A coronary artery analysis processing method and system based on magnetic resonance CNR value

Abstract

The present application relates to the technical field of medical data analysis, and particularly relates to a coronary artery analysis processing method and system based on magnetic resonance CNR value, which comprises the following steps: S1: collecting coronary image data, obtaining pretreatment data after pretreating the collected coronary image data, inputting the pretreatment data into an ROI identification model to obtain a target ROI; S2: using an MRCA method based on the target ROI to obtain an MRCA image of the target ROI, and using a final dynamic adaptability formula to obtain a CNR value of a correlation reference region in the MRCA image. The present application improves the efficiency and accuracy of target region extraction by using the image enhancement and region extraction technology combining the edge preserving normal gradient vector flow with the active contour model, and improves the work efficiency of radiologists by using the method combining the MRCA image and the CNR value to diagnose the coronary artery stenosis degree.

Description

Technical Field

[0001] This invention relates to the field of medical data analysis technology, and in particular to a method and system for coronary artery analysis and processing based on magnetic resonance CNR values. Background Technology

[0002] Magnetic resonance coronary angiography (MRCA) is a non-invasive, radiation-free, and non-contrast-enhanced coronary angiography technique, highly suitable for early screening of coronary heart disease and possessing significant clinical guiding value. With advancements in MRCA hardware technology, its imaging quality has improved year by year, clearly displaying the main coronary arteries and proximal branches. This provides a safe and reliable means for non-invasive examination of coronary heart disease, enabling early screening and accurate diagnosis. It is of great practical significance for preventing fatal sudden cardiac death and for implementing effective scientific treatment.

[0003] However, MRCA technology has problems such as relatively complex operation, low success rate and low image spatial resolution. Moreover, MRCA currently relies solely on the subjective judgment of the diagnostic physician to determine the degree of coronary artery stenosis. This method has problems of strong subjectivity, low accuracy and large variability. Furthermore, MRCA is an magnetic resonance imaging examination. If the subjective judgment of the reduction in the diameter of the coronary artery lumen and the decrease in signal intensity is used alone, it is impossible to accurately diagnose the degree of coronary artery stenosis. Therefore, it is very necessary to relieve the workload of radiologists in reconstructing images and assist them in completing image diagnosis work objectively and efficiently. Summary of the Invention

[0004] To overcome the inability to accurately diagnose the degree of coronary artery stenosis, this invention provides a coronary artery analysis and processing method and system based on magnetic resonance CNR values.

[0005] Technical solution: A method for coronary artery analysis and processing based on magnetic resonance CNR values, comprising the following steps:

[0006] S1: Collect coronary artery image data, preprocess the collected coronary artery image data to obtain preprocessed data, input the preprocessed data into the ROI recognition model to obtain the target ROI;

[0007] S2: Based on the target ROI, use the MRCA method to obtain the MRCA image of the target ROI, and use the final dynamic adaptive formula to obtain the CNR value of the associated reference region in the MRCA image;

[0008] S3: Obtain the comprehensive CNR value using a weighted average method based on the CNR values ​​of the associated reference regions in the MRCA image;

[0009] S4: Perform stenosis ratio characteristic analysis based on the comprehensive CNR value to obtain the stenosis data of the corresponding cross-section.

[0010] Preferably, the process of collecting coronary artery image data, preprocessing the collected coronary artery image data to obtain a preprocessed coronary artery image, and inputting the preprocessed coronary artery image into an ROI recognition model to obtain a target ROI includes: using multi-plane imaging and multi-angle reconstruction methods based on minimizing artifacts and anatomical coverage criteria to screen the collected coronary artery image data to obtain a standard coronary artery image, performing data augmentation and contrast enhancement preprocessing on the standard coronary artery image to obtain a preprocessed coronary artery image, and inputting the preprocessed coronary artery image into an ROI recognition model to obtain a target ROI.

