Diabetes kidney lesion risk detection method and system based on ultrasound

Abstract

The invention discloses an ultrasound-based diabetic renal lesion risk detection method and system, and relates to the technical field of ultrasonic detection.According to the system, through continuous scanning of a two-dimensional ultrasonic probe, an elastic imaging probe and a Doppler probe, the shape, structure, blood flow characteristics and cortex images of the kidney are obtained; an image processing technology is adopted for preprocessing, space registration and fusion, and a kidney area is automatically segmented; according to the extracted kidney tissue characteristics, calculating an atrophy risk coefficient and a fibrosis risk coefficient, and comparing with a standard threshold value to evaluate the kidney lesion risk; calculating a comprehensive risk coefficient in combination with blood glucose, urine protein and diet data, and evaluating diet influence risks; a renal lesion risk assessment model is constructed by using a convolutional neural network and a machine learning algorithm, a renal lesion prediction function is generated, and the system can accurately assess the renal lesion risk of a diabetic patient and provide early diagnosis and personalized treatment suggestions.

Description

Technical Field

[0001] The present invention relates to the technical field of ultrasonic detection, and specifically to a method and system for detecting the risk of kidney lesions in diabetic patients based on ultrasound. Background Art

[0002] Diabetes is a common chronic metabolic disease. Long-term hyperglycemia can lead to kidney lesions, such as kidney atrophy and kidney fibrosis. However, there are multiple problems with current methods for monitoring kidney lesions in diabetic patients. Traditional methods for detecting kidney lesions, such as renal function tests, CT or MRI examinations, although they can provide relatively detailed information, are costly, complex to operate, and usually involve radiation, posing certain risks to the health of patients. In addition, the detection cycle of these methods is long, real-time monitoring cannot be achieved, and there is a lack of sufficient dynamic data, resulting in the inability to detect early signs of kidney lesions in a timely manner.

[0003] In clinical practice, existing ultrasonic image processing technologies have not been able to effectively combine multi-modal images and the physiological data of patients, especially lacking precise evaluation criteria in the early detection of kidney atrophy and kidney fibrosis. Even high-resolution ultrasonic images are difficult to automatically identify and quantify minor kidney lesions in a short time, lacking real-time performance and high efficiency, which limits their wide application. Summary of the Invention

[0004] In view of the deficiencies of the prior art, the present invention provides a method and system for detecting the risk of kidney lesions in diabetic patients based on ultrasound to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention is realized through the following technical solutions: A method for detecting the risk of kidney lesions in diabetic patients based on ultrasound, comprising the following steps:

[0006] S1. Set a two-dimensional ultrasound probe, an elastography probe, and a Doppler probe in the kidney area of the patient. The probes obtain the kidney tissue by continuous scanning, and collect ultrasonic images of the kidney morphology, structure, kidney blood flow characteristics, and kidney cortex;

[0007] S2. Use image processing technology to preprocess the ultrasonic images. After image processing, perform spatial registration on the ultrasonic images and then fuse them, and automatically segment the fused kidney area into several regions;

[0008] S3. By extracting the kidney tissue area Ss of the i-th kidney region i the kidney cyst tissue area Snz i the roundness Sly of the kidney contour i the smoothness Slp of the kidney contour i the curvature Sql of the kidney contour iand the texture features Swl of the kidney i , calculate the atrophy risk coefficient Wsfx of the kidney, compare it with the first standard threshold, evaluate whether the diabetic patient has the risk of kidney atrophy, and generate a kidney atrophy risk data report;

[0009] S4. By extracting the renal cortex thickness Cor of the i-th kidney region i , the hardness ZYD of the kidney tissue i , the blood perfusion rate BLG i , calculate the renal fibrosis risk coefficient XWHfx, compare it with the second standard threshold, evaluate whether the diabetic patient has the risk of renal fibrosis, and generate a renal fibrosis risk data report;

[0010] S5. By collecting the blood glucose value XTZ, urine protein value NDB of the diabetic patient and the dietary characteristics of the patient, calculate the comprehensive risk coefficient ZHXS, compare it with the third standard threshold, evaluate whether the daily diet of the diabetic patient has an impact risk on kidney lesions, and generate a dietary impact risk data report;

[0011] S6. Establish a kidney lesion risk assessment model, train and test the model, use a machine learning algorithm to generate a prediction function, and predict the risk of kidney lesions in diabetic patients.

