Image processing method and device, storage medium and electronic device

Abstract

The invention discloses an image processing method and device, a storage medium and an electronic device. The method comprises the following steps: receiving an image to be detected, wherein the image to be detected is obtained by performing magnetic resonance imaging scanning on the liver of a detector; determining field intensity used by the to-be-detected image to obtain target field intensity; selecting a value of a target feature corresponding to the target field intensity from the feature data of the to-be-detected image to obtain a to-be-detected numerical value, the target feature satisfying a preset condition of high diagnosis efficiency; according to the boundary value corresponding to the target feature in the target field intensity, the grading result of the liver inflammation of the detector is determined, the grading result of the liver inflammation can be accurately determined, and the diagnosis efficiency and accuracy are improved.

Description

Technical Field

[0001] This article relates to image recognition technology, and in particular to an image processing method, device, storage medium and electronic device. Background Art

[0002] In medical institutions, the diagnostic process for liver inflammation typically follows these steps: First, the doctor conducts a detailed interview to understand the patient's medical history, symptoms, previous medical conditions, and possible triggers, thereby identifying suspected areas of liver inflammation or areas relevant to the diagnosis. Subsequently, medical imaging scans, such as ultrasound, CT, or magnetic resonance imaging (MRI), are performed on these areas to obtain images of the liver's internal structure. However, the key issue lies in the subsequent image analysis process. This step is highly dependent on the doctor's personal experience. The doctor needs to rely on his or her accumulated professional knowledge and clinical experience to identify and interpret subtle changes in the scanned images to determine the presence of liver inflammation and the type and extent of the inflammation. Due to differences in experience and professional level among different doctors, this may lead to deviations in the interpretation of the same image, which in turn affects the final diagnosis and increases the risk of misdiagnosis. Summary of the Invention

[0003] Embodiments of the present application provide an image processing method, apparatus, storage medium, and electronic device.

[0004] An image processing method, comprising:

[0005] receiving an image to be tested, wherein the image to be tested is obtained by performing a magnetic resonance imaging scan of a subject's liver;

[0006] Determining the field strength used by the image to be measured to obtain a target field strength;

[0007] Selecting a value of a target feature corresponding to the target field strength from the feature data of the image to be measured to obtain a value to be measured, wherein the target feature satisfies a preset condition of high diagnostic efficacy;

[0008] The grading result of the liver inflammation of the test subject is determined according to the threshold value of the target feature corresponding to the target field strength.

[0009] An image processing device, comprising:

[0010] A receiving module configured to receive an image to be tested, wherein the image to be tested is obtained by performing a magnetic resonance imaging scan of a subject's liver;

[0011] A first determining module is configured to determine the field strength used by the image to be measured to obtain a target field strength;

[0012] A selection module is configured to select a value of a target feature corresponding to the target field strength from the feature data of the image to be measured to obtain a value to be measured, wherein the target feature satisfies a preset condition of high diagnostic efficiency;

[0013] The second determination module is configured to determine the grade result of the liver inflammation of the test subject according to the threshold value corresponding to the target field strength of the target feature.

[0014] A storage medium stores a computer program, wherein the computer program is configured to execute the method described above when running.

[0015] An electronic device includes a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to perform the method described above.

[0016] In the embodiment of the present application, by obtaining the target field strength of the image to be tested and selecting the target features that meet the high diagnostic efficiency conditions, the grading results of liver inflammation can be accurately determined based on the threshold value corresponding to the target field strength of the target features, thereby improving the diagnostic efficiency and accuracy.

[0017] Other features and advantages of the present application will be described in the following description, and in part will become apparent from the description, or will be understood by practicing the present application. Other advantages of the present application can be realized and obtained by the solutions described in the description and the drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The accompanying drawings are used to provide an understanding of the technical solution of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the technical solution of the present application and do not constitute a limitation on the technical solution of the present application.

[0019] Figure 1 A flowchart of an image recognition method provided in an embodiment of the present application;

[0020] Figure 2 A schematic diagram of the structure of the image recognition device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0021] This application describes multiple embodiments, but this description is exemplary rather than restrictive, and it will be apparent to those skilled in the art that there may be more embodiments and implementations within the scope of the embodiments described herein. Although many possible feature combinations are shown in the drawings and discussed in the detailed description, many other combinations of the disclosed features are also possible. Unless specifically limited, any feature or element of any embodiment may be used in combination with any other feature or element in any other embodiment, or may replace any other feature or element in any other embodiment.

