultrasound device to screen the multi-source adjustment parameters of the ultrasound device corresponding to high-quality ultrasound images.

[0011] Preferably, step S1 comprises the following steps:

[0012] Step S11: custom-adjusting the ultrasound device according to the preset ultrasound device multi-source adjustment parameter group to obtain an adjusted ultrasound device, and performing an ultrasound image acquisition operation on the patient through the adjusted ultrasound device;

[0013] Step S12: performing ultrasonic signal acquisition based on the ultrasonic image acquisition operation to obtain an ultrasonic signal;

[0014] Step S13: Performing ultrasonic image conversion based on the ultrasonic signal to obtain ultrasonic image data.

[0015] The present invention performs custom adjustments on the ultrasonic device according to the preset ultrasonic device multi-source adjustment parameter group to obtain an adjusted ultrasonic device, thereby collecting corresponding ultrasonic images under different adjustment parameters, and analyzing the corresponding relationship between the parameters of the ultrasonic device and the ultrasonic image quality in a personalized manner, providing diversified data samples for subsequent image quality evaluation, so as to more comprehensively analyze and compare the image performance under different adjustment parameters. Ultrasonic signals are collected from patients, and these signals are the basis for generating ultrasonic images. By collecting ultrasonic signals, raw data containing rich anatomical and tissue information is obtained, providing necessary input for subsequent image conversion and processing. The collected ultrasonic signals are converted into visual ultrasonic image data, and the process includes signal processing, imaging algorithms and data reconstruction, which can convert the original ultrasonic signals into medically usable two-dimensional or three-dimensional images, showing tissue structure and lesion information.

[0016] Preferably, step S13 comprises the following steps:

[0017] Step S131: performing spectrum data conversion according to the ultrasonic signal to generate ultrasonic spectrum data;

[0018] Step S132: performing ultrasonic pixel block analysis according to the ultrasonic spectrum data to obtain ultrasonic pixel block data;

[0019] Step S133: performing ultrasonic image conversion according to the ultrasonic pixel block data to obtain ultrasonic image data.

[0020] The present invention converts ultrasonic signals into spectrum data, effectively extracts the frequency components and energy distribution of the signals, and the spectrum data reflects the frequency domain characteristics of ultrasonic signals. Ultrasonic spectrum data can capture the frequency response characteristics of different tissues or structures, providing a rich information basis for subsequent analysis, and helping to more accurately reflect the visual situation in the image. Converting ultrasonic spectrum data into pixel block data helps to perform more detailed local feature analysis of the image in space. These pixel block data can capture the local structure and characteristics of different areas in the image, providing a more detailed analysis basis. By analyzing pixel block data instead of the entire image, the complexity of processing and calculation is effectively reduced while ensuring data accuracy, and the processing efficiency and speed are improved, especially in the case of large data volumes. The processed ultrasonic pixel block data is converted into the final ultrasonic image data, including image reconstruction processing, so that the ultrasonic signal is converted into the corresponding ultrasonic image, which is used to analyze the quality and classification status of the ultrasonic image.

[0021] Preferably, step S2 comprises the following steps:

[0022] Step S21: performing image standardization processing on the ultrasonic image data to generate standard ultrasonic image data;

[0023] Step S22: performing grayscale value distribution analysis on the standard ultrasound image data to generate image grayscale value distribution data;

[0024] Step S23: performing image texture distribution analysis on the standard ultrasound image data to generate image texture distribution data;

[0025] Step S24: performing ultrasound image visual feature distribution analysis based on the image gray value distribution data and the image texture distribution data to generate ultrasound image visual feature distribution data.

[0026] The present invention ensures that the ultrasound image data collected from different sources and conditions have a consistent data format and range through standardized processing, reduces data deviations caused by equipment differences or operational changes, and improves the comparability and credibility of the data. Standardized ultrasound image data is easier to perform subsequent analysis and processing, such as feature extraction or model training, because the basic characteristics and statistical properties of the data have been adjusted and standardized. Through gray value distribution analysis, the distribution characteristics of gray levels in ultrasound images are quantitatively extracted, which not only reflects the contrast and brightness distribution of the image, but also can reveal the gray feature differences of different tissue structures or lesion areas. The gray value distribution data provides the internal characteristics of different tissue types. Texture distribution analysis reveals the local structure and texture features in ultrasound images, such as spots, roughness, and granularity, which helps to distinguish different tissue types and lesion states, and provides a more detailed description of image features. Texture distribution data provides doctors with additional image information. Combining grayscale value and texture distribution data, a comprehensive analysis of the visual features of ultrasound images is performed, which not only integrates the brightness and contrast information of the image, but also includes the subtle structure and texture features in the image, providing a detailed visual feature description for subsequent image quality assessment and lesion analysis. The visual feature distribution data of ultrasound images provides a deeper and more comprehensive basis for image analysis, and helps to provide effective feature information in ultrasound images.

[0027] Preferably, step S23 includes the following steps:

[0028] Step S231: performing grayscale value gradient calculation on the image grayscale value distribution data to generate image grayscale value gradient distribution data;

[0029] Step S232: performing grayscale value gradient feature division processing on the image grayscale value gradient distribution data according to the preset grayscale value gradient difference data to generate grayscale value gradient feature division data;

[0030] Step S233: performing image texture distribution analysis on the standard ultrasound image data according to the gray value gradient feature division data to generate image texture distribution data.

[0031] The present invention extracts more detailed grayscale change information from the ultrasound image by calculating the grayscale value gradient. These data reflect the grayscale transition and change rate of different regions in the image, which helps to describe the edges and subtle structures in the image. Grayscale value gradient distribution data can accurately reflect the contrast changes in the image, thereby enhancing the visual differences and distinctions between different regions. Grayscale value gradient feature division classifies and divides the grayscale value gradient data according to the preset gradient difference, thereby quantifying the characteristics of grayscale changes in the image, effectively reducing the complexity of the data, making subsequent analysis more efficient and controllable, and the divided feature data makes the grayscale gradient features of different images comparable and consistent, which helps to accurately compare and analyze between different samples. Texture distribution analysis based on grayscale value gradient feature division data can describe the texture features in the ultrasound image in detail, reflect the structure, roughness and texture laws of different regions in the image, and provide a basis for the detailed feature extraction of the image. Image texture distribution data provides more information about the surface and structural characteristics of the image, which helps further visual analysis and feature extraction. These data can be used in applications such as image classification and target detection to improve the comprehensiveness and accuracy of image analysis.

[0032] Preferably, step S3 comprises the following steps:

[0033] Step S31: performing local grid division processing on the ultrasound image visual feature distribution data to generate grid visual feature distribution data;

[0034] Step S32: performing grid visual feature correlation analysis according to the grid visual feature distribution data to generate grid visual feature correlation data;

[0035] Step S33: using a visual feature information entropy evaluation algorithm to calculate the ultrasonic image visual feature information entropy on the grid visual feature correlation data to generate ultrasonic image visual feature information entropy data;

[0036] Step S34: Performing intelligent analysis of effective information quantification of the ultrasonic image based on the ultrasonic image visual feature information entropy data to generate effective information quantification data of the ultrasonic image.

[0037] The present invention performs local grid division processing on the ultrasonic image visual feature distribution data. The grid division can effectively integrate and divide the ultrasonic image visual feature data in space, so that the feature data in each grid unit is more concentrated and controllable. The analysis of the grid visual feature distribution data helps to extract more refined local feature information, which reflects the visual characteristics and structural differences of different regions of the image. The grid visual feature correlation analysis reveals the degree of association between visual features in different grid units, which helps to understand the interaction and dependency between local regions in the image. Through the correlation data, the relationship between the features of different parts in the image is visualized and quantified, which provides a reliable basis for subsequent image analysis and understanding. The ultrasonic image visual feature information entropy data quantifies the complexity and information richness of the visual features in the image, and provides a measure of the diversity and complexity of the image content. The information entropy calculation can reflect the diversity and distribution balance of the visual features in different grid units, which helps to evaluate the overall visual expression and information content of the image. The ultrasonic image effective information quantification data can more accurately quantify the effective information content in the image by quantitatively analyzing the information entropy data, standardize the information processing flow in the image evaluation process, and make different ultrasonic image data comparable and consistent in terms of information quantification.

[0038] Preferably, step S32 includes the following steps:

[0039] Step S321: performing cluster analysis on the grid visual feature distribution data according to a clustering algorithm to generate clustered grid visual feature distribution data;

[0040] Step S322: performing grid visual feature similarity evaluation processing on the clustered grid visual feature distribution data to generate grid visual feature similarity data;

[0041] Step S323: performing grid visual feature correlation analysis based on the grid visual feature similarity data to generate grid visual feature correlation data.