[0012] Preferably, step S1 includes:

[0013] In the kidney region of the diabetic patient, set a two-dimensional ultrasound probe for collecting ultrasound images of the basic shape of the kidney and ultrasound images of the internal tissue structure; set an elastography probe for collecting ultrasound images of the hardness of the kidney tissue and ultrasound images of the renal cortex thickness; set a Doppler probe for collecting ultrasound images of the renal blood flow characteristics.

[0014] Preferably, step S2 includes:

[0015] S21. Use wavelet denoising technology to denoise the collected ultrasound images, decompose the data information in several tissue layer images and kidney structure layout images through wavelet transform, remove the noise components in the wavelet domain, and use histogram equalization to adjust the brightness, stretch the image range, and adjust the contrast of the image;

[0016] S22. Perform spatial registration on the ultrasound images so that different modality images can be aligned in the same coordinate system, and then perform fusion. Automatically segment the fused kidney region into several regions, and label them as regions L1, L2, L3,..., Ln respectively, where n represents the total number of regions.

[0017] Preferably, step S3 includes:

[0018] S31. By extracting the renal tissue area Ss of the i-th renal region i , the renal cyst tissue area Snz i , the roundness Sly of the renal contour i , the smoothness Slp of the renal contour i , the curvature Sql of the renal contour i and the texture feature Swl of the kidney i , after dimensionless processing, calculate and obtain the atrophy risk coefficient Wsfx of the kidney, and the formula is as follows:

[0019]

[0020] In the formula, n represents the total number of renal regions, S normal represents the reference area of a healthy kidney, R iderl represents the ideal renal contour feature, Swl mean represents the texture mean inside the kidney, Swl total represents the texture mean inside a healthy kidney, and w1, w2, and w3 represent weight coefficients;

[0021] S32. By presetting the first standard threshold Q1 in advance and comparing and analyzing the atrophy risk coefficient Wsfx of the kidney with the first standard threshold Q1, obtain the first evaluation result, including:

[0022] When the atrophy risk coefficient Wsfx of the kidney < the first standard threshold Q1, it indicates that the kidney of the diabetic patient has no atrophy risk, and continuous monitoring is carried out;

[0023] When the atrophy risk coefficient Wsfx of the kidney ≥ the first standard threshold Q1, it indicates that the kidney of the diabetic patient has an atrophy risk, trigger the first warning instruction, and generate a kidney atrophy risk data report.

[0024] Preferably, step S4 includes:

[0025] S41. By extracting the renal cortex thickness Cor of the i-th renal region i , the hardness ZYD of the renal tissue i , the blood perfusion rate BLG i , after dimensionless processing, calculate and obtain the renal fibrosis risk coefficient XWHfx, and the formula is as follows:

[0026]

[0027] In the formula, Cor normal represents the cortex thickness of a healthy kidney, ZYD iderl represents the ideal hardness of the renal tissue, BLG normal represents the blood perfusion rate of a healthy kidney, and w4, w5, and w6 represent weight coefficients;

[0028] S42. By presetting a second standard threshold Q2 in advance and comparing and analyzing the renal fibrosis risk coefficient XWHfx with the second standard threshold Q2, a second evaluation result is obtained, including:

[0029] When the renal fibrosis risk coefficient XWHfx < the second standard threshold Q2, it indicates that the kidneys of diabetic patients have no fibrosis risk, and continuous monitoring is carried out;

[0030] When the renal fibrosis risk coefficient XWHfx ≥ the second standard threshold Q2, it indicates that the kidneys of diabetic patients have a fibrosis risk, triggering a first warning instruction and generating a renal fibrosis risk data report.

[0031] Preferably, step S5 includes:

[0032] S51. By collecting the blood glucose value XTZ, urine protein value NDB of diabetic patients and the dietary characteristics of the patients, after dimensionless processing, the comprehensive risk coefficient ZHXS is calculated and obtained. The formula is as follows:

[0033]

[0034] In the formula, XTZ min represents the minimum value of the blood glucose value variable, XTZ max represents the maximum value of the blood glucose value variable, NDB min represents the minimum value of the urine protein value variable, NDB max represents the maximum value of the urine protein value variable, Tang represents the daily sugar intake of the patient, Zhi represents the daily fat intake of the patient, Tang normal represents the standard daily sugar intake of the patient, Zhi normal represents the standard daily fat intake of the patient, w7, w8 and w9 represent weight coefficients;

[0035] S52. By presetting a third standard threshold Q3 in advance and comparing and analyzing the comprehensive risk coefficient ZHXS with the third standard threshold Q3, a third evaluation result is obtained, including:

[0036] When the comprehensive risk coefficient ZHXS < the third standard threshold Q3, it indicates that the daily diet of diabetic patients has no impact risk on kidney lesions, and continuous monitoring is carried out;

[0037] When the comprehensive risk coefficient ZHXS ≥ the third standard threshold Q3, it indicates that the daily diet of diabetic patients has an impact risk on kidney lesions, triggering a third warning instruction and generating a dietary impact risk data report.