[0011] Preferably, the step of inputting the preprocessed coronary artery image into the ROI recognition model to obtain the target ROI includes: the ROI recognition model uses a method combining edge-preserving normal gradient vector flow and an active contour model to extract the cardiac region. Specifically, it trains a convolutional neural network model using historical coronary artery image data, enhances the preprocessed coronary artery image using edge-preserving normal gradient vector flow to obtain an enhanced image, and determines the target contour of the image by minimizing the energy functional based on the enhanced image, removing tissues of non-interest of interest; performs coronary artery segmentation based on a regularized flow structure network with affine coupling layers, iteratively trains the network parameters using a gradient descent optimization algorithm using manually labeled coronary artery data; detects the vessel centerline based on the Kalman filter algorithm, and corrects and optimizes the automatic segmentation results by combining a prior knowledge-driven vessel identification re-identification program, finally obtaining the target ROI.

[0012] Preferably, the step of using the MRCA method to obtain an MRCA image of the target ROI based on the target ROI, and using the final dynamic adaptive formula to obtain the CNR value of the associated reference region in the MRCA image, includes: selecting the normal vessel segment closest to the lesion but unaffected as the proximal reference point, calculating the proximal reference point based on the MRCA image and the final dynamic adaptive formula to obtain the CNR value of the proximal reference lumen of the lesion; selecting the normal vessel segment farthest from the lesion but unaffected as the distal reference point, calculating the distal reference point based on the MRCA image and the final dynamic adaptive formula to obtain the CNR value of the distal reference lumen of the lesion.

[0013] Preferably, the step of obtaining an MRCA image of the target ROI using the MRCA method and obtaining the CNR value of the associated reference region in the MRCA image using a final dynamic adaptive formula includes: obtaining relevant data of the target patient, relevant data of the proximal reference point, and relevant data of the distal reference point, and adjusting the final dynamic adaptive formula according to the relevant data of the target patient, wherein the final dynamic adaptive formula is:

[0014] ;

[0015] In the formula, The cross-sectional contrast noise ratio should be measured. To measure the signal intensity within the lumen of a coronary artery segment cross-section; This represents the absolute value of the signal intensity from the chest wall muscles. The standard deviation of the air background noise signal intensity; This is a dynamic adjustment factor used to adjust the CNR calculation based on individual patient characteristics.

[0016] Preferably, the step of acquiring relevant data of the target patient, relevant data of the proximal reference point, and relevant data of the distal reference point, and adjusting the final dynamic adaptive formula based on the relevant data of the target patient, includes: the relevant data of the target patient includes the target patient's age, weight, and underlying disease data; and a dynamic adjustment factor is obtained by using a dynamic adjustment formula based on the relevant data of the target patient, wherein the dynamic adjustment formula is:

[0017] ;

[0018] In the formula, For dynamic adjustment factors; This is the weighting adjustment coefficient; The age of the target patient; The average age of the target population; The target patient's weight; The average weight of the target population; For the target patient Binary markers for diseases; For the first Weight of each disease.

[0019] Preferably, obtaining the composite CNR value using a weighted average method based on the CNR values ​​of the associated reference regions in the MRCA image includes: obtaining the composite CNR value using a comprehensive calculation formula based on the CNR values ​​of the proximal reference lumen and the distal reference lumen of the lesion, wherein the comprehensive calculation formula is:

[0020] ;

[0021] In the formula, The overall CNR value; For dynamic weighting factors; The reference lumen CNR value is the value of the lumen proximal to the lesion. The reference lumen CNR value is used for the distal end of the lesion.

[0022] Preferably, the step of performing stenosis ratio feature analysis based on the comprehensive CNR value to obtain the stenosis degree data of the corresponding cross section includes: obtaining the stenosis ratio based on the comprehensive CNR value and the normal lumen CNR value; classifying the stenosis degree of the cross section into a preset number of levels based on the stenosis ratio; and determining the reliability of the stenosis rate formula through plotting statistical analysis.

[0023] Preferably, the method of determining the reliability of the stenosis rate formula through plotted point statistical analysis includes: analyzing the mean, variance, and skewness of the CNR value distribution on different segments to evaluate the applicability of the formula; and analyzing the classification ability of the comprehensive CNR value to predict the stenosis rate through ROC curve analysis, and calculating the AUC value to verify the performance of the formula.