[0042] The present invention performs cluster analysis on grid visual feature distribution data according to a clustering algorithm. Cluster analysis combines similar features in grid visual feature distribution data into clusters, thereby effectively simplifying complex data structures into several representative categories, which is convenient for subsequent data understanding and processing. Through cluster analysis, patterns and categories hidden in grid visual feature data are discovered, helping to identify different visual feature categories and their distribution in the image. Grid visual feature similarity data quantifies the similarity and difference between visual features in different grid units, and provides a specific measure of the correlation between features in different regions of the image. Similarity data reveals the degree of correlation between visual features in different regions of the image, which helps to understand the image structure and the distribution law of visual information. Grid visual feature correlation analysis deeply explores the correlation between visual features in different grid units, helps to understand the mutual influence and synergy of each grid image in the image, and shows the degree of correlation between features in different parts of the image, providing reliable data support for further image analysis and processing.

[0043] Preferably, the visual feature information entropy evaluation algorithm in step S33 is as follows:

[0044]

[0045] In the formula, H represents the information entropy data of the visual features of the ultrasound image, N represents the number of visual feature types, and α i Expressed as the weight coefficient of the i-th visual feature type, β i Denoted as the scaling parameter of the i-th visual feature type, γ i It is represented as the i-th visual feature correlation influence parameter, and e is represented as a natural constant, ∈ i is represented as the exponential scaling parameter of the i-th visual feature type, M is represented as the number of subgrids of the standard ultrasound image data, and θ ij It is expressed as the weight coefficient of the i-th visual feature type of the j-th sub-grid, η ij Represented as the grid visual feature correlation data corresponding to the i-th visual feature type of the j-th sub-grid.

[0046] The present invention utilizes a visual feature information entropy evaluation algorithm to evaluate the effective information content in standard ultrasound image data through grid visual feature correlation data, intelligently quantify the image information value reflected by the ultrasound image, and output the result ultrasound image visual feature information entropy data reflecting the visual feature information content of the standard ultrasound image. The part reflects the importance of various visual feature types for the evaluation of the effective information content of the overall ultrasound image, that is, it reflects the effective information content of the overall ultrasound image, among which the i-th visual feature correlation influence parameter γ iIt reflects the influence of the shape, spatial distribution or scale parameters of different features on the correlation of visual features of ultrasound images. It is used to index the i-th visual feature type, allowing the algorithm to take into account the nonlinear relationship between different features when evaluating the complexity and information content of the features. The information entropy contribution of the i-th visual feature type of the j-th subgrid is quantified, and processed by a logarithmic function to effectively process a large amount of data, while minimizing the contribution of larger values, thereby enhancing the algorithm's sensitivity to data changes. The above algorithm is used to calculate the effective information content of the image specifically reflected in the overall ultrasound image.

[0047] Preferably, step S4 comprises the following steps:

[0048] When the ultrasonic image effective information quantization data corresponding to the standard ultrasonic image data is greater than a preset ultrasonic image effective information quantization threshold, the standard ultrasonic image data is identified as effective ultrasonic image data; or, when the ultrasonic image effective information quantization data corresponding to the standard ultrasonic image data is not greater than the preset ultrasonic image effective information quantization threshold, the standard ultrasonic image data is identified as abnormal ultrasonic image data;

[0049] When abnormal ultrasonic image data is identified, the abnormal ultrasonic image data is transmitted to the terminal to perform an ultrasonic image acquisition abnormality feedback operation.

[0050] The present invention can accurately classify standard ultrasonic image data into two categories: effective and abnormal, based on the quantified data of effective information of ultrasonic images and preset thresholds. This classification method is based on the objective analysis of quantitative data, avoiding the uncertainty and deviation of traditional subjective evaluation methods. Through the automated discrimination process, real-time classification of ultrasonic image data is achieved, which improves the efficiency and accuracy of data processing, and is particularly important for the rapid processing and management of large amounts of ultrasonic image data. After identifying abnormal ultrasonic image data, these data can be quickly transmitted to the terminal to perform abnormal feedback operations for ultrasonic image acquisition. This real-time feedback mechanism helps to adjust ultrasonic equipment or acquisition conditions in a timely manner to improve image quality. The abnormal data collected through the feedback mechanism is deeply analyzed and evaluated, providing valuable reference information for equipment adjustment, formulation of operating specifications, and technical training in the process of ultrasonic image acquisition, thereby continuously improving and optimizing ultrasonic image quality.

[0051] Preferably, step S5 comprises the following steps:

[0052] Step S51: extracting ultrasonic image contour features according to the effective ultrasonic image data to generate ultrasonic image contour feature data;

[0053] Step S52: performing contour feature discretization processing on the contour feature data of the ultrasound image to generate contour feature discretization data;

[0054] Step S53: performing contour feature key node analysis according to the contour feature discretization data to generate contour feature key node data;

[0055] Step S54: performing key node descriptor analysis on the key node data of the contour feature to generate key node descriptor data;

[0056] Step S55 performs key node fuzzy diffusion feature analysis on the key node data of the contour feature according to the key node descriptor data to generate key node fuzzy diffusion feature data;

[0057] Step S56: performing ultrasound image contour fuzziness analysis on the ultrasound image contour feature data according to the key node fuzzy diffusion feature data to generate ultrasound image contour fuzziness degree data;

[0058] Step S57: performing image quality intelligent identification processing on the effective ultrasonic image data according to the ultrasonic image contour blur degree data to generate quality-identified ultrasonic image data;

[0059] Step S58: Performing an ultrasonic image intelligent classification task based on the quality-marked ultrasonic image data.

[0060] The present invention can accurately describe the shape and boundary information of organs and structures in the image by extracting the contour features of the ultrasonic image. In the process of contour feature extraction, the edge information, central node, deformation node, curvature node and other features in the contour information of the ultrasonic image are extracted, which is helpful for subsequent quantitative analysis and comparison. Discretization of contour feature data and conversion of edge feature information into discrete nodes are helpful to reduce data complexity while retaining important feature information, making subsequent analysis more efficient and reliable. Discretization can convert contour features into a form that is easy to compare and process, providing a clearer data basis for subsequent feature analysis and pattern recognition. By analyzing the key nodes of contour features, important structural features and key points in the ultrasonic image, such as deformation nodes, curvature nodes and other key nodes, are identified and extracted, which is helpful to further understand the structural composition of the image. Key node analysis helps focus on key structural features in the image, providing important clues for subsequent feature description and pattern recognition. Descriptor analysis is performed on key node data to more finely describe and quantify the features and attributes of each key node in the ultrasound image. The generated descriptor data enables a deeper comparison and analysis of the key nodes, providing more detailed and accurate information for the feature analysis and pattern recognition of the image. By performing fuzzy diffusion feature analysis on key node descriptor data, the expressive ability of key node features can be expanded and enhanced, making the structural features in the image more comprehensive and diverse. Fuzzy diffusion feature analysis converts the original key node descriptor data into richer and more complex feature data, thereby improving the representation ability of image information. Ultrasound image contour blur degree analysis is performed based on key node fuzzy diffusion feature data to quantify the blur degree of the image contour, that is, the clarity and accuracy of the image boundary. The ultrasound image contour blur degree data provides an objective quantitative basis for subsequent image quality assessment and helps to judge the visual clarity and information richness of the image. Intelligent identification processing is performed based on the ultrasound image contour blur degree data to generate quality identification ultrasound image data to indicate the quality level or category of the image. The quality identification ultrasound image data can provide an important reference for subsequent automated processing and decision-making, and support intelligent classification and further analysis of the image. Intelligent classification of ultrasound images based on quality-labeled ultrasound image data can improve the precision and accuracy of classification, thereby effectively distinguishing ultrasound images of different quality levels. Intelligent classification can classify images in an automated manner, optimize the processing flow, and improve work efficiency and quality control levels.

[0061] The beneficial effect of the present application is that the method for intelligent detection and classification of image quality based on ultrasound images of the present invention objectively evaluates the quality of ultrasound images, eliminates the uncertainty and errors of human subjective identification in traditional methods, ensures the objectivity and consistency of quality evaluation, and unifies the standards for image quality detection and classification of quantitative ultrasound images, so as to screen out high-quality ultrasound images of ultrasound equipment adjusted under different parameters. In combination with artificial intelligence technology, an intelligent algorithm model is developed, which can learn and identify key features of ultrasound images from a large amount of data, improve the accuracy and generalization ability of image quality evaluation, and effectively solve the problems of accuracy, efficiency and generalization ability of ultrasound imaging quality evaluation. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 A schematic diagram of the steps of an intelligent detection and classification method for image quality based on ultrasound images according to the present invention;

[0063] Figure 2 for Figure 1 Detailed implementation steps of step S3 in FIG.

[0064] Figure 3 for Figure 1 Detailed implementation steps of step S5;

[0065] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0066] The technical method of the present invention is described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.

[0067] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.