[0022] The present application includes and contemplates combinations of features and elements known to those of ordinary skill in the art. The embodiments, features, and elements disclosed in this application may also be combined with any conventional features or elements to form a unique inventive solution. Any features or elements of any embodiment may also be combined with features or elements from other inventive solutions to form another unique inventive solution. Therefore, it should be understood that any feature shown and / or discussed in this application may be implemented individually or in any appropriate combination. Therefore, except for the limitations made according to the appended claims and their equivalents, the embodiments are not subject to other limitations. In addition, various modifications and changes may be made within the scope of protection of the appended claims.

[0023] In addition, when describing representative embodiments, the specification may have presented the method and / or process as a specific sequence of steps. However, to the extent that the method or process does not rely on the specific order of the steps described herein, the method or process should not be limited to the steps in the specific order described. As will be understood by those skilled in the art, other orders of steps are also possible. Therefore, the specific order of the steps set forth in the specification should not be interpreted as a limitation to the claims. In addition, the claims for the method and / or process should not be limited to performing their steps in the order written, and those skilled in the art can readily understand that these orders can be changed and still remain within the spirit and scope of the embodiments of the present application.

[0024] Diffusion-weighted imaging (DWI) is a technique that uses the Brownian motion of water molecules diffusing within living tissues to reveal the internal functional and structural characteristics of tissues and organs. This technique, which uses magnetic resonance imaging (MRI) to scan the human body, offers unique advantages such as speed, safety, and quantitative analysis. Its application is increasingly effective in differential diagnosis of benign and malignant liver lesions and quantitative analysis of liver inflammation.

[0025] The apparent diffusion coefficient (ADC) is a key quantitative metric in DWI imaging, reflecting the Brownian motion of water molecules in tissue. In patients with chronic hepatitis B (CHB), liver inflammation restricts the diffusion of water molecules, leading to changes in the liver's ADC value. By observing changes in the liver's ADC value, the severity of liver inflammation can be quantitatively analyzed.

[0026] During the implementation of the technical solution of this application, it was further discovered that there were significant statistical differences in the ADC values between CHB patients and healthy individuals, and that the ADC values between different inflammatory groups also showed statistical differences. This indicates that the ADC value has a high diagnostic value and can be improved by combining it with the characteristic imaging manifestations of conventional MRI. It should be noted that the liver ADC value of the same CHB patient at a field strength of 3.0T is generally lower than the liver ADC value at a field strength of 1.5T.

[0027] Based on these findings, the embodiments of the present application propose a solution that aims to improve diagnostic efficiency by utilizing differences in DWI imaging technology and changes in ADC values.

[0028] Figure 1 Schematic diagram of the process of establishing the image recognition model provided in the embodiment of the present application. Figure 1 As shown, the method includes:

[0029] Step 101: Receive an image to be tested.

[0030] The purpose of this step is to obtain the basic data source for subsequent liver inflammation grading analysis, namely the image to be tested. The image to be tested here specifically refers to the image obtained by performing a magnetic resonance imaging scan of the liver of the test subject. Magnetic resonance imaging (MRI) is a technology that uses the principle of resonance of atomic nuclei in a magnetic field to generate images of the internal structure of the human body. It has many advantages such as being non-invasive and radiation-free. It can more clearly present information such as the different signal characteristics of liver tissue, providing a key imaging basis for subsequent image processing and analysis.

[0031] Step 102: Determine the field strength used by the image to be measured to obtain the target field strength.

[0032] This step aims to clarify the field strength of the MRI scanner corresponding to the image to be measured, that is, to determine the target field strength. Field strength is a key parameter of MRI equipment, and different field strengths may differ in imaging principles, signal characteristics, and image quality. Determining the target field strength is crucial for the subsequent targeted selection of appropriate feature data and accurate liver inflammation grading, because the liver's characteristic data such as the apparent diffusion coefficient (ADC) will have different performances at different field strengths, and these differences will directly affect the accuracy of inflammation grading.

[0033] Step 103: Select the value of the target feature corresponding to the target field strength from the feature data of the image to be measured to obtain the value to be measured.

[0034] The purpose of this step is to screen out the target feature values that meet the preset high diagnostic efficiency conditions from the numerous feature data of the image to be tested based on the determined target field strength, and then obtain the measured value. Feature data is obtained by performing various analyses and extractions on the image, and can reflect the different characteristics of liver tissue in the image. Target features refer to those features that have been determined to have high diagnostic value for liver inflammation grading under specific field strengths through preliminary research, verification, and other processes. Selecting appropriate target features and their values can extract key information closely related to the liver inflammation grade from a large amount of data, laying the foundation for subsequent accurate judgment of the inflammation grade.