[0024] Preferably, a coronary artery analysis and processing system based on magnetic resonance CNR values ​​includes:

[0025] The data acquisition and preprocessing module is used to collect coronary artery image data, screen valid images for standardization, and use multi-section imaging and multi-angle reconstruction methods to screen and obtain standard coronary artery images.

[0026] The region of interest identification module is used to segment the cardiac region and coronary arteries using convolutional neural networks and active contour models, combined with edge-preserving normal gradient vector flow.

[0027] The CNR value calculation module is used to obtain the CNR value of the associated reference region in the MRCA image using a dynamic adjustment formula.

[0028] The comprehensive CNR analysis module is used to obtain a comprehensive CNR value based on the CNR values ​​of the associated reference regions in the MRCA image using a weighted average method.

[0029] The stenosis analysis module is used to analyze and classify the stenosis ratio characteristics of the target segment based on the comprehensive CNR value, and to determine the reliability of the stenosis rate formula.

[0030] The beneficial effects of this invention are:

[0031] 1. This invention improves the accuracy of region of interest identification by employing image enhancement and region extraction techniques that combine edge-preserving normal gradient vector flow with an active contour model. At the same time, it improves the efficiency and accuracy of target region extraction by using an affine coupling regularized network to accurately segment coronary arteries.

[0032] 2. This system introduces dynamic adjustment factors and combines them with the patient's individual characteristics, such as age, weight and underlying diseases, to dynamically optimize the CNR calculation, thereby improving the personalized adaptability of the CNR value calculation and enabling it to more accurately reflect the coronary artery stenosis characteristics of different patients.

[0033] 3. By using a weighted average formula to comprehensively calculate the CNR values ​​of the proximal and distal reference regions, a comprehensive analysis of the degree of coronary artery stenosis is ensured, the bias caused by a single CNR value is eliminated, and the stability of stenosis ratio analysis is improved.

[0034] 4. The system calculates the AUC value by statistically analyzing the distribution characteristics of CNR values ​​through plotted points and ROC curve analysis, which fully verifies the classification performance and applicability of the narrowing rate formula and effectively improves the reliability and stability of narrowing rate analysis.

[0035] 5. As a non-invasive analytical method based on MRCA, it has no radiation risk and is easy to operate, reducing the risk and discomfort of patients during the diagnostic process, while avoiding complications caused by invasive procedures. Attached Figure Description

[0036] Figure 1 This is a flowchart of the coronary artery analysis and processing method based on magnetic resonance CNR values ​​according to the present invention;

[0037] Figure 2 This is a flowchart of the coronary artery analysis and processing system based on magnetic resonance CNR values ​​according to the present invention. Detailed Implementation

[0038] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0039] Example 1: A method for coronary artery analysis and processing based on magnetic resonance CNR values, such as... Figure 1 and Figure 2 As shown, it includes the following steps:

[0040] S1: Collect coronary artery image data, preprocess the collected coronary artery image data to obtain preprocessed data, input the preprocessed data into the ROI recognition model to obtain the target ROI;

[0041] Using multi-plane imaging and multi-angle reconstruction methods based on minimizing artifacts and anatomical coverage criteria, the collected coronary artery image data is filtered to obtain standard coronary artery images. The standard coronary artery images are then preprocessed with data augmentation and contrast enhancement to obtain preprocessed coronary artery images. These preprocessed coronary artery images are then input into an ROI recognition model to obtain target ROIs.

[0042] It should be noted that, using a magnetic resonance coronary angiography (MRCA) device, coronary artery image data of patients are acquired. Artifact detection is performed on each image, and images with artifacts exceeding a threshold, such as 20% signal change, are removed. Based on anatomical coverage criteria, clear imaging data of the main coronary artery region are prioritized. Translation, rotation, and symmetry transformations are applied to each image to generate additional data. Contrast stretching and histogram equalization are used to improve local vessel contrast. Finally, a preprocessed coronary artery image dataset is obtained.