[0068] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, a first unit may be referred to as a second unit, and a similar second unit may be referred to as a first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0069] To achieve this, please refer to Figures 1 to 3 The present invention provides an intelligent detection and classification method for image quality based on ultrasound images, comprising the following steps:

[0070] Step S1: custom-adjusting the ultrasound device according to a preset ultrasound device multi-source adjustment parameter group to obtain an adjusted ultrasound device, and performing an ultrasound image acquisition operation on a patient through the adjusted ultrasound device; and performing ultrasound image acquisition based on the ultrasound image acquisition operation to obtain ultrasound image data;

[0071] Step S2: performing image standardization processing on the ultrasonic image data to generate standard ultrasonic image data; based on the standard ultrasonic image data; performing ultrasonic image visual feature distribution analysis to generate ultrasonic image visual feature distribution data;

[0072] Step S3: performing local grid division processing on the ultrasonic image visual feature distribution data to generate grid visual feature distribution data; performing ultrasonic image visual feature information entropy analysis based on the grid visual feature distribution data to generate ultrasonic image visual feature information entropy data; performing ultrasonic image effective information quantization intelligent analysis based on the ultrasonic image visual feature information entropy data to generate ultrasonic image effective information quantization data;

[0073] Step S4: dividing the standard ultrasonic image data into valid and abnormal ultrasonic image data according to the ultrasonic image effective information quantization data to obtain valid ultrasonic image data or abnormal ultrasonic image data; when abnormal ultrasonic image data is identified, transmitting the abnormal ultrasonic image data to the terminal to perform ultrasonic image acquisition abnormality feedback operation;

[0074] Step S5: extract ultrasonic image contour features based on the effective ultrasonic image data to generate ultrasonic image contour feature data; perform ultrasonic image contour blurriness analysis on the ultrasonic image contour feature data to generate ultrasonic image contour blurriness data; perform image quality intelligent identification processing on the effective ultrasonic image data based on the ultrasonic image contour blurriness data to generate quality-identified ultrasonic image data; and perform ultrasonic image intelligent classification based on the quality-identified ultrasonic image data.

[0075] The present invention customizes the ultrasonic device according to the preset multi-source adjustment parameter group, ensures that the images collected from different ultrasonic devices are different in parameter settings, so as to obtain ultrasonic images of different qualities. The ultrasonic device after adjustment is used to collect ultrasonic images of the patient, and ultrasonic image data with different acquisition parameters can be obtained. The ultrasonic device is precisely adjusted by using the preset ultrasonic device multi-source adjustment parameter group to obtain ultrasonic images of different qualities. These parameters involve ultrasonic frequency, gain, depth setting, etc., and provide diversified data samples for subsequent image quality evaluation, so as to more comprehensively analyze and compare the performance of images under different adjustment parameters. By performing image standardization processing on the collected ultrasonic image data, consistent standard ultrasonic image data is generated, so as to ensure that the image differences caused by image settings are eliminated, so that the ultrasonic image quality evaluation is carried out under the same conditions, and then the ultrasonic image visual feature distribution analysis is performed based on these standardized data, so as to deeply understand the spatial distribution of different visual features in the image, and provide accurate and reliable basic data for subsequent quality evaluation and intelligent classification. By performing local grid division processing on the ultrasonic image visual feature distribution data, grid visual feature distribution data is generated, which can effectively capture the changes and distribution of local features in the image, and provide a more detailed and accurate data basis for subsequent analysis. Subsequently, the information entropy of the ultrasonic image visual features is analyzed according to the grid visual feature distribution data to generate the information entropy data of the ultrasonic image visual features, quantify the complexity and randomness of the information in the image, so as to determine whether the image has useful valuable information or no available information, and help further understand and evaluate the visual feature quality of the ultrasonic image. According to the quantified data of the effective information of the ultrasonic image, the standard ultrasonic image data is divided into effective and abnormal, and the high-quality effective ultrasonic image data and the abnormal ultrasonic image data with problems are accurately distinguished, such as the abnormal ultrasonic image that is blank or only black and white without any effective ultrasonic information. In particular, when the abnormal ultrasonic image data is identified, the data can be transmitted to the terminal in time for abnormal feedback operation, so as to promote equipment adjustment or operation improvement, and improve the overall quality and accuracy of ultrasonic image acquisition. By extracting the contour features of the effective ultrasonic image data, the key shape and contour information are extracted from the image. These features are very important for identifying and describing the structure and morphology in the image, and provide key data support for subsequent analysis. The blur degree analysis of the extracted contour feature data can quantify the clarity or blur degree of the contour in the image, which helps to evaluate the visual quality of the image and determine whether the image has sufficient clarity and accuracy. According to the contour blur degree data, the valid ultrasound image data is intelligently labeled, that is, the image quality is marked with a corresponding quality label, and the image is automatically classified into high quality, medium quality or low quality, providing accurate image grading for medical diagnosis or research.Based on the ultrasound image data after quality identification, intelligent classification operations are performed, including classifying or sorting the images according to their quality levels, and associating them with the multi-source adjustment parameters of the ultrasound device to screen the multi-source adjustment parameters of the ultrasound device corresponding to high-quality ultrasound images.

[0076] In the embodiment of the present invention, reference Figure 1 The above is a schematic diagram of the steps of an intelligent detection and classification method for image quality based on ultrasound images of the present invention. In this embodiment, the intelligent detection and classification method for image quality based on ultrasound images includes the following steps:

[0077] Step S1: custom-adjusting the ultrasound device according to a preset ultrasound device multi-source adjustment parameter group to obtain an adjusted ultrasound device, and performing an ultrasound image acquisition operation on a patient through the adjusted ultrasound device; and performing ultrasound image acquisition based on the ultrasound image acquisition operation to obtain ultrasound image data;

[0078] In an embodiment of the present invention, settings and adjustments are performed based on a plurality of pre-set groups of ultrasound device adjustment parameter groups, including but not limited to ultrasonic frequency, gain, depth, sensitivity and other parameters. The selection of these parameters is usually based on specific clinical needs and the operator's experience. To customize the ultrasound device according to the preset ultrasound device multi-source adjustment parameter group, it is necessary to enter the setting interface of the ultrasound device and manually input or select the preset parameter value to adjust the working mode and output of the device. Use the adjusted ultrasound device to perform ultrasound image acquisition operations on the patient, place the ultrasound probe in a specific position, and adjust the device to obtain digital information collected by the ultrasound device. During the acquisition operation, the ultrasound device will emit ultrasound waves and receive echoes, convert the echoes into electrical signals, and further process them into visual ultrasound image data.

[0079] Step S2: performing image standardization processing on the ultrasonic image data to generate standard ultrasonic image data; based on the standard ultrasonic image data; performing ultrasonic image visual feature distribution analysis to generate ultrasonic image visual feature distribution data;

[0080] In an embodiment of the present invention, the obtained ultrasound image data is preprocessed to ensure that the image data meets the standardization requirements in terms of quality and format, including unified noise reduction processing, adjusting the contrast and brightness of the image, and converting the image to a standard format and size. The standardized ultrasound image data is marked as standard ultrasound image data for subsequent analysis and processing. Visual feature distribution analysis is performed based on the standard ultrasound image data, and the visual features (such as texture, color, shape, etc.) in the image are quantitatively analyzed through image processing technology, and the distribution of the corresponding visual features is analyzed to generate ultrasound image visual feature distribution data.

[0081] Step S3: performing local grid division processing on the ultrasonic image visual feature distribution data to generate grid visual feature distribution data; performing ultrasonic image visual feature information entropy analysis based on the grid visual feature distribution data to generate ultrasonic image visual feature information entropy data; performing ultrasonic image effective information quantization intelligent analysis based on the ultrasonic image visual feature information entropy data to generate ultrasonic image effective information quantization data;

[0082] In an embodiment of the present invention, the ultrasound image visual feature distribution data is divided into small areas, each area is called a grid, and the grid is defined according to a preset size and shape. For example, the image is divided into 10x10 grids, such as a local gridding process for the visual feature data of an ultrasound image, and the entire image is divided into 20x20 grids, each grid size is 1 / 400 of the image. The visual feature data in each grid is statistically analyzed to generate the visual feature distribution data of each grid, and the statistical features such as the mean, variance, and frequency distribution of each grid are calculated to form a grid-level feature description. The grayscale value of the ultrasound image in each 20x20 grid is statistically analyzed to generate feature data such as the average grayscale value and grayscale value range of each grid. Based on the generated grid visual feature distribution data, the visual feature information entropy of each grid is calculated. The information entropy can reflect the complexity and diversity of the visual features in each grid for further analysis and quantification. The information entropy of the ultrasound image visual features of each grid is calculated, and mathematical methods such as entropy calculation are used to quantify the information entropy of the ultrasound image visual features within the grid, which is used to evaluate the uncertainty and complexity of the gray value distribution within the grid, and generate ultrasound image visual feature information entropy data, which reflects the changes and distribution characteristics of the visual features in different areas of the ultrasound image, thereby quantifying the effective information in each ultrasound image. The generated ultrasound image visual feature information entropy data is used to perform intelligent analysis of the effective information quantification of the ultrasound image, such as setting a threshold or using a machine learning model to intelligently quantify the effective information of each grid, such as the complexity of the texture or the degree of change of the gray value, so as to distinguish whether the ultrasound image can provide effective information such as anatomical information.