[0035] Step 104: Determine the grade of the subject's liver inflammation based on the threshold corresponding to the target feature at the target field strength.

[0036] This step is the core application link of the entire method. Its function is to grade the liver inflammation of the subject according to the preset threshold value of the target feature under the target field strength. The threshold value is a critical value determined through the analysis and statistics of a large amount of sample data. The value of the target feature is compared with the threshold value to judge the severity of the liver inflammation and ultimately determine the grade of the subject's liver inflammation. This step combines the feature data extracted in the early stage with the actual clinical diagnosis needs, achieving the purpose of quantitative grading and evaluation of liver inflammation using image processing technology, and providing valuable reference information for clinical diagnosis.

[0037] The method provided in the embodiment of the present application obtains the target field strength of the image to be tested, selects target features that meet the conditions of high diagnostic efficiency, and accurately determines the grading results of liver inflammation based on the threshold value corresponding to the target field strength of the target feature, thereby improving diagnostic efficiency and accuracy.

[0038] The following describes the method provided in the embodiment of the present application:

[0039] In an exemplary embodiment, the target feature and its threshold are obtained by:

[0040] Step A1: Obtain a sample data set.

[0041] The purpose of this step is to establish a data foundation for analysis and training. The sample dataset must contain at least two sample images for each subject, and these images must meet the following conditions:

[0042] Same imaging parameters: All images must be acquired under the same imaging parameters (such as layer thickness, scanning sequence, parameter b value, etc.) to eliminate the interference of parameter differences on the results.

[0043] Different field strengths: Images of the same subject need to be acquired using MRI equipment with different field strengths (e.g., 1.5T and 3.0T) to study the effect of field strength on feature data.

[0044] Clear grouping: Subjects need to be divided into a control group (such as healthy people or chronic hepatitis B patients with inflammation grade ≤G1) and a case group (such as patients with inflammation grade ≥G2) so that the characteristic differences between the two groups can be distinguished through statistical methods in the future.

[0045] In practical applications, image data that meets the above conditions is obtained and the accuracy of data annotation (field intensity, grouping) is ensured. This step provides a standardized data source for subsequent feature screening and threshold determination.

[0046] Step A2: Acquire characteristic data under the same field intensity.

[0047] The purpose of this step is to extract quantitative indicators related to field strength for subsequent analysis. The specific implementation requires:

[0048] Unified field strength grouping: The sample data sets are classified by field strength (such as 1.5T group and 3.0T group).

[0049] Extracting feature data: Extract feature data from each set of images using image processing tools (such as FireVoxel software).

[0050] This step ensures field-strength specificity for subsequent analyses by normalizing the data.

[0051] Step A3: Determine the target feature and its threshold.

[0052] The purpose of this step is to screen out features with diagnostic significance through statistical methods and determine their classification threshold (cutoff value). The specific implementation includes:

[0053] Group comparative analysis: Under the same field intensity, the characteristic data of the case group and the control group are compared, and the characteristics with significant differences between the two groups are screened out through hypothesis testing (such as t-test, ROC curve analysis).

[0054] Cutoff calculation: Based on the selected features, the classification cutoff is determined using the maximum Youden index method or the optimal cutoff point of the receiver operating characteristic (ROC) curve. For example, if the cutoff value of a texture feature at a field strength of 1.5 Tesla is X, then samples exceeding X are classified as ≥G2 inflammation.

[0055] This step ensures that the target features and their thresholds have high diagnostic efficacy through a data-driven approach, providing a reliable basis for subsequent clinical applications.

[0056] By constructing a multi-field strength sample dataset, standardizing feature extraction, and conducting statistical analysis, we achieved scientific screening of target features and their thresholds. The core of this approach is to utilize field-strength-specific data and group comparison to address the issue of consistent diagnostic indicators across different magnetic field intensities, laying the foundation for precise grading of liver inflammation.

[0057] In an exemplary embodiment, the method further includes a verification operation, specifically comprising:

[0058] Step B1: Obtain a validation dataset.

[0059] The purpose of this step is to construct a data set for verifying the reliability of target features and thresholds. The verification data set contains at least two sample images of the verifier, and these images must meet the following conditions: the images of the same verifier must be obtained by performing DWI scans of the liver using magnetic resonance imaging equipment with different field strengths (such as 1.5T and 3.0T) under the same imaging parameters (such as scanning sequence, parameter b value). When implemented, it is necessary to prospectively collect imaging data of patients with chronic hepatitis B (CHB) that meet clinical criteria through multiple centers or single centers to ensure the diversity and representativeness of the data. This step provides a real, multi-field strength imaging data foundation for subsequent verification.