[0043] The ROI recognition model uses a combination of edge-preserving normal gradient vector flow and an active contour model to extract the cardiac region. Specifically, it trains a convolutional neural network model using historical coronary artery image data, enhances the preprocessed coronary artery image using edge-preserving normal gradient vector flow, and obtains an enhanced image. Based on the enhanced image, it combines an active contour model to determine the target contour of the image by minimizing the energy functional and removing non-interest-bearing tissues. A regularized flow structure network based on affine coupling layers is used for coronary artery segmentation. Using manually labeled coronary artery data, a gradient descent-based optimization algorithm is used to iteratively train the network parameters. A Kalman filter algorithm is used to detect the vessel centerline, and a prior knowledge-driven vessel marker re-identification program is combined to correct and optimize the automatic segmentation results, finally obtaining the target ROI.

[0044] It should be noted that the preprocessed coronary artery image is enhanced using the edge-preserving GVF algorithm. This algorithm calculates the normal gradient vector flow for each pixel, preserving strong gradient regions, such as the vessel wall edges, while suppressing weak gradient regions. The resulting enhanced image exhibits clearer coronary artery boundary features. Based on the enhanced image, the following energy functional is minimized using an active contour model. In the formula, Image intensity energy, For contour smoothing energy, To balance the coefficients, the extracted results include the preliminary contours of the cardiac region and the main coronary artery. Affine-coupled regularized flow network is used to complete coronary artery segmentation. The network parameters are optimized by gradient descent. Manually labeled vessel contour data is used during training, and the resulting image contains a segmentation mask of the vessel region. Based on the Kalman filter algorithm, the position of the vessel centerline is detected and false segmentation is eliminated. A prior knowledge-driven re-identification procedure is used to correct errors in vessel branch identification. Finally, the target ROI is obtained, including complete coronary artery region and centerline information.

[0045] S2: Based on the target ROI, use the MRCA method to obtain the MRCA image of the target ROI, and use the final dynamic adaptive formula to obtain the CNR value of the associated reference region in the MRCA image;

[0046] The closest but unaffected normal vessel segment to the lesion is selected as the proximal reference point. The CNR value of the proximal reference lumen is calculated based on the MRCA image and the final dynamic adaptive formula. The farthest but unaffected normal vessel segment to the lesion is selected as the distal reference point. The CNR value of the distal reference lumen is calculated based on the MRCA image and the final dynamic adaptive formula.

[0047] It should be noted that, from the MRCA image, the normal vessel segment upstream of the lesion region that is closest to the lesion and unaffected by the lesion is selected as the proximal reference point; from the MRCA image, the normal vessel segment downstream of the lesion region that is farthest from the lesion and unaffected by the lesion is selected as the distal reference point. In this embodiment, one approach is to keep the distance from the proximal reference point to the lesion vessel segment equal to the distance from the distal reference point to the lesion vessel segment.

[0048] Acquire relevant data from the target patient, proximal reference points, and distal reference points, and adjust the final dynamic adaptive formula based on the target patient's data. The final dynamic adaptive formula is as follows:

[0049] ;

[0050] In the formula, The cross-sectional contrast noise ratio should be measured. To measure the signal intensity within the lumen of a coronary artery segment cross-section; This represents the absolute value of the signal intensity from the chest wall muscles. The standard deviation of the air background noise signal intensity; This is a dynamic adjustment factor used to adjust the CNR calculation based on individual patient characteristics.

[0051] The process of acquiring relevant data of the target patient, relevant data of the proximal reference point, and relevant data of the distal reference point, and adjusting the final dynamic adaptive formula based on the relevant data of the target patient, includes: the relevant data of the target patient includes the target patient's age, weight, and underlying disease data; and the dynamic adjustment factor is obtained by using the dynamic adjustment formula based on the relevant data of the target patient, wherein the dynamic adjustment formula is:

[0052] ;

[0053] In the formula, For dynamic adjustment factors; This is the weighting adjustment coefficient; The age of the target patient; The average age of the target population; The target patient's weight; The average weight of the target population; For the target patient Binary markers for diseases; For the first Weight of each disease.