[0083] Step S4: dividing the standard ultrasonic image data into valid and abnormal ultrasonic image data according to the ultrasonic image effective information quantization data to obtain valid ultrasonic image data or abnormal ultrasonic image data; when abnormal ultrasonic image data is identified, transmitting the abnormal ultrasonic image data to the terminal to perform ultrasonic image acquisition abnormality feedback operation;

[0084] In an embodiment of the present invention, a preset algorithm or model is used to calculate and analyze the effective information quantization data of the ultrasound image for standard ultrasound image data, such as quantifying the effective information amount of the image based on statistical features, machine learning models or other data analysis methods. According to the calculation results of the effective information quantization data of the ultrasound image, a preset effective information quantization threshold of the ultrasound image is set, and images exceeding the threshold are classified as effective ultrasound image data, and those not reaching the threshold are classified as abnormal ultrasound image data, thereby screening out ultrasound images that can display anatomical features and marking them as effective ultrasound image data, and marking ultrasound images that cannot display anatomical features as abnormal ultrasound image data. In the process of classification, images whose effective information quantization data of the ultrasound image does not reach the preset threshold are identified, that is, abnormal ultrasound image data. Once the abnormal ultrasound image data is identified, the system automatically transmits the abnormal data to the terminal, involving data transmission protocols, network connections and data security measures, to ensure that the abnormal ultrasound image data can be transmitted to the terminal device in a timely and secure manner, and the terminal device uniformly organizes the abnormal ultrasound image data and the corresponding ultrasound equipment acquisition parameters, and automatically notifies the operator or professional to conduct further inspection or adjust the ultrasound equipment parameters.

[0085] Step S5: extract ultrasonic image contour features based on the effective ultrasonic image data to generate ultrasonic image contour feature data; perform ultrasonic image contour blurriness analysis on the ultrasonic image contour feature data to generate ultrasonic image contour blurriness data; perform image quality intelligent identification processing on the effective ultrasonic image data based on the ultrasonic image contour blurriness data to generate quality-identified ultrasonic image data; and perform ultrasonic image intelligent classification based on the quality-identified ultrasonic image data.

[0086] In an embodiment of the present invention, image processing technology (such as edge detection algorithm) is used to extract contour features from valid ultrasound image data, identify and extract edge or contour information in the image, usually through grayscale gradient or edge detection method. The extracted ultrasound image contour feature data is further analyzed to evaluate the clarity and blurriness of the image contour, and the blurriness of the image contour is quantified by measuring the sharpness, continuity or other image features of the edge. The effective ultrasound image data is intelligently marked for image quality, including setting a threshold or standard for quality marking, and automatically marking high-quality or low-quality ultrasound image data according to the contour blurriness data. Based on the quality-marked ultrasound image data, an intelligent classification operation of the ultrasound image is performed, which involves using a machine learning algorithm or a preset classification model to classify the ultrasound image data into images of different quality levels or types.

[0087] Preferably, step S1 comprises the following steps:

[0088] Step S11: custom-adjusting the ultrasound device according to the preset ultrasound device multi-source adjustment parameter group to obtain an adjusted ultrasound device, and performing an ultrasound image acquisition operation on the patient through the adjusted ultrasound device;

[0089] Step S12: performing ultrasonic signal acquisition based on the ultrasonic image acquisition operation to obtain an ultrasonic signal;

[0090] Step S13: Performing ultrasonic image conversion based on the ultrasonic signal to obtain ultrasonic image data.

[0091] The present invention performs custom adjustments on the ultrasonic device according to the preset ultrasonic device multi-source adjustment parameter group to obtain an adjusted ultrasonic device, thereby collecting corresponding ultrasonic images under different adjustment parameters, and analyzing the corresponding relationship between the parameters of the ultrasonic device and the ultrasonic image quality in a personalized manner, providing diversified data samples for subsequent image quality evaluation, so as to more comprehensively analyze and compare the image performance under different adjustment parameters. Ultrasonic signals are collected from patients, and these signals are the basis for generating ultrasonic images. By collecting ultrasonic signals, raw data containing rich anatomical and tissue information is obtained, providing necessary input for subsequent image conversion and processing. The collected ultrasonic signals are converted into visual ultrasonic image data, and the process includes signal processing, imaging algorithms and data reconstruction, which can convert the original ultrasonic signals into medically usable two-dimensional or three-dimensional images, showing tissue structure and lesion information.

[0092] In an embodiment of the present invention, according to a preset ultrasonic device multi-source adjustment parameter group, the ultrasonic device multi-source adjustment parameter group is a plurality of groups of different ultrasonic multi-element adjustment parameters, and the ultrasonic device is customized and adjusted, including adjusting the parameters such as the frequency, depth, gain, and filter setting of the ultrasonic device to optimize the acquisition quality of the ultrasonic image, for example, using the adjustment interface provided by the ultrasonic device manufacturer, according to the preset ultrasonic device multi-source adjustment parameter group, adjusting the probe frequency, depth setting, gain setting, and filter setting of the plurality of groups of ultrasonic devices. Based on the ultrasonic image acquisition operation, the adjusted ultrasonic device is used to perform the acquisition of ultrasonic signals, and the ultrasonic signal is transmitted through the probe and the echo signal is received, which is used for subsequent image generation, for example, the adjusted ultrasonic device probe is placed at a specific position of the patient's body, the ultrasonic wave is transmitted and the echo is received, and the returned signal is recorded by the ultrasonic signal collector to obtain the ultrasonic signal. Based on the collected ultrasonic signal, the ultrasonic image is converted and processed by using digital signal processing, ultrasonic image generation algorithm, filtering and enhancement technology, etc., to generate visualized ultrasonic image data, for example, the collected ultrasonic signal is converted into a grayscale or color ultrasonic image by using a digital signal processing algorithm, and filtering and enhancement technology is applied to improve the image quality.

[0093] Preferably, step S13 comprises the following steps:

[0094] Step S131: performing spectrum data conversion according to the ultrasonic signal to generate ultrasonic spectrum data;

[0095] Step S132: performing ultrasonic pixel block analysis according to the ultrasonic spectrum data to obtain ultrasonic pixel block data;

[0096] Step S133: performing ultrasonic image conversion according to the ultrasonic pixel block data to obtain ultrasonic image data.

[0097] The present invention converts ultrasonic signals into spectrum data, effectively extracts the frequency components and energy distribution of the signals, and the spectrum data reflects the frequency domain characteristics of ultrasonic signals. Ultrasonic spectrum data can capture the frequency response characteristics of different tissues or structures, providing a rich information basis for subsequent analysis, and helping to more accurately reflect the visual situation in the image. Converting ultrasonic spectrum data into pixel block data helps to perform more detailed local feature analysis of the image in space. These pixel block data can capture the local structure and characteristics of different areas in the image, providing a more detailed analysis basis. By analyzing pixel block data instead of the entire image, the complexity of processing and calculation is effectively reduced while ensuring data accuracy, and the processing efficiency and speed are improved, especially in the case of large data volumes. The processed ultrasonic pixel block data is converted into the final ultrasonic image data, including image reconstruction processing, so that the ultrasonic signal is converted into the corresponding ultrasonic image, which is used to analyze the quality and classification status of the ultrasonic image.

[0098] In an embodiment of the present invention, raw data is obtained from an ultrasonic signal. These data are usually waveform data in the time domain, representing the changes of the ultrasonic signal in time. The ultrasonic signal is converted into ultrasonic spectrum data using Fourier transform or similar frequency domain conversion technology, reflecting the composition and intensity distribution of the ultrasonic signal in frequency. Ultrasonic pixel blocks are defined based on ultrasonic spectrum data, and information such as the intensity of a specific frequency range or frequency band in the spectrum is converted into grayscale value information of the ultrasonic pixel block, such as selecting a spectrum region of interest or extracting pixel blocks with specific frequency characteristics through filtering technology. The data obtained from the analysis of the ultrasonic pixel block is converted into ultrasonic image data, which involves remapping the spectrum data into a two-dimensional or three-dimensional space to generate an ultrasonic image, including reflecting the distribution and intensity of the pixel block in space, thereby forming the final ultrasonic image data.

[0099] Preferably, step S2 comprises the following steps:

[0100] Step S21: performing image standardization processing on the ultrasonic image data to generate standard ultrasonic image data;

[0101] Step S22: performing grayscale value distribution analysis on the standard ultrasound image data to generate image grayscale value distribution data;

[0102] Step S23: performing image texture distribution analysis on the standard ultrasound image data to generate image texture distribution data;

[0103] Step S24: performing ultrasound image visual feature distribution analysis based on the image gray value distribution data and the image texture distribution data to generate ultrasound image visual feature distribution data.