[0060] Step B2: extracting feature data of the verification image.

[0061] This step quantifies the pathological features of the verification images and provides data support for grading. Image analysis software (such as FireVoxel) is used to perform ADC histogram analysis on each verifier's DWI images to extract feature data. This requires image preprocessing (such as denoising and normalization) and automatic generation of feature data based on a histogram algorithm. This step converts the raw images into quantifiable and comparable numerical metrics.

[0062] Step B3: Generate the grading results to be verified.

[0063] The purpose of this step is to use the target features and their cutoff values to predict the grade of liver inflammation in the verifier. The target features (such as ADC mean) and their corresponding cutoff values at different field strengths must be determined (such as the cutoff value at 1.5T is 0.8×10-3mm 2 / s), compares the feature data of the same verifier at different field intensities, and outputs the inflammation grade (e.g., ≤G1 or ≥G2) at each field strength. This implementation requires establishing a mapping table between field strength, feature, and threshold, and performing numerical matching via an automated script. This step verifies the applicability of the target feature at different field intensities.

[0064] Step B4: Calculate the accuracy of the target feature.

[0065] The purpose of this step is to evaluate the diagnostic reliability of the target feature and the cutoff value. The grading results to be verified by the validator under the same field strength need to be compared with their benchmark grading results (such as the pathological results of liver puncture biopsy), and the sensitivity, specificity, positive predictive value and other indicators need to be statistically analyzed, and the comprehensive accuracy needs to be calculated. When implementing, it is necessary to draw the ROC curve through statistical software (such as SPSS), calculate the area under the curve (AUC), and use the preset accuracy threshold (such as AUC ≥ 0.85) as the screening criterion. This step uses quantitative indicators to screen out target features with high diagnostic efficacy.

[0066] Step B5: Screen target features involved in detection.

[0067] The purpose of this step is to ensure that the target features ultimately used for detection are clinically practical. Based on the accuracy calculation results, only target features that meet the preset threshold (such as sensitivity ≥ 90%) are retained for subsequent detection image processing. During implementation, a feature elimination mechanism needs to be established to eliminate features with insufficient accuracy and update the field strength-feature database. This step improves the overall diagnostic performance of liver inflammation grading through dynamic optimization.

[0068] The accuracy and universality of liver inflammation grading are improved through multi-field strength verification and dynamic feature optimization. Specifically: First, by obtaining a multi-field strength verification data set and extracting feature data, the repeatability of the features under different magnetic field strengths (such as 1.5T and 3.0T) is ensured; secondly, the sensitivity, specificity and other indicators are calculated in combination with the pathological benchmark results (such as puncture biopsy), and the diagnostic efficacy of the target feature is evaluated using the ROC curve, and features with high accuracy (such as AUC ≥ 0.85) are selected for grading to avoid misjudgment of ADC values due to field strength differences; finally, through multi-center external verification and cross-device testing, the feature selection mechanism is optimized to enhance the generalization ability of the method in real clinical scenarios.

[0069] In an exemplary embodiment, the cutoff value of the target feature is determined by combining the manufacturer information of the imaging device, thereby solving the problem of the difference in imaging performance between devices from different manufacturers on the grading of liver inflammation. Specifically, due to differences in hardware design (such as gradient coils, radio frequency calibration) and software algorithms (such as image reconstruction, noise suppression) of magnetic resonance equipment from different manufacturers, even at the same field strength (such as 3.0T), the image features (such as ADC mean, entropy value) generated by different devices may show systematic deviations. For example, the ADC mean of manufacturer A's equipment is generally 10% higher than that of manufacturer B. If a single cutoff value is used uniformly, the grading results will be offset.

[0070] The specific implementation method is:

[0071] Manufacturer-field strength matching database: The device manufacturer of each image in the sample dataset is annotated, and the images are grouped by "manufacturer + field strength" combination (such as manufacturer A-1.5T, manufacturer B-3.0T). The distribution of feature data of each group is statistically analyzed in combination with pathological results (such as G1 / G2 grade), and the optimal threshold is determined for each group.

[0072] Dynamic calibration mechanism: When a new device or manufacturer is added, the threshold values are recalculated through the verification data set and the database is updated to ensure that the threshold values are synchronized with the device characteristics.

[0073] Gradual application: When processing images from the tester, the corresponding threshold is applied according to the device manufacturer and field emphasis. For example, the ADC mean threshold of the manufacturer C-3.0T device is set to value A, rather than value B of other manufacturers.