[0054] It should be noted that, for example, the following personalized characteristics of the target patient are collected: age: 60 years old; weight: 75 kg; presence of underlying diseases: hypertension (yes), diabetes (no); the CNR formula is adjusted based on the target patient data to obtain the dynamic adjustment factor for the target patient; proximal reference point selection: proximal reference points of the target ROI are selected using MRCA images, and the signal intensity and background noise of the relevant region are measured; distal reference point selection: similarly, distal reference points of the target ROI are selected, and their signal intensity and background noise are measured.

[0055] S3: Obtain the comprehensive CNR value using a weighted average method based on the CNR values ​​of the associated reference regions in the MRCA image;

[0056] A comprehensive CNR value is obtained by using a combined calculation formula based on the CNR values ​​of the proximal reference lumen and the distal reference lumen of the lesion. The comprehensive calculation formula is as follows:

[0057] ;

[0058] In the formula, The overall CNR value; For dynamic weighting factors; The reference lumen CNR value is the value of the lumen proximal to the lesion. The reference lumen CNR value is used for the distal end of the lesion.

[0059] It should be noted that the dynamically adjusted CNR values ​​obtained above are used: that is, the CNR values ​​of the proximal reference lumen and the CNR values ​​of the distal reference lumen of the lesion, weighted according to the distance weight of the reference point in the lesion region, where... equal In this embodiment, the specific process of using the comprehensive calculation formula is as follows: The CNR value of the coronary artery lumen for each segment greater than or equal to 2 mm is measured; the location of the chest wall muscle tissue and the standard deviation of the air background noise signal intensity on the MRCA image data are determined; and the... Both are set to 0.5, meaning the comprehensive calculation formula becomes: Then, the comprehensive CNR value is obtained using the comprehensive calculation formula.

[0060] S4: Perform stenosis ratio characteristic analysis based on the comprehensive CNR value to obtain the stenosis data of the corresponding cross-section.

[0061] Based on the comprehensive CNR value and the normal lumen CNR value, a stenosis ratio is obtained. The degree of stenosis in the cross-section is divided into a preset number of levels based on the stenosis ratio. The reliability of the stenosis rate formula is then determined through plotting statistical analysis.

[0062] The mean, variance, and skewness of CNR values ​​in different segments were analyzed to evaluate the applicability of the formula. The classification ability of the comprehensive CNR value in predicting stenosis rate was analyzed by ROC curve analysis, and the AUC value was calculated to verify the performance of the formula.

[0063] It should be noted that the normal lumen CNR value is measured by selecting a lesion-free area in the healthy coronary artery segment of the target patient using MRCA images; the comprehensive CNR value is obtained according to the comprehensive calculation formula mentioned above. For example, the ratio of the comprehensive CNR value to the normal lumen CNR value is used as the stenosis ratio. According to the preset stenosis level standard, the degree of stenosis in the cross-section is divided into the following four levels: Level 0 (no stenosis): stenosis ratio greater than 0.9; Level 1 (minor stenosis): stenosis ratio greater than 0.75 and less than or equal to 0.9; Level 2 (mild stenosis): stenosis ratio greater than or equal to 0.51 and less than or equal to 0.75; Level 3 (moderate stenosis): stenosis ratio greater than or equal to 0.31 and less than or equal to 0.5; Level 4 (severe stenosis): stenosis ratio greater than or equal to 0.01 and less than or equal to 0.3. Grade 5 (Occlusion): Stenosis ratio equals 0; The reliability of the stenosis rate formula was analyzed using plotted statistics and ROC curves. Plotted statistics analysis involved segment division: the coronary artery was divided into three segments: proximal, mid-segment, and distal; statistical parameter extraction: the mean, variance, and skewness coefficient of the CNR value distribution were calculated for each segment; based on the mean and variance, the CNR value distribution in the proximal segment was more concentrated and the data skewness was smaller, resulting in better applicability of the formula; while the distal segment showed increased skewness, limiting the predictive ability of the formula; ROC curve analysis involved: real data collection: actual clinical coronary artery stenosis rate classification data were paired with the comprehensive CNR value to generate classification samples; ROC curve plotting: the comprehensive CNR value was used as input to the classifier, sensitivity and specificity were calculated, and the ROC curve was plotted; AUC value calculation: the AUC value was calculated based on the ROC curve.