[0038] Preferably, step S6 further includes:

[0039] S61. Using a convolutional neural network, a risk assessment model for kidney lesions is constructed, and the initial model of the convolutional neural network is trained and tested with a comprehensive risk coefficient. At the same time, the kidney lesion features extracted by the kidney atrophy risk coefficient and the kidney fibrosis risk coefficient are used to identify the feature information, and the risk assessment model for kidney lesions is trained and tested with the obtained feature information. A prediction function is generated using a machine learning algorithm to predict the risk of kidney lesions in diabetic patients.

[0040] Preferably, an ultrasound-based risk detection system for kidney lesions in diabetic patients includes:

[0041] A first acquisition unit for acquiring ultrasound images of the kidney morphology, structure, kidney blood flow characteristics, and kidney cortex;

[0042] An image processing unit for preprocessing the ultrasound images. After image processing, the ultrasound images are spatially registered and then fused;

[0043] A region division unit for automatically segmenting the fused kidney region into several regions;

[0044] A first calculation unit for calculating and obtaining the kidney atrophy risk coefficient Wsfx;

[0045] A first evaluation unit for evaluating whether a diabetic patient has a risk of kidney atrophy and generating a kidney atrophy risk data report;

[0046] A second calculation unit for calculating and obtaining the kidney fibrosis risk coefficient XWHfx;

[0047] A second evaluation unit for evaluating whether a diabetic patient has a risk of kidney fibrosis and generating a kidney fibrosis risk data report;

[0048] A second acquisition unit for acquiring the blood glucose value XTZ, urine protein value NDB of a diabetic patient, and the dietary characteristics of the patient;

[0049] A third calculation unit for calculating and obtaining the comprehensive risk coefficient ZHXS;

[0050] A third evaluation unit for evaluating whether the daily diet of a diabetic patient has a risk of affecting kidney lesions and generating a dietary impact risk data report;

[0051] A model construction unit for constructing a risk assessment model for kidney lesions and predicting the risk of kidney lesions in diabetic patients.

[0052] The present invention provides an ultrasound-based risk detection method and system for kidney lesions in diabetic patients. It has the following beneficial effects:

[0053] (1) The method and system for detecting the risk of kidney lesions in diabetic patients based on ultrasound can achieve early warning of kidney lesions in diabetic patients by fusing multimodal ultrasound images and combining the risk coefficients of kidney atrophy and fibrosis. This method helps doctors detect problems in a timely manner when the lesions are still in the early stage by accurately evaluating the morphology, structure, blood flow characteristics of the kidneys and the changes in the renal cortex, avoiding missing the best treatment opportunity.

[0054] (2) The method and system for detecting the risk of kidney lesions in diabetic patients based on ultrasound generate a detailed risk data report by automatically segmenting and analyzing the kidney region and combining the calculated atrophy risk coefficient and fibrosis risk coefficient. This automated process reduces human error and improves the evaluation efficiency, helping doctors make more rapid and accurate diagnostic decisions.

[0055] (3) The method and system for detecting the risk of kidney lesions in diabetic patients based on ultrasound integrate ultrasound image data and combine physiological data such as the patient's blood glucose, urine protein, and dietary characteristics to monitor the dynamic changes in the risk of kidney atrophy and fibrosis in diabetic patients in real time, helping doctors formulate more precise intervention measures for patients, thereby effectively reducing the incidence of diabetes-related kidney complications.

[0056] (4) The method and system for detecting the risk of kidney lesions in diabetic patients based on ultrasound generate a comprehensive risk assessment model through machine learning algorithms. This model improves the prediction accuracy of kidney lesions in diabetic patients by considering multiple factors. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 It is a schematic diagram of the steps of a method for detecting the risk of kidney lesions in diabetic patients based on ultrasound according to the present invention;

[0058] Figure 2 It is a schematic diagram of the block diagram process of a system for detecting the risk of kidney lesions in diabetic patients based on ultrasound according to the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0059] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0060] Embodiment 1

[0061] Please refer to Figure 1 , the present invention provides a method for detecting the risk of kidney lesions in diabetic patients based on ultrasound, including the following steps:

[0062] S1. Set up a two-dimensional ultrasound probe, an elastography probe, and a Doppler probe in the renal region of the patient. The probes obtain the renal tissue through continuous scanning, and collect ultrasonic images of the renal morphology, structure, renal blood flow characteristics, and renal cortex.