[0074] The advantages of the above processing operation are:

[0075] Eliminate device bias: Avoid systematic errors in feature data due to differences in manufacturers and improve the consistency of diagnosis across devices;

[0076] Enhance clinical applicability: Support multi-center collaboration and cross-hospital follow-up of patients, ensuring unified grading standards when different institutions use equipment from different manufacturers.

[0077] Optimize diagnostic performance: Improve sensitivity (reduce missed diagnoses) and specificity (reduce misdiagnoses) through manufacturer-specific cutoffs.

[0078] The above implementation method, through refined threshold management, incorporates device heterogeneity into the grading model, significantly improving the accuracy and clinical operability of liver inflammation assessment.

[0079] In an exemplary embodiment, the feature data includes texture feature data extracted based on an image and primary statistical data extracted based on an ADC histogram corresponding to the image.

[0080] The texture feature data records the tissue structure and heterogeneity of liver tissue.

[0081] Tissue structure data reflects the normal structure and morphology of liver tissue, such as the arrangement of hepatocytes and the structure of liver lobules.

[0082] Heterogeneity data describes the heterogeneity within liver tissue, including the differences between normal tissue and diseased tissue (e.g., plaques). Texture feature data of plaques (e.g., size, shape, density, and distribution) are part of heterogeneity data.

[0083] By acquiring texture feature data, liver lesions such as inflammation, fibrosis, and tumors can be more accurately identified and evaluated, thus providing important basis for clinical diagnosis and treatment.

[0084] Specifically, texture features include the following feature parameters:

[0085] A1. Voxel count

[0086] Definition: Voxel count refers to the number of voxels contained in the ROI.

[0087] Function: Voxel count directly reflects the size of ROI and can be used to compare volume differences between different ROIs.

[0088] A2, Volume

[0089] Definition: Volume refers to the size of the ROI area.

[0090] Function: Volume measurement provides the actual size of ROI in three-dimensional space, which is important for evaluating the size and changes of lesions.

[0091] A3. Voxel count (voxel count)

[0092] Definition: The number of voxels within an ROI that have been classified based on their signal intensity or other characteristics.

[0093] Function: This classified voxel count can help identify and quantify regions with different characteristics within the ROI, such as high signal intensity areas and low signal intensity areas.

[0094] A4, Nonuniformity

[0095] Definition: Nonuniformity is a measure of the uniformity of signal intensity or voxel distribution within a ROI.

[0096] Effect: The higher the inhomogeneity, the more uneven the signal intensity or voxel distribution within the ROI, which may indicate a higher heterogeneity of the lesion.

[0097] The above characteristic parameters describe the shape, size, and internal structure of ROI from different perspectives, providing rich quantitative information for imaging omics analysis and helping to more comprehensively evaluate the morphological and functional changes of tissues.

[0098] In addition, in the embodiment of the present application, the texture features also include:

[0099] A5. Num Blobs (Number of Blobs)

[0100] Definition: The number of independent patches or connected regions identified within the ROI.

[0101] Role: The number of plaques can reflect the distribution of lesions within the ROI. Multiple small plaques may indicate diffuse lesions, while a small number of large plaques may indicate localized lesions. This helps distinguish different types of liver diseases, such as diffuse inflammation versus localized tumors.

[0102] A6. Blob stdev (standard deviation of plaques)

[0103] Definition: Blob stdev refers to the standard deviation of voxel count or signal intensity of each plaque.

[0104] Purpose: Plaque standard deviation reflects the heterogeneity between plaques. A larger standard deviation indicates greater differences between plaques. This helps assess the homogeneity of lesions. For example, a higher standard deviation may indicate lesion diversity or complexity.

[0105] In radiomics analysis, the combined use of texture features A5 (Num Blobs) and A6 (Blob stdev) has the following advantages, including:

[0106] Comprehensive assessment: Provides comprehensive information on the distribution and heterogeneity of lesions. This helps to more accurately diagnose and differentiate different types of liver diseases.

[0107] Improved diagnostic accuracy: By quantifying the distribution and heterogeneity of lesions, these features can improve the diagnostic performance of the model, especially when distinguishing complex lesions.

[0108] Decision support: These texture features can provide clinicians with additional objective information to assist in formulating treatment plans and predicting disease progression.

[0109] From the above analysis, it can be seen that in the assessment of liver inflammation and fibrosis, A5 and A6 can help identify the pattern and nature of lesions, providing strong support for non-invasive diagnosis and grading.