[0064] Example 2: Based on Example 1, a coronary artery analysis and processing system based on magnetic resonance CNR values ​​includes:

[0065] The data acquisition and preprocessing module is used to collect coronary artery image data, screen valid images for standardization, and use multi-section imaging and multi-angle reconstruction methods to screen and obtain standard coronary artery images.

[0066] The region of interest identification module is used to segment the cardiac region and coronary arteries using convolutional neural networks and active contour models, combined with edge-preserving normal gradient vector flow.

[0067] The CNR value calculation module is used to obtain the CNR value of the associated reference region in the MRCA image using a dynamic adjustment formula.

[0068] The comprehensive CNR analysis module is used to obtain a comprehensive CNR value based on the CNR values ​​of the associated reference regions in the MRCA image using a weighted average method.

[0069] The stenosis analysis module is used to analyze and classify the stenosis ratio characteristics of the target segment based on the comprehensive CNR value, and to determine the reliability of the stenosis rate formula.

[0070] It should be understood that this embodiment is for illustrative purposes only and is not intended to limit the scope of the invention. Furthermore, it should be understood that after reading the teachings of this invention, those skilled in the art can make various alterations or modifications to the invention, and these equivalent forms also fall within the scope defined by the appended claims.

Claims

1. A method for coronary artery analysis and processing based on magnetic resonance CNR values, characterized in that, Includes the following steps: S1: Collect coronary artery image data, preprocess the collected coronary artery image data to obtain preprocessed data, input the preprocessed data into the ROI recognition model to obtain the target ROI; S2: Based on the target ROI, use the MRCA method to obtain the MRCA image of the target ROI, and use the final dynamic adaptive formula to obtain the CNR value of the associated reference region in the MRCA image; S3: Obtain the comprehensive CNR value using a weighted average method based on the CNR values ​​of the associated reference regions in the MRCA image; S4: Perform stenosis ratio characteristic analysis based on the comprehensive CNR value to obtain stenosis degree data for the corresponding cross-section; the corresponding cross-section is the cross-section of the lesion area within the target ROI; The process involves using the MRCA method based on the target ROI to obtain an MRCA image of the target ROI, and then using a final dynamic adaptive formula to obtain the CNR value of the associated reference region in the MRCA image. This includes: obtaining relevant data of the target patient, relevant data of the proximal reference point, and relevant data of the distal reference point, and adjusting the final dynamic adaptive formula based on the relevant data of the target patient. The final dynamic adaptive formula is as follows: ； In the formula, To measure the contrast-to-noise ratio of coronary artery segment cross-sections; To measure the signal intensity within the lumen of a coronary artery segment cross-section; This represents the absolute value of the signal intensity from the chest wall muscles. The standard deviation of the air background noise signal intensity; To adjust the dynamic adjustment factor for CNR calculation based on individual patient characteristics; The process of acquiring relevant data of the target patient, relevant data of the proximal reference point, and relevant data of the distal reference point, and adjusting the final dynamic adaptive formula based on the relevant data of the target patient, includes: the relevant data of the target patient includes the target patient's age, weight, and underlying disease data; and the dynamic adjustment factor is obtained by using the dynamic adjustment formula based on the relevant data of the target patient, wherein the dynamic adjustment formula is: ； In the formula, For dynamic adjustment factors; This is the weighting adjustment coefficient; The age of the target patient; The average age of the target population; The target patient's weight; The average weight of the target population; For the target patient A binary label for a disease; D = ( , () is a set of diseases of the target patients; For the first The weight of each disease; The step of using the MRCA method based on the target ROI to obtain the MRCA image of the target ROI and using the final dynamic adaptive formula to obtain the CNR value of the associated reference region in the MRCA image includes: selecting the normal vessel segment closest to the lesion but unaffected as the proximal reference point, calculating the proximal reference point based on the MRCA image and the final dynamic adaptive formula to obtain the CNR value of the proximal reference lumen of the lesion; selecting the normal vessel segment farthest from the lesion but unaffected as the distal reference point, calculating the distal reference point based on the MRCA image and the final dynamic adaptive formula to obtain the CNR value of the distal reference lumen of the lesion. The step of obtaining the comprehensive CNR value using a weighted average method based on the CNR values ​​of the associated reference regions in the MRCA image includes: obtaining the comprehensive CNR value using a comprehensive calculation formula based on the CNR values ​​of the proximal reference lumen and the distal reference lumen of the lesion, wherein the comprehensive calculation formula is: ; In the formula, The overall CNR value; For dynamic weighting factors; The reference lumen CNR value is the value of the lumen proximal to the lesion. The reference lumen CNR value is used for the distal end of the lesion.