[0104] The present invention ensures that the ultrasound image data collected from different sources and conditions have a consistent data format and range through standardized processing, reduces data deviations caused by equipment differences or operational changes, and improves the comparability and credibility of the data. Standardized ultrasound image data is easier to perform subsequent analysis and processing, such as feature extraction or model training, because the basic characteristics and statistical properties of the data have been adjusted and standardized. Through gray value distribution analysis, the distribution characteristics of gray levels in ultrasound images are quantitatively extracted, which not only reflects the contrast and brightness distribution of the image, but also can reveal the gray feature differences of different tissue structures or lesion areas. The gray value distribution data provides the internal characteristics of different tissue types. Texture distribution analysis reveals the local structure and texture features in ultrasound images, such as spots, roughness, and granularity, which helps to distinguish different tissue types and lesion states, and provides a more detailed description of image features. Texture distribution data provides doctors with additional image information. Combining grayscale value and texture distribution data, a comprehensive analysis of the visual features of ultrasound images is performed, which not only integrates the brightness and contrast information of the image, but also includes the subtle structure and texture features in the image, providing a detailed visual feature description for subsequent image quality assessment and lesion analysis. The visual feature distribution data of ultrasound images provides a deeper and more comprehensive basis for image analysis, and helps to provide effective feature information in ultrasound images.

[0105] In an embodiment of the present invention, the original ultrasound image data collected by the ultrasound device is obtained, which includes different brightness levels and contrasts and may be affected by the device settings, environmental conditions or patient factors. In order to standardize the image, the following are performed: according to the preset image processing standardization operation, the brightness and contrast of the image are adjusted to ensure that all images have consistent visual effects, and digital signal processing technology, such as smoothing filtering or wavelet denoising, is applied to remove noise or interference in the image, thereby improving the quality and consistency of the image, for example, noise reduction parameters are set according to the background noise to perform noise reduction processing. Gray value distribution analysis is performed on the standardized ultrasound image data, and the number or proportion of pixels at each gray level in the image is counted to calculate the distribution of the gray histogram, and statistical features, such as mean, variance, skewness and kurtosis, are extracted from the gray distribution to describe the gray characteristics of the image, and finally the image gray value distribution data is obtained. Texture analysis is performed on the standardized ultrasound image data to describe the texture features in the image. Texture analysis algorithms, such as co-occurrence matrix, gray-level co-occurrence matrix (GLCM) or wavelet texture analysis, are applied to extract texture features from the image. Statistics of texture features, such as energy, contrast, uniformity, etc., are calculated to quantify the complexity and degree of variation of texture in the image and generate image texture distribution data. Based on the gray value distribution data and texture distribution data of the image, a comprehensive analysis of the visual features of the ultrasound image is performed. The image gray value distribution data and the image texture distribution data are merged or combined to form a comprehensive visual feature description. The correlation between different features is analyzed to determine which features are most critical for describing image quality or specific visual features, so as to analyze the distribution of ultrasound image visual features and generate ultrasound image visual feature distribution data.

[0106] Preferably, step S23 includes the following steps:

[0107] Step S231: performing grayscale value gradient calculation on the image grayscale value distribution data to generate image grayscale value gradient distribution data;

[0108] Step S232: performing grayscale value gradient feature division processing on the image grayscale value gradient distribution data according to the preset grayscale value gradient difference data to generate grayscale value gradient feature division data;

[0109] Step S233: performing image texture distribution analysis on the standard ultrasound image data according to the gray value gradient feature division data to generate image texture distribution data.

[0110] The present invention extracts more detailed grayscale change information from the ultrasound image by calculating the grayscale value gradient. These data reflect the grayscale transition and change rate of different regions in the image, which helps to describe the edges and subtle structures in the image. Grayscale value gradient distribution data can accurately reflect the contrast changes in the image, thereby enhancing the visual differences and distinctions between different regions. Grayscale value gradient feature division classifies and divides the grayscale value gradient data according to the preset gradient difference, thereby quantifying the characteristics of grayscale changes in the image, effectively reducing the complexity of the data, making subsequent analysis more efficient and controllable, and the divided feature data makes the grayscale gradient features of different images comparable and consistent, which helps to accurately compare and analyze between different samples. Texture distribution analysis based on grayscale value gradient feature division data can describe the texture features in the ultrasound image in detail, reflect the structure, roughness and texture laws of different regions in the image, and provide a basis for the detailed feature extraction of the image. Image texture distribution data provides more information about the surface and structural characteristics of the image, which helps further visual analysis and feature extraction. These data can be used in applications such as image classification and target detection to improve the comprehensiveness and accuracy of image analysis.

[0111] In an embodiment of the present invention, grayscale gradient calculation is performed on the standardized ultrasound image data. The grayscale gradient represents the rate of change of the grayscale value of each pixel in the image, and is achieved by calculating the difference or gradient of the grayscale values ​​of the neighboring pixels around each pixel. This calculation is completed by applying a difference operator or a gradient operator, such as a Sobel operator or a Prewitt operator, to obtain the grayscale gradient data of each pixel. According to the preset grayscale gradient difference data, the image grayscale gradient distribution data is subjected to feature division processing, and the grayscale gradient features in the image are classified or grouped according to the size or range of the grayscale gradient. For example, different grayscale gradient intervals are defined, and the number or proportion of pixels in each interval is counted, so as to generate grayscale gradient feature division data. Using the grayscale gradient feature division data, the standard ultrasound image data is subjected to image texture distribution analysis, and the texture features in the image, i.e., the spatial arrangement and grayscale change pattern between pixels, are identified and described, including extracting and quantifying the texture features of the image based on the grayscale co-occurrence matrix (GLCM) or wavelet transform and other technologies, and finally generating image texture distribution data.

[0112] Preferably, step S3 comprises the following steps:

[0113] Step S31: performing local grid division processing on the ultrasound image visual feature distribution data to generate grid visual feature distribution data;

[0114] Step S32: performing grid visual feature correlation analysis according to the grid visual feature distribution data to generate grid visual feature correlation data;

[0115] Step S33: using a visual feature information entropy evaluation algorithm to calculate the ultrasonic image visual feature information entropy on the grid visual feature correlation data to generate ultrasonic image visual feature information entropy data;

[0116] Step S34: Performing intelligent analysis of effective information quantification of the ultrasonic image based on the ultrasonic image visual feature information entropy data to generate effective information quantification data of the ultrasonic image.

[0117] The present invention performs local grid division processing on the ultrasonic image visual feature distribution data. The grid division can effectively integrate and divide the ultrasonic image visual feature data in space, so that the feature data in each grid unit is more concentrated and controllable. The analysis of the grid visual feature distribution data helps to extract more refined local feature information, which reflects the visual characteristics and structural differences of different regions of the image. The grid visual feature correlation analysis reveals the degree of association between visual features in different grid units, which helps to understand the interaction and dependency between local regions in the image. Through the correlation data, the relationship between the features of different parts in the image is visualized and quantified, which provides a reliable basis for subsequent image analysis and understanding. The ultrasonic image visual feature information entropy data quantifies the complexity and information richness of the visual features in the image, and provides a measure of the diversity and complexity of the image content. The information entropy calculation can reflect the diversity and distribution balance of the visual features in different grid units, which helps to evaluate the overall visual expression and information content of the image. The ultrasonic image effective information quantification data can more accurately quantify the effective information content in the image by quantitatively analyzing the information entropy data, standardize the information processing flow in the image evaluation process, and make different ultrasonic image data comparable and consistent in terms of information quantification.

[0118] As an example of the present invention, refer to Figure 2 As shown, Figure 1 Detailed implementation steps of step S3 in the flowchart, in this example, step S3 includes:

[0119] Step S31: performing local grid division processing on the ultrasound image visual feature distribution data to generate grid visual feature distribution data;

[0120] In an embodiment of the present invention, the ultrasound image visual feature distribution data is processed by local grid division, and the ultrasound image is divided into grid areas. The pixels in each area are regarded as a local visual feature unit. This division can be performed according to a preset grid size and shape, such as a square or rectangular grid. For example, an appropriate grid size and shape are selected to ensure that the entire image is covered and the spatial continuity of the local features is maintained, and local visual features such as grayscale distribution, texture features or other predefined visual features are extracted in each grid unit to generate grid visual feature distribution data.

[0121] Step S32: performing grid visual feature correlation analysis according to the grid visual feature distribution data to generate grid visual feature correlation data;

[0122] In an embodiment of the present invention, a grid visual feature correlation analysis is performed based on grid visual feature distribution data to determine the degree of correlation between visual features in different grid units. Clustering methods and information theory techniques, such as the K-Means clustering algorithm, are applied to evaluate similar features between visual features in different grid units. The similarity between adjacent grids or the similarity between a local grid and an overall grid is calculated based on the similar features. The calculated similarity is then used to perform a grid visual feature correlation analysis through information theory techniques to generate grid visual feature correlation data.