[0110] The ADC histogram is generated by statistically analyzing the distribution of ADC values in DWI images. By generating the corresponding ADC histogram for a selected ROI on the image, a series of first-order statistical features, such as mean, standard deviation, skewness, and kurtosis, can be obtained. These features reflect the overall distribution of water molecule diffusion in liver tissue.

[0111] The first-order statistics based on the ADC histogram include the following characteristic parameters:

[0112] B1, Mean

[0113] Definition: The mean is the average signal intensity of all pixels within the ROI.

[0114] Function: It reflects the average level of ADC value in the entire ROI area and can be used to preliminarily judge the average diffusion of liver tissue.

[0115] B2, Std Dev (standard deviation)

[0116] Definition: Standard deviation is an indicator that measures the dispersion of pixel signal intensity values within a ROI.

[0117] Function: Indicates the degree of dispersion of ADC values. A higher standard deviation may indicate the presence of uneven diffusion restriction within the tissue, which may be related to the uneven degree of inflammation.

[0118] B3. Variance

[0119] Definition: Variance is the square of the standard deviation and measures the dispersion of signal strength values.

[0120] Effect: The larger the variance, the greater the fluctuation in signal intensity and the higher the tissue heterogeneity may be.

[0121] B4. Skewness

[0122] Definition: Skewness describes the symmetry of the signal strength distribution.

[0123] Effect: A skewness of 0 indicates a symmetrical distribution, a positive skewness indicates a longer right tail, and a negative skewness indicates a longer left tail.

[0124] B5, Kurtosis

[0125] Definition: Kurtosis describes the sharpness of the signal strength distribution.

[0126] Effect: The larger the kurtosis, the sharper the distribution and the signal strength is concentrated near the mean; the smaller the kurtosis, the flatter the distribution and the dispersed the signal strength.

[0127] B6. Entropy

[0128] Definition: Entropy measures the complexity of the signal strength distribution.

[0129] Effect: The greater the entropy, the more complex the signal intensity distribution is and the higher the tissue heterogeneity may be.

[0130] These characteristic parameters describe the statistical characteristics of the signal intensity within the ROI from different perspectives, providing rich quantitative information for imaging omics analysis and helping to more comprehensively evaluate the morphological and functional changes of tissues.

[0131] In addition, in the embodiment of the present application, the first-order statistical parameters also include:

[0132] B7. Inhomogeneity is a data definition used to indicate whether the grayscale distribution in the ADC histogram is uniform: Inhomogeneity measures the uniformity of the signal intensity distribution.

[0133] Effect: The greater the unevenness, the more uneven the signal intensity distribution is, and the higher the tissue heterogeneity may be.

[0134] B7 (Inhomogeneity) is set in the first-order statistical parameters. Its main advantages are:

[0135] High sensitivity: It is sensitive to small changes in grayscale distribution in images and can effectively detect early lesions of organs such as the liver.

[0136] Quantitative heterogeneity: Provides objective quantitative indicators of tissue heterogeneity to assist doctors in determining the nature and extent of lesions.

[0137] Auxiliary diagnosis: By analyzing the differences in grayscale distribution, it helps to distinguish different types of liver diseases, such as inflammation, fibrosis, and tumors.

[0138] Comprehensiveness: Combined with other first-order statistical parameters, it comprehensively describes the grayscale distribution characteristics of the image and improves diagnostic accuracy.

[0139] In radiomics, although nonuniformity and inhomogeneity have similar names, there are some differences in their specific definitions and applications.

[0140] 1. Nonuniformity

[0141] Definition: Nonuniformity is often used to describe the uneven distribution of pixel values in an image. In texture analysis, it measures whether the distribution of different grayscale values in an image is uniform. For example, in a gray-level co-occurrence matrix (GLCM), nonuniformity can measure whether the frequencies of different grayscale value pairs are uniform.

[0142] Calculation method: Nonuniformity can be measured by calculating the variance or entropy of the frequency distribution of grayscale values. For example, if the distribution of grayscale values is very concentrated, the nonuniformity value will be low; if the distribution is dispersed, the nonuniformity value will be high.

[0143] Application: Nonuniformity is used in texture analysis to describe image heterogeneity, and is particularly useful in evaluating the internal structure of lesions such as tumors. For example, internal heterogeneity in a tumor may indicate a higher degree of malignancy.

[0144] 2. Inhomogeneity

[0145] Definition: Inhomogeneity is often used to describe the overall grayscale inhomogeneity in an image. It focuses more on whether there are areas with obvious grayscale differences in the image, such as plaques and nodules.

[0146] Calculation method: Inhomogeneity can be measured by calculating statistics such as the standard deviation and variance of the grayscale values of the image. For example, images with a larger standard deviation generally have a higher inhomogeneity.