2. The method for coronary artery analysis and processing based on magnetic resonance CNR values ​​according to claim 1, characterized in that, The process of collecting coronary artery image data, preprocessing the collected coronary artery image data to obtain preprocessed coronary artery images, and inputting the preprocessed coronary artery images into an ROI recognition model to obtain target ROIs includes: using multi-plane imaging and multi-angle reconstruction methods based on minimizing artifacts and anatomical coverage criteria to filter the collected coronary artery image data to obtain standard coronary artery images, performing data augmentation and contrast enhancement preprocessing on the standard coronary artery images to obtain preprocessed coronary artery images, and inputting the preprocessed coronary artery images into an ROI recognition model to obtain target ROIs.

3. The method for coronary artery analysis and processing based on magnetic resonance CNR values ​​according to claim 2, characterized in that, The process of inputting the preprocessed coronary artery image into the ROI recognition model to obtain the target ROI includes: the ROI recognition model extracting the cardiac region using a combination of edge-preserving normal gradient vector flow and an active contour model; specifically, training a convolutional neural network model using historical coronary artery image data, enhancing the preprocessed coronary artery image using edge-preserving normal gradient vector flow to obtain an enhanced image, determining the target contour of the image based on the enhanced image and the active contour model by minimizing the energy functional, and removing tissues of non-interest of interest; performing coronary artery segmentation based on a regularized flow structure network with affine coupling layers, iteratively training the network parameters using manually labeled coronary artery data and a gradient descent-based optimization algorithm; detecting the vessel centerline based on a Kalman filter algorithm, and correcting and optimizing the automatic segmentation results using a prior knowledge-driven vessel identification re-identification program, finally obtaining the target ROI.

4. The method for coronary artery analysis and processing based on magnetic resonance CNR values ​​according to claim 1, characterized in that, The step of performing stenosis ratio feature analysis based on the comprehensive CNR value to obtain stenosis data of the corresponding cross-section includes: obtaining a stenosis ratio based on the comprehensive CNR value and the normal lumen CNR value; classifying the stenosis of the cross-section into a preset number of levels based on the stenosis ratio; and determining the reliability of the stenosis rate formula through plotting statistical analysis.

5. The method for coronary artery analysis and processing based on magnetic resonance CNR values ​​according to claim 4, characterized in that, The method of determining the reliability of the stenosis rate formula through plotted statistical analysis includes: analyzing the mean, variance, and skewness of the CNR value distribution on different segments to evaluate the applicability of the formula; and analyzing the classification ability of the comprehensive CNR value to predict the stenosis rate through ROC curve analysis, and calculating the AUC value to verify the performance of the formula.

6. A coronary artery analysis and processing system based on magnetic resonance CNR values, and a coronary artery analysis and processing method based on magnetic resonance CNR values ​​according to any one of claims 1-5, characterized in that, include: The data acquisition and preprocessing module is used to collect coronary artery image data and screen valid images for standardized processing; Standard coronary artery images were obtained by screening using multi-plane imaging and multi-angle reconstruction methods. The region of interest identification module is used to segment the cardiac region and coronary arteries using convolutional neural networks and active contour models, combined with edge-preserving normal gradient vector flow. The CNR value calculation module is used to obtain the CNR value of the associated reference region in the MRCA image using a dynamic adjustment formula. The comprehensive CNR analysis module is used to obtain a comprehensive CNR value based on the CNR values ​​of the associated reference regions in the MRCA image using a weighted average method. The stenosis analysis module is used to analyze and classify the stenosis ratio characteristics of the target segment based on the comprehensive CNR value, and to determine the reliability of the stenosis rate formula.

Images