[0063] S2. Use image processing technology to preprocess the ultrasonic images. After image processing, perform spatial registration on the ultrasonic images and then fuse them, and automatically segment the fused renal region into several regions.

[0064] S3. By extracting the renal tissue area Ss i of the i-th renal region, the renal cyst tissue area Snz i the roundness Sly of the renal contour i the smoothness Slp of the renal contour i the curvature Sql of the renal contour i and the texture feature Swl of the kidney i , calculate and obtain the renal atrophy risk coefficient Wsfx, and compare and analyze it with the first standard threshold to evaluate whether the diabetic patient has the risk of renal atrophy, and generate a renal atrophy risk data report.

[0065] S4. By extracting the renal cortex thickness Cor i of the i-th renal region, the hardness ZYD of the renal tissue i the blood perfusion rate BLG i , calculate and obtain the renal fibrosis risk coefficient XWHfx, and compare and analyze it with the second standard threshold to evaluate whether the diabetic patient has the risk of renal fibrosis, and generate a renal fibrosis risk data report.

[0066] S5. By collecting the blood glucose value XTZ, urine protein value NDB of the diabetic patient, and the dietary characteristics of the patient, calculate and obtain the comprehensive risk coefficient ZHXS, and compare and analyze it with the third standard threshold to evaluate whether the daily diet of the diabetic patient has an impact risk on renal lesions, and generate a dietary impact risk data report.

[0067] S6. Establish a renal lesion risk assessment model, train and test the model, use machine learning algorithms to generate a prediction function, and predict the risk of renal lesions in diabetic patients.

[0068] In this embodiment, by comprehensively using multimodal ultrasonic images and the physiological data of the patient, the atrophy and fibrosis risks of the kidneys of diabetic patients can be accurately evaluated. By constructing a renal lesion risk assessment model based on a convolutional neural network, the accurate prediction of renal lesion risk is realized, and a case report can be generated according to the specific situation of each patient, effectively helping doctors formulate early intervention measures and reducing the occurrence and progression of diabetes-related renal lesions.

[0069] Example 2

[0070] This embodiment is an explanatory description based on Embodiment 1. Specifically, step S1 includes:

[0071] At the kidney region of a diabetic patient, a two-dimensional ultrasound probe is set up to collect ultrasound images of the basic morphology and internal tissue structure of the kidney; an elastography probe is set up to collect ultrasound images of the kidney tissue hardness and the thickness of the kidney cortex; a Doppler probe is set up to collect ultrasound images of the kidney blood flow characteristics.

[0072] In this embodiment, by simultaneously using a two-dimensional ultrasound probe, an elastography probe, and a Doppler probe, this method can comprehensively obtain the morphology, structure, blood flow characteristics, and cortical state of the kidney, providing multi-dimensional kidney image information. This comprehensive scanning method significantly improves the comprehensiveness and accuracy of image acquisition, providing more refined and reliable basic data for subsequent risk assessment and lesion analysis.

[0073] Example 3

[0074] This embodiment is an explanatory description based on Embodiment 1. Specifically, step S2 includes:

[0075] S21. Use wavelet denoising technology to denoise the collected ultrasound images. Decompose the data information in several tissue layer images and kidney structure layout images through wavelet transform, remove the noise components in the wavelet domain, and use histogram equalization to adjust the brightness, stretch the image range, and adjust the contrast of the image;

[0076] S22. Perform spatial registration on the ultrasound images so that images of different modalities can be aligned in the same coordinate system for fusion. Automatically segment the fused kidney region into several regions, which are respectively labeled as region L1, L2, L3,..., Ln, where n represents the total number of regions.

[0077] In this embodiment, through wavelet denoising technology and histogram equalization, the quality of the ultrasound images is significantly improved, the noise is removed, and the contrast of the images is enhanced, making the details of the kidney tissue clearer. In addition, the spatial registration technology ensures the accurate alignment of multi-modal images, further improving the accuracy of image fusion, and then realizing the automatic segmentation and accurate labeling of the kidney region, providing a reliable image basis for subsequent analysis, and improving the accuracy and meticulousness of region division.