[0147] Application: In radiomics, inhomogeneity is used to assess the overall inhomogeneity of images. It is particularly useful in assessing lesions such as inflammation and fibrosis in organs like the liver. For example, liver inflammation can lead to an uneven distribution of grayscale values in liver tissue, and inhomogeneity can quantify this inhomogeneity.

[0148] As can be seen from the above, although nonuniformity and inhomogeneity are both used to describe image nonuniformity, they differ in their definitions, calculation methods, and application scenarios. Nonuniformity focuses more on the distribution characteristics of pixel values, while inhomogeneity focuses more on the overall nonuniformity of the image.

[0149] In magnetic resonance imaging, especially in diffusion-weighted imaging, the b-value is a key parameter that represents the product of the strength and duration of the applied diffusion sensitivity gradient. The unit of the b-value is seconds per square millimeter (s / mm 2 ). It is used to control and quantify the effect of water molecule diffusion within tissue on image signal intensity.

[0150] In the embodiment of the present application, the parameter b in the DWI is greater than a preset high threshold.

[0151] Specifically, when using low b values (usually less than 200s / mm2 ), the measured tissue signal attenuation is mostly due to the perfusion effect of microcirculatory capillaries, rather than just the diffusion of water molecules. This is because low b values are less sensitive to the diffusion of water molecules and are more susceptible to the influence of microcirculatory blood flow. In this case, the signal attenuation mainly reflects the perfusion of the tissue, rather than the simple diffusion of water molecules. Therefore, the embodiments of the present application use high b values (usually greater than 800s / mm 2 ) can improve the accuracy and diagnostic value of DWI because it can reduce the influence of perfusion effect and more accurately reflect the diffusion of water molecules in tissues.

[0152] In practical applications, the image scanning operation can be scheduled on the same day as liver puncture, which not only increases the accuracy and comprehensiveness of the research parameters, but also makes the research results more realistic and credible, and has higher application value for clinical diagnosis and treatment.

[0153] Among them, the main considerations for arranging the scan on the day of puncture are as follows:

[0154] 1. Ensure data consistency and accuracy: By performing MRI scans and liver puncture on the same day, the changes in liver inflammation status that may be caused by the time interval can be minimized, thereby ensuring the correspondence between imaging examination results and pathological examination results, and improving the reliability and accuracy of research data.

[0155] 2. Facilitate comparative analysis: Scheduling both examinations on the same day facilitates direct comparison and analysis of imaging assessment results, such as DWI and texture features, with pathological findings obtained by puncture. This helps to more accurately assess the diagnostic efficacy of imaging methods in grading liver inflammation and identify more effective assessment indicators.

[0156] 3. Improved research efficiency: For subjects, completing both tests on the same day can reduce the inconvenience and burden of multiple trips to the hospital, improving subject compliance and study efficiency. It also helps the research team obtain complete data more quickly, advancing the research process.

[0157] 4. Increase the scientificity and credibility of the research: This arrangement meets the requirements for concurrent controls in scientific research, making the research results more scientific and credible. By directly comparing with pathological results, the effectiveness of imaging assessment methods can be more objectively verified, providing stronger evidence for subsequent clinical application.

[0158] In summary, arranging the scan on the day of puncture is to ensure the quality and reliability of research data, facilitate comparative analysis, improve research efficiency and scientificity, and better achieve the research goals of this project in non-invasive grading assessment of liver inflammation.

[0159] Figure 2 This is a schematic diagram of the structure of the image processing device provided in the embodiment of the present application. Figure 2 As shown, the device includes:

[0160] A receiving module 201 is configured to receive an image to be tested, wherein the image to be tested is obtained by performing a magnetic resonance imaging scan of a subject's liver;

[0161] A first determination module 202 is configured to determine the field strength used by the image to be measured to obtain a target field strength;

[0162] The selection module 203 is configured to select a value of a target feature corresponding to the target field strength from the feature data of the image to be measured to obtain a value to be measured, wherein the target feature satisfies a preset high diagnostic efficiency condition;

[0163] The second determination module 204 is configured to determine the grade of the liver inflammation of the subject according to the threshold value corresponding to the target field strength of the target feature.

[0164] The device provided in the embodiment of the present application obtains the target field strength of the image to be tested, selects target features that meet the conditions of high diagnostic efficiency, and accurately determines the grading results of liver inflammation based on the threshold value corresponding to the target feature at the target field strength, thereby improving diagnostic efficiency and accuracy.