[0123] Step S33: using a visual feature information entropy evaluation algorithm to calculate the ultrasonic image visual feature information entropy on the grid visual feature correlation data to generate ultrasonic image visual feature information entropy data;

[0124] In the embodiment of the present invention, the information entropy of the visual features of the ultrasound image is calculated for the grid visual feature correlation data using the visual feature information entropy evaluation algorithm. The information entropy is used to measure the uncertainty or randomness of the visual features in the grid unit, that is, it reflects the value information of the overall ultrasound image. For example, the ultrasound image with higher information entropy reflects the anatomical contour features that can be identified, while the information entropy is a blurred black and white image. According to the selected information entropy evaluation algorithm, the information entropy of the visual features in each grid unit is calculated. The information entropy calculation results are analyzed to understand the complexity and information content of the visual features in each grid unit, as well as the information distribution pattern between different grid units, and to generate the information entropy data of the ultrasound image visual features.

[0125] Step S34: Performing intelligent analysis of effective information quantification of the ultrasonic image based on the ultrasonic image visual feature information entropy data to generate effective information quantification data of the ultrasonic image.

[0126] In an embodiment of the present invention, effective information quantification intelligent analysis of the ultrasonic image is performed based on the ultrasonic image visual feature information entropy data, the effectiveness and information content of the visual features in the ultrasonic image are quantified and evaluated, and the classification model is based on the ultrasonic image visual feature information entropy data to analyze the effective information quantification data of the ultrasonic image visual features, and identify and describe important visual feature areas or patterns in the image. The ultrasonic image effective information quantification data is used for subsequent image intelligent processing and classification tasks to generate ultrasonic image effective information quantification data.

[0127] Preferably, step S32 includes the following steps:

[0128] Step S321: performing cluster analysis on the grid visual feature distribution data according to a clustering algorithm to generate clustered grid visual feature distribution data;

[0129] Step S322: performing grid visual feature similarity evaluation processing on the clustered grid visual feature distribution data to generate grid visual feature similarity data;

[0130] Step S323: performing grid visual feature correlation analysis based on the grid visual feature similarity data to generate grid visual feature correlation data.

[0131] The present invention performs cluster analysis on grid visual feature distribution data according to a clustering algorithm. Cluster analysis combines similar features in grid visual feature distribution data into clusters, thereby effectively simplifying complex data structures into several representative categories, which is convenient for subsequent data understanding and processing. Through cluster analysis, patterns and categories hidden in grid visual feature data are discovered, helping to identify different visual feature categories and their distribution in the image. Grid visual feature similarity data quantifies the similarity and difference between visual features in different grid units, and provides a specific measure of the correlation between features in different regions of the image. Similarity data reveals the degree of correlation between visual features in different regions of the image, which helps to understand the image structure and the distribution law of visual information. Grid visual feature correlation analysis deeply explores the correlation between visual features in different grid units, helps to understand the mutual influence and synergy of each grid image in the image, and shows the degree of correlation between features in different parts of the image, providing reliable data support for further image analysis and processing.

[0132] In an embodiment of the present invention, a clustering algorithm is applied to analyze and process the grid visual feature distribution data. The clustering analysis aims to group similar grid units together to form cluster groups with similar visual feature distribution patterns. A suitable clustering algorithm is selected according to data characteristics and analysis purposes, such as K-means clustering, hierarchical clustering, density clustering, etc. Clustering features are randomly selected according to a pre-set number of clusters or an automatically determined number of clusters, and the grid visual feature distribution data is grouped. The grid units in each cluster group are regarded as having similar visual feature distribution patterns, and clustered grid visual feature distribution data is generated. The clustered grid visual feature distribution data is evaluated for grid visual feature similarity. After quantifying the visual feature similarity or difference between each grid unit, a mathematical metric or distance metric method (such as Euclidean distance, Manhattan distance, etc.) is used to calculate the visual feature similarity between different grid units. Based on the calculated similarity data, the grid visual feature distribution is quantitatively evaluated to reveal the visual feature correlation and distribution pattern between different grid units, and generate grid visual feature similarity data. Grid visual feature correlation analysis is performed based on the grid visual feature similarity data. This analysis aims to understand and describe the correlation or degree of correlation between visual features of different grid cells, and use statistical analysis methods or information theory techniques, such as Pearson correlation coefficient, mutual information quantification, etc., to analyze and quantify the correlation between visual features within grid cells, and finally generate grid visual feature correlation data.

[0133] Preferably, the visual feature information entropy evaluation algorithm in step S33 is as follows:

[0134]

[0135] In the formula, H represents the information entropy data of the visual features of the ultrasound image, N represents the number of visual feature types, and α i Expressed as the weight coefficient of the i-th visual feature type, β i Denoted as the scaling parameter of the i-th visual feature type, γ i It is represented as the i-th visual feature correlation influence parameter, and e is represented as a natural constant, ∈ i is represented as the exponential scaling parameter of the i-th visual feature type, M is represented as the number of subgrids of the standard ultrasound image data, and θ ij It is expressed as the weight coefficient of the i-th visual feature type of the j-th sub-grid, η ij Represented as the grid visual feature correlation data corresponding to the i-th visual feature type of the j-th sub-grid.

[0136] The present invention utilizes a visual feature information entropy evaluation algorithm to evaluate the effective information content in standard ultrasound image data through grid visual feature correlation data, intelligently quantify the image information value reflected by the ultrasound image, and output the result ultrasound image visual feature information entropy data reflecting the visual feature information content of the standard ultrasound image. The part reflects the importance of various visual feature types for the evaluation of the effective information content of the overall ultrasound image, that is, it reflects the effective information content of the overall ultrasound image, among which the i-th visual feature correlation influence parameter γ i It reflects the influence of the shape, spatial distribution or scale parameters of different features on the correlation of visual features of ultrasound images. It is used to index the i-th visual feature type, allowing the algorithm to take into account the nonlinear relationship between different features when evaluating the complexity and information content of the features. The information entropy contribution of the i-th visual feature type of the j-th subgrid is quantified, and processed by a logarithmic function to effectively process a large amount of data, while minimizing the contribution of larger values, thereby enhancing the algorithm's sensitivity to data changes. The above algorithm is used to calculate the effective information content of the image specifically reflected in the overall ultrasound image.

[0137] Preferably, step S4 comprises the following steps:

[0138] When the ultrasonic image effective information quantization data corresponding to the standard ultrasonic image data is greater than a preset ultrasonic image effective information quantization threshold, the standard ultrasonic image data is identified as effective ultrasonic image data; or, when the ultrasonic image effective information quantization data corresponding to the standard ultrasonic image data is not greater than the preset ultrasonic image effective information quantization threshold, the standard ultrasonic image data is identified as abnormal ultrasonic image data;

[0139] When abnormal ultrasonic image data is identified, the abnormal ultrasonic image data is transmitted to the terminal to perform an ultrasonic image acquisition abnormality feedback operation.

[0140] The present invention can accurately classify standard ultrasonic image data into two categories: effective and abnormal, based on the quantified data of effective information of ultrasonic images and preset thresholds. This classification method is based on the objective analysis of quantitative data, avoiding the uncertainty and deviation of traditional subjective evaluation methods. Through the automated discrimination process, real-time classification of ultrasonic image data is achieved, which improves the efficiency and accuracy of data processing, and is particularly important for the rapid processing and management of large amounts of ultrasonic image data. After identifying abnormal ultrasonic image data, these data can be quickly transmitted to the terminal to perform abnormal feedback operations for ultrasonic image acquisition. This real-time feedback mechanism helps to adjust ultrasonic equipment or acquisition conditions in a timely manner to improve image quality. The abnormal data collected through the feedback mechanism is deeply analyzed and evaluated, providing valuable reference information for equipment adjustment, formulation of operating specifications, and technical training in the process of ultrasonic image acquisition, thereby continuously improving and optimizing ultrasonic image quality.

[0141] In the embodiment of the present invention, a quantization threshold of effective information of ultrasonic images is pre-designed. The threshold is determined according to actual needs and application scenarios, and is determined by analyzing a large amount of data or empirical rules. It is used to compare the quantization data of effective information of ultrasonic images corresponding to standard ultrasonic image data. When the quantization data of effective information of ultrasonic images corresponding to the standard ultrasonic image data is greater than the preset quantization threshold of effective information of ultrasonic images, it reflects that the ultrasonic image data is able to identify anatomical structure features, such as anatomical structure lines, anatomical structure contours, etc. However, when the quantization data of effective information of ultrasonic images corresponding to the standard ultrasonic image data is not greater than the preset quantization threshold of effective information of ultrasonic images, it indicates that the ultrasonic image is a fuzzy feature of black and white, and the anatomical structure features cannot be identified. Therefore, the ultrasonic image data is divided by the quantization threshold of effective information of ultrasonic images, so as to obtain effective ultrasonic image data or abnormal ultrasonic image data. When abnormal ultrasonic image data is identified, the abnormal ultrasonic image data is transmitted to the terminal to perform an abnormal feedback operation of ultrasonic image acquisition, including adjusting the settings of the ultrasonic device, re-acquiring the image, or taking other corrective measures to improve the image quality. According to the result of the abnormal feedback operation, the ultrasonic image acquisition process or the device settings are adjusted to optimize the acquisition of image quality and information content.