[0078] Example 4

[0079] This embodiment is an explanatory description based on Embodiment 1. Specifically, step S3 includes:

[0080] S31. By extracting the kidney tissue area Ss of the i-th kidney region i , the kidney cyst tissue area Snz i , the roundness Sly of the kidney contour i , the smoothness Slp of the kidney contour i , the curvature Sql of the kidney contour i and the texture feature Swl of the kidney i , after dimensionless processing, calculate and obtain the atrophy risk coefficient Wsfx of the kidney, and the formula is as follows:

[0081]

[0082] In the formula, n represents the total number of kidney regions, S normal represents the reference area of a healthy kidney, R iderl represents the ideal kidney contour feature, Swl mean represents the texture mean inside the kidney, Swl total represents the texture mean inside a healthy kidney, w1, w2 and w3 represent weight coefficients, 0 < w1 < 1, 0 < w2 < 1 and 0 < w3 < 1, and w1 + w2 + w3 = 1;

[0083] S32. By presetting the first standard threshold Q1 in advance and comparing and analyzing the atrophy risk coefficient Wsfx of the kidney with the first standard threshold Q1, obtain the first evaluation result, including:

[0084] When the atrophy risk coefficient Wsfx of the kidney < the first standard threshold Q1, it means that the kidney of the diabetic patient has no atrophy risk, and continuous monitoring is carried out;

[0085] When the atrophy risk coefficient Wsfx of the kidney ≥ the first standard threshold Q1, it means that the kidney of the diabetic patient has an atrophy risk, trigger the first warning instruction, and generate a kidney atrophy risk data report.

[0086] In this embodiment, by extracting multi-dimensional features of the kidney region, such as kidney tissue area, contour roundness, smoothness and texture features, and combining dimensionless processing, the atrophy risk coefficient of the kidney can be accurately calculated. This method not only provides a comprehensive evaluation based on multiple structural and texture features, but also improves the accuracy of risk assessment through flexible adjustment of weight coefficients, helps to detect the kidney atrophy risk of diabetic patients at an early stage, so as to take intervention measures in advance; by setting the standard threshold Q1 and comparing it with the atrophy risk coefficient of the kidney, the automatic evaluation of the kidney atrophy risk of diabetic patients can be realized. When the risk coefficient exceeds the preset threshold, the system can trigger an alarm in time and generate a detailed risk report, providing important decision-making support for clinicians. This automatic warning mechanism effectively improves the monitoring efficiency of kidney diseases and provides an opportunity for early intervention for patients.

[0087] Example 5

[0088] This example is an explanatory note carried out in Example 1. Specifically, step S4 includes:

[0089] S41. Extract the renal cortex thickness Cor of the i-th kidney region i , the hardness ZYD of the renal tissue i , the blood perfusion rate BLG i . After dimensionless processing, calculate and obtain the renal fibrosis risk coefficient XWHfx. The formula is as follows:

[0090]

[0091] In the formula, Cor normal represents the cortex thickness of a healthy kidney, ZYD iderl represents the hardness of an ideal renal tissue, BLG normal represents the blood perfusion rate of a healthy kidney, and w4, w5, and w6 represent weight coefficients, where 0 < w4 < 1, 0 < w5 < 1, and 0 < w6 < 1, and w4 + w5 + w6 = 1;

[0092] S42. Preset a second standard threshold Q2 in advance, and compare and analyze the renal fibrosis risk coefficient XWHfx with the second standard threshold Q2 to obtain a second evaluation result, including:

[0093] When the renal fibrosis risk coefficient XWHfx < the second standard threshold Q2, it indicates that the kidney of the diabetic patient has no fibrosis risk, and continuous monitoring is carried out;

[0094] When the renal fibrosis risk coefficient XWHfx ≥ the second standard threshold Q2, it indicates that the kidney of the diabetic patient has a fibrosis risk, trigger a first warning instruction, and generate a renal fibrosis risk data report.

[0095] In this example, by extracting key parameters such as the renal cortex thickness, tissue hardness, and blood perfusion rate, and calculating the renal fibrosis risk coefficient XWHfx, the fibrosis degree of the kidneys of diabetic patients can be accurately evaluated. This method is based on the comprehensive data obtained after dimensionless processing, providing comprehensive and detailed renal fibrosis risk information for clinicians, thus helping to timely detect potential fibrosis risks; by comparing and analyzing the renal fibrosis risk coefficient XWHfx with the preset second standard threshold Q2, early warning of renal fibrosis can be achieved. When the risk coefficient exceeds the set threshold, the system will automatically trigger a warning and generate a fibrosis risk data report to help doctors timely detect the renal fibrosis risk of diabetic patients, so as to take early intervention measures and effectively prevent the further deterioration of the condition.