[0165] In addition, an embodiment of the present application further provides a storage medium, in which a computer program is stored, wherein the computer program is configured to execute the method described above when running.

[0166] An embodiment of the present application further provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute the method described above.

[0167] It will be appreciated by those skilled in the art that all or some of the steps, systems, and functional modules / units in the methods disclosed above may be implemented as software, firmware, hardware, and appropriate combinations thereof. In hardware implementations, the division between the functional modules / units mentioned in the above description does not necessarily correspond to the division of physical components; for example, a physical component may have multiple functions, or a function or step may be performed by several physical components in cooperation. Some or all components may be implemented as software executed by a processor, such as a digital signal processor or a microprocessor, or implemented as hardware, or implemented as an integrated circuit, such as an application-specific integrated circuit. Such software may be distributed on a computer-readable medium, which may include a computer storage medium (or non-transitory medium) and a communication medium (or temporary medium). As is well known to those skilled in the art, the term "computer storage medium" includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to store the desired information and can be accessed by a computer. In addition, it is well known to those skilled in the art that communication media generally embodies computer-readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transport mechanism, and may include any information delivery media.

Claims

1. An image processing method, comprising: receiving an image to be tested, wherein the image to be tested is obtained by performing a magnetic resonance imaging scan of a subject's liver; Determining the field strength used by the image to be measured to obtain a target field strength; Selecting a value of a target feature corresponding to the target field strength from the feature data of the image to be measured to obtain a value to be measured, wherein the target feature satisfies a preset condition of high diagnostic efficacy; The grading result of the liver inflammation of the test subject is determined according to the threshold value of the target feature corresponding to the target field strength.

2. The method according to claim 1, characterized in that Methods for obtaining the target features and their thresholds include: Acquiring a sample data set, wherein the sample data set includes at least two sample images for each subject, wherein the at least two sample images of the same subject are obtained by performing magnetic resonance imaging scans of the subject's liver under the same conditions of various imaging parameters and different field intensities, wherein each subject is grouped into a control group or a case group; Obtaining feature data of sample images of all subjects under the same field intensity; The target features and their thresholds at each field strength are determined based on the group to which each subject belongs and the feature data of all subjects at the same field strength.

3. The method according to claim 2, characterized in that The method further comprises: Acquiring a validation data set, wherein the validation data set includes at least two sample images of a verifier, wherein at least two images of the same verifier are obtained by performing magnetic resonance imaging scans of the verifier's liver under the same imaging parameters and different field intensities; Obtain feature data for each image of each verifier; According to the target features and their thresholds under different field intensities, the grading results of liver inflammation of each verifier under different field intensities are obtained, and the grading results to be verified of each verifier are obtained; Based on the baseline grading results of each verifier under the same field strength and the grading results to be verified, the accuracy of the target features corresponding to each field strength is determined, wherein the target features with an accuracy greater than a preset accuracy threshold participate in the processing of the image of the detector.

4. The method according to any one of claims 1 to 3, characterized in that: The threshold value of the target feature is the threshold value of the target manufacturer corresponding to the target field strength, wherein the target manufacturer is the manufacturer of the device that generates the image.

5. The method according to any one of claims 1 to 3, characterized in that The feature data includes texture feature data extracted based on the image and primary statistical data extracted based on an apparent diffusion coefficient ADC histogram corresponding to the image.

6. The method according to claim 5, characterized in that: The texture feature data includes the number of patches and the standard deviation of the patches; The primary statistical data includes characteristic data for indicating whether the grayscale distribution in the ADC histogram is uniform.

7. The method according to claim 1, characterized in that The image is a diffusion weighted image (DWI), wherein a parameter b in the DWI is greater than a preset high threshold, wherein the parameter b represents the product of the intensity of the applied diffusion sensitivity gradient and time.

8. An image processing device, comprising: A receiving module configured to receive an image to be tested, wherein the image to be tested is obtained by performing a magnetic resonance imaging scan of a subject's liver; A first determining module is configured to determine the field strength used by the image to be measured to obtain a target field strength; A selection module is configured to select a value of a target feature corresponding to the target field strength from the feature data of the image to be measured to obtain a value to be measured, wherein the target feature satisfies a preset condition of high diagnostic efficiency; The second determination module is configured to determine the grade result of the liver inflammation of the test subject according to the threshold value corresponding to the target field strength of the target feature.

9. A storage medium, characterized in that: The storage medium stores a computer program, wherein the computer program is configured to execute the method according to any one of claims 1 to 7 when executed.

10. An electronic device comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to run the computer program to perform the method according to any one of claims 1 to 7.

Images