[0142] Preferably, step S5 comprises the following steps:

[0143] Step S51: extracting ultrasonic image contour features according to the effective ultrasonic image data to generate ultrasonic image contour feature data;

[0144] Step S52: performing contour feature discretization processing on the contour feature data of the ultrasound image to generate contour feature discretization data;

[0145] Step S53: performing contour feature key node analysis according to the contour feature discretization data to generate contour feature key node data;

[0146] Step S54: performing key node descriptor analysis on the key node data of the contour feature to generate key node descriptor data;

[0147] Step S55 performs key node fuzzy diffusion feature analysis on the key node data of the contour feature according to the key node descriptor data to generate key node fuzzy diffusion feature data;

[0148] Step S56: performing ultrasound image contour fuzziness analysis on the ultrasound image contour feature data according to the key node fuzzy diffusion feature data to generate ultrasound image contour fuzziness degree data;

[0149] Step S57: performing image quality intelligent identification processing on the effective ultrasonic image data according to the ultrasonic image contour blur degree data to generate quality-identified ultrasonic image data;

[0150] Step S58: Performing an ultrasonic image intelligent classification task based on the quality-marked ultrasonic image data.

[0151] The present invention can accurately describe the shape and boundary information of organs and structures in the image by extracting the contour features of the ultrasonic image. In the process of contour feature extraction, the edge information, central node, deformation node, curvature node and other features in the contour information of the ultrasonic image are extracted, which is helpful for subsequent quantitative analysis and comparison. Discretization of contour feature data and conversion of edge feature information into discrete nodes are helpful to reduce data complexity while retaining important feature information, making subsequent analysis more efficient and reliable. Discretization can convert contour features into a form that is easy to compare and process, providing a clearer data basis for subsequent feature analysis and pattern recognition. By analyzing the key nodes of contour features, important structural features and key points in the ultrasonic image, such as deformation nodes, curvature nodes and other key nodes, are identified and extracted, which is helpful to further understand the structural composition of the image. Key node analysis helps focus on key structural features in the image, providing important clues for subsequent feature description and pattern recognition. Descriptor analysis is performed on key node data to more finely describe and quantify the features and attributes of each key node in the ultrasound image. The generated descriptor data enables a deeper comparison and analysis of the key nodes, providing more detailed and accurate information for the feature analysis and pattern recognition of the image. By performing fuzzy diffusion feature analysis on key node descriptor data, the expressive ability of key node features can be expanded and enhanced, making the structural features in the image more comprehensive and diverse. Fuzzy diffusion feature analysis converts the original key node descriptor data into richer and more complex feature data, thereby improving the representation ability of image information. Ultrasound image contour blur degree analysis is performed based on key node fuzzy diffusion feature data to quantify the blur degree of the image contour, that is, the clarity and accuracy of the image boundary. The ultrasound image contour blur degree data provides an objective quantitative basis for subsequent image quality assessment and helps to judge the visual clarity and information richness of the image. Intelligent identification processing is performed based on the ultrasound image contour blur degree data to generate quality identification ultrasound image data to indicate the quality level or category of the image. The quality identification ultrasound image data can provide an important reference for subsequent automated processing and decision-making, and support intelligent classification and further analysis of the image. Intelligent classification of ultrasound images based on quality-labeled ultrasound image data can improve the precision and accuracy of classification, thereby effectively distinguishing ultrasound images of different quality levels. Intelligent classification can classify images in an automated manner, optimize the processing flow, and improve work efficiency and quality control levels.

[0152] As an example of the present invention, refer to Figure 3 As shown, Figure 1 FIG. 5 is a flow chart of detailed implementation steps of step S5 in FIG. 5 . In this example, step S5 includes:

[0153] Step S51: extracting ultrasonic image contour features according to the effective ultrasonic image data to generate ultrasonic image contour feature data;

[0154] In an embodiment of the present invention, an appropriate edge detection algorithm, such as a Canny edge detection algorithm or a Sobel operator, is applied to extract contour features of an ultrasonic image based on valid ultrasonic image data, and preliminary information of the contour is extracted from the preprocessed image. Based on the image obtained by edge detection, a contour extraction algorithm (such as a contour search function in OpenCV) is used to extract the actual contour in the image. The contour includes a closed contour or a continuous contour line of an edge in the image. Each contour is represented as a set of feature points or feature vectors. These features include the length, area, shape descriptor (such as Hu moment), or other geometric and topological features of the contour. The extracted and represented contour feature data is saved to obtain ultrasonic image contour feature data.

[0155] Step S52: performing contour feature discretization processing on the contour feature data of the ultrasound image to generate contour feature discretization data;

[0156] In an embodiment of the present invention, the contour feature data of the ultrasound image is analyzed to understand the numerical range and distribution of the features, and the contour features are discretized according to the distribution of the feature data and application requirements, such as equal-interval discretization, adaptive discretization based on data distribution, or a discretization method defined based on business rules. The contour feature data is converted into a discretized form, and continuous numerical features are converted into discrete categories or intervals. This is achieved by setting thresholds or segmentation points to ensure that the generated contour feature discretization data can capture important feature information, including discretized contour feature values ​​or feature categories, such as discretized nodes, shapes, etc.

[0157] Step S53: performing contour feature key node analysis according to the contour feature discretization data to generate contour feature key node data;

[0158] In an embodiment of the present invention, a key node analysis of the contour feature is performed based on the discretized data of the contour feature to define the key nodes in the contour feature. The key nodes can be specific categories and corresponding node values, such as deformation feature nodes, curvature change nodes, etc., which are used to represent important change points or characteristic value intervals of the contour feature. Data mining or image processing technology, such as threshold setting or rule definition, is used to extract key nodes from the discretized contour feature data to generate contour feature key node data.

[0159] Step S54: performing key node descriptor analysis on the key node data of the contour feature to generate key node descriptor data;

[0160] In an embodiment of the present invention, a key node descriptor analysis is performed on the key node data of the contour feature, and the key node descriptor information is extracted from the visual change characteristics of the feature nodes by using image processing technology, for example, the fuzzy diffusion intensity and fuzzy diffusion range of the feature nodes are calculated by applying a point diffusion calculation function, and the extracted key node descriptor information is statistically analyzed by a statistical algorithm to generate key node descriptor data. A key node descriptor analysis is performed on the key node data of the contour feature, and the key node descriptor information is extracted from the visual change characteristics of the feature nodes by using image processing technology, for example, the gray value change diffusion characteristics or contrast change characteristics of the feature nodes and the surroundings are calculated by applying a point diffusion calculation function, and the extracted key node descriptor information is statistically analyzed by a statistical algorithm to generate key node descriptor data.

[0161] Step S55 performs key node fuzzy diffusion feature analysis on the key node data of the contour feature according to the key node descriptor data to generate key node fuzzy diffusion feature data;

[0162] In an embodiment of the present invention, a fuzzy feature diffusion algorithm, such as a power spectral density function or an autocorrelation function, is applied to calculate the fuzzy diffusion degree of the key node descriptor data in space and attributes to reflect the fuzziness or uncertainty of the changes of these nodes, and through statistical analysis, spatial analysis or model inference, the fuzzy diffusion intensity and fuzzy diffusion range of the key node descriptor data are calculated to capture the fuzzy change pattern or trend between nodes. The key node fuzzy diffusion feature identification of the contour feature key node data is performed through the calculation results, reflecting the fuzzy diffusion features corresponding to each contour feature key node data, and generating key node fuzzy diffusion feature data.

[0163] Step S56: performing ultrasound image contour fuzziness analysis on the ultrasound image contour feature data according to the key node fuzzy diffusion feature data to generate ultrasound image contour fuzziness degree data;

[0164] In an embodiment of the present invention, an intelligent blur degree scoring engine is designed by scoring the clarity of contour lines or other visual quality assessment indicators, and the blur degree of key node blur diffusion feature data is scored by the intelligent blur degree scoring engine. The ultrasound image contour feature data is marked with ultrasound image contour blur degree based on the blur degree scoring result, and ultrasound image contour blur degree data is generated.

[0165] Step S57: performing image quality intelligent identification processing on the effective ultrasonic image data according to the ultrasonic image contour blur degree data to generate quality-identified ultrasonic image data;

[0166] In an embodiment of the present invention, according to a pre-designed scoring standard, the effective ultrasound image data is automatically scored for image quality using ultrasound image contour blur degree data, involving the application of logical judgment, machine learning model or rule engine to generate quality-marked ultrasound image data.

[0167] Step S58: Performing an ultrasonic image intelligent classification task based on the quality-marked ultrasonic image data.

[0168] In an embodiment of the present invention, a trained classification model or algorithm is prepared for classifying the quality-marked ultrasound image data, including supervised learning, unsupervised learning or semi-supervised learning methods. The quality-marked ultrasound image data is input into the classification model for processing. Ensure that the data format and features meet the requirements of the model. According to the trained model, each quality-marked ultrasound image is intelligently classified, and the model assigns the ultrasound image to a predefined classification or category, such as normal, local abnormality, abnormality, and other levels. Output the classification result of each ultrasound image, and provide feedback or record when necessary.

[0169] The beneficial effect of the present application is that the method for intelligent detection and classification of image quality based on ultrasound images of the present invention eliminates the uncertainty and error of subjective human identification in traditional methods by objectively evaluating the quality of ultrasound images, ensures the objectivity and consistency of quality evaluation, unifies the standard for image quality detection and classification of quantitative ultrasound images, thereby screening out high-quality ultrasound images of ultrasound equipment adjusted under different parameters, and combines artificial intelligence technology to develop intelligent algorithm models, which can learn and identify key features of ultrasound images from a large amount of data, improve the accuracy and generalization ability of image quality evaluation, and effectively solve the problems of accuracy, efficiency and generalization ability of ultrasound imaging quality evaluation. Therefore, no matter from which point of view, the embodiments should be regarded as exemplary and non-restrictive, and the scope of the present invention is limited by the appended claims rather than the above description, and it is intended to include all changes within the meaning and scope of the equivalent elements of the application documents in the present invention.

[0170] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. An intelligent detection and classification method for image quality based on ultrasound images, characterized in that: The following steps are involved: Step S1: custom-adjusting the ultrasound device according to a preset ultrasound device multi-source adjustment parameter group to obtain an adjusted ultrasound device, and performing an ultrasound image acquisition operation on a patient through the adjusted ultrasound device; and performing ultrasound image acquisition based on the ultrasound image acquisition operation to obtain ultrasound image data; Step S2: performing image standardization processing on the ultrasonic image data to generate standard ultrasonic image data; based on the standard ultrasonic image data; performing ultrasonic image visual feature distribution analysis to generate ultrasonic image visual feature distribution data; Step S3: performing local grid division processing on the ultrasonic image visual feature distribution data to generate grid visual feature distribution data; performing ultrasonic image visual feature information entropy analysis based on the grid visual feature distribution data to generate ultrasonic image visual feature information entropy data; performing ultrasonic image effective information quantization intelligent analysis based on the ultrasonic image visual feature information entropy data to generate ultrasonic image effective information quantization data; Step S4: dividing the standard ultrasonic image data into valid and abnormal ultrasonic image data according to the ultrasonic image effective information quantization data to obtain valid ultrasonic image data or abnormal ultrasonic image data; when abnormal ultrasonic image data is identified, transmitting the abnormal ultrasonic image data to the terminal to perform ultrasonic image acquisition abnormality feedback operation; Step S5: extract ultrasonic image contour features based on the effective ultrasonic image data to generate ultrasonic image contour feature data; perform ultrasonic image contour blurriness analysis on the ultrasonic image contour feature data to generate ultrasonic image contour blurriness data; perform image quality intelligent identification processing on the effective ultrasonic image data based on the ultrasonic image contour blurriness data to generate quality-identified ultrasonic image data; and perform ultrasonic image intelligent classification based on the quality-identified ultrasonic image data.

2. The method for intelligent detection and classification of image quality based on ultrasonic images according to claim 1, characterized in that: Step S1 includes the following steps: Step S11: custom-adjusting the ultrasound device according to the preset ultrasound device multi-source adjustment parameter group to obtain an adjusted ultrasound device, and performing an ultrasound image acquisition operation on the patient through the adjusted ultrasound device; Step S12: performing ultrasonic signal acquisition based on the ultrasonic image acquisition operation to obtain an ultrasonic signal; Step S13: Performing ultrasonic image conversion based on the ultrasonic signal to obtain ultrasonic image data.

3. The method for intelligent detection and classification of image quality based on ultrasonic images according to claim 2, characterized in that: Step S13 includes the following steps: Step S131: performing spectrum data conversion according to the ultrasonic signal to generate ultrasonic spectrum data; Step S132: performing ultrasonic pixel block analysis according to the ultrasonic spectrum data to obtain ultrasonic pixel block data; Step S133: performing ultrasonic image conversion according to the ultrasonic pixel block data to obtain ultrasonic image data.

4. The method for intelligent detection and classification of image quality based on ultrasonic images according to claim 1, characterized in that: Step S2 includes the following steps: Step S21: performing image standardization processing on the ultrasonic image data to generate standard ultrasonic image data; Step S22: performing grayscale value distribution analysis on the standard ultrasound image data to generate image grayscale value distribution data; Step S23: performing image texture distribution analysis on the standard ultrasound image data to generate image texture distribution data; Step S24: performing ultrasound image visual feature distribution analysis based on the image gray value distribution data and the image texture distribution data to generate ultrasound image visual feature distribution data.

5. The method for intelligent detection and classification of image quality based on ultrasonic images according to claim 4, characterized in that: Step S23 includes the following steps: Step S231: performing grayscale value gradient calculation on the image grayscale value distribution data to generate image grayscale value gradient distribution data; Step S232: performing grayscale value gradient feature division processing on the image grayscale value gradient distribution data according to the preset grayscale value gradient difference data to generate grayscale value gradient feature division data; Step S233: performing image texture distribution analysis on the standard ultrasound image data according to the gray value gradient feature division data to generate image texture distribution data.

6. The method for intelligent detection and classification of image quality based on ultrasonic images according to claim 1, characterized in that: Step S3 includes the following steps: Step S31: performing local grid division processing on the ultrasound image visual feature distribution data to generate grid visual feature distribution data; Step S32: performing grid visual feature correlation analysis according to the grid visual feature distribution data to generate grid visual feature correlation data; Step S33: using a visual feature information entropy evaluation algorithm to calculate the ultrasonic image visual feature information entropy on the grid visual feature correlation data to generate ultrasonic image visual feature information entropy data; Step S34: Performing intelligent quantitative analysis of effective information of the ultrasonic image based on the ultrasonic image visual feature information entropy data to generate quantified data of effective information of the ultrasonic image.

7. The method for intelligent detection and classification of image quality based on ultrasonic images according to claim 6, characterized in that: Step S32 includes the following steps: Step S321: performing cluster analysis on the grid visual feature distribution data according to a clustering algorithm to generate clustered grid visual feature distribution data; Step S322: performing grid visual feature similarity evaluation processing on the clustered grid visual feature distribution data to generate grid visual feature similarity data; Step S323: performing grid visual feature correlation analysis based on the grid visual feature similarity data to generate grid visual feature correlation data.

8. The method for intelligent detection and classification of image quality based on ultrasonic images according to claim 6, characterized in that: The visual feature information entropy evaluation algorithm in step S33 is as follows: In the formula, H represents the information entropy data of the visual features of the ultrasound image, N represents the number of visual feature types, and α i Expressed as the weight coefficient of the i-th visual feature type, β i Denoted as the scaling parameter of the i-th visual feature type, γ i It is represented as the i-th visual feature correlation influence parameter, and e is represented as a natural constant, ∈ i is represented as the exponential scaling parameter of the i-th visual feature type, M is represented as the number of subgrids of the standard ultrasound image data, and θ ij It is expressed as the weight coefficient of the i-th visual feature type of the j-th sub-grid, η ij Represented as the grid visual feature correlation data corresponding to the i-th visual feature type of the j-th sub-grid.

9. The method for intelligent detection and classification of image quality based on ultrasonic images according to claim 1, characterized in that: Step S4 includes the following steps: When the ultrasonic image effective information quantization data corresponding to the standard ultrasonic image data is greater than a preset ultrasonic image effective information quantization threshold, the standard ultrasonic image data is identified as effective ultrasonic image data; or, when the ultrasonic image effective information quantization data corresponding to the standard ultrasonic image data is not greater than the preset ultrasonic image effective information quantization threshold, the standard ultrasonic image data is identified as abnormal ultrasonic image data; When abnormal ultrasonic image data is identified, the abnormal ultrasonic image data is transmitted to the terminal to perform an ultrasonic image acquisition abnormality feedback operation.

10. The method for intelligent detection and classification of image quality based on ultrasonic images according to claim 1, characterized in that: Step S5 includes the following steps: Step S51: extracting ultrasonic image contour features according to the effective ultrasonic image data to generate ultrasonic image contour feature data; Step S52: performing contour feature discretization processing on the contour feature data of the ultrasound image to generate contour feature discretization data; Step S53: performing contour feature key node analysis according to the contour feature discretization data to generate contour feature key node data; Step S54: performing key node descriptor analysis on the key node data of the contour feature to generate key node descriptor data; Step S55 performs key node fuzzy diffusion feature analysis on the key node data of the contour feature according to the key node descriptor data to generate key node fuzzy diffusion feature data; Step S56: performing ultrasound image contour fuzziness analysis on the ultrasound image contour feature data according to the key node fuzzy diffusion feature data to generate ultrasound image contour fuzziness degree data; Step S57: performing image quality intelligent identification processing on the effective ultrasonic image data according to the ultrasonic image contour blur degree data to generate quality-identified ultrasonic image data; Step S58: Performing an ultrasonic image intelligent classification task based on the quality-marked ultrasonic image data.

Images