[0096] Example 6

[0097] This example is an explanatory description based on Example 1. Specifically, step S5 includes:

[0098] S51. By collecting the blood glucose value XTZ, urine protein value NDB of diabetic patients and the dietary characteristics of the patients, after dimensionless processing, calculate and obtain the comprehensive risk coefficient ZHXS. The formula is as follows:

[0099]

[0100] In the formula, XTZ min represents the minimum value of the blood glucose value variable, XTZ max represents the maximum value of the blood glucose value variable, NDB min represents the minimum value of the urine protein value variable, NDB max represents the maximum value of the urine protein value variable, Tang represents the daily sugar intake of the patient, Zhi represents the daily fat intake of the patient, Tang normal represents the standard daily sugar intake of the patient, Zhi normal represents the standard daily fat intake of the patient, w7, w8 and w9 represent weight coefficients, 0 < w7 < 1, 0 < w8 < 1 and 0 < w9 < 1, and w7 + w8 + w9 = 1;

[0101] S52. By presetting the third standard threshold Q3 in advance and comparing and analyzing the comprehensive risk coefficient ZHXS with the third standard threshold Q3, obtain the third evaluation result, including:

[0102] When the comprehensive risk coefficient ZHXS < the third standard threshold Q3, it means that the daily diet of diabetic patients has no impact risk on kidney lesions, and continuous monitoring is carried out;

[0103] When the comprehensive risk coefficient ZHXS ≥ the third standard threshold Q3, it means that the daily diet of diabetic patients has an impact risk on kidney lesions, trigger the third warning instruction, and generate a diet impact risk data report.

[0104] In this example, by collecting and analyzing the blood glucose, urine protein and dietary characteristics of diabetic patients, calculating the comprehensive risk coefficient ZHXS, personalized kidney lesion risk assessment can be provided for each patient. This method can accurately reflect the health status of patients and provide a refined risk management plan based on the eating habits and physiological indicators of patients, which helps doctors formulate personalized treatment and intervention measures for different patients, thereby effectively reducing the occurrence probability of kidney lesions.

[0105] Example 7

[0106] This example is an explanatory description based on Example 1. Specifically, step S6 includes:

[0107] S61. Using a convolutional neural network, construct a risk assessment model for kidney lesions, and train and test the initial model of the convolutional neural network with a comprehensive risk coefficient. At the same time, use the kidney lesion features extracted by the kidney atrophy risk coefficient and the kidney fibrosis risk coefficient to identify feature information, and train and test the kidney lesion risk assessment model with the obtained feature information. Use a machine learning algorithm to generate a prediction function to predict the risk of kidney lesions in diabetic patients.

[0108] In this embodiment, the risk assessment model for kidney lesions constructed by a convolutional neural network (CNN), combined with the comprehensive risk coefficient, atrophy risk coefficient, and fibrosis risk coefficient, can accurately predict the risk of kidney lesions in diabetic patients. The application of the machine learning algorithm enables the model to be continuously optimized and trained based on the multi-dimensional data of the patient, improving the prediction accuracy. This technology can identify potential kidney lesion risks at an early stage, thus providing reliable decision-making support for doctors, intervening in advance, and effectively delaying the development of the disease.

[0109] Example 8

[0110] A risk detection system for kidney lesions in diabetic patients based on ultrasound, please refer to Figure 2 , including:

[0111] The first acquisition unit is used to acquire ultrasound images of the kidney morphology, structure, kidney blood flow characteristics, and kidney cortex;

[0112] The image processing unit is used to preprocess the ultrasound images. After image processing, the ultrasound images are spatially registered and then fused;

[0113] The region division unit is used to automatically segment the fused kidney region into several regions;

[0114] The first calculation unit is used to calculate and obtain the kidney atrophy risk coefficient Wsfx;

[0115] The first evaluation unit is used to evaluate whether a diabetic patient has a risk of kidney atrophy and generate a kidney atrophy risk data report;

[0116] The second calculation unit is used to calculate and obtain the kidney fibrosis risk coefficient XWHfx;

[0117] The second evaluation unit is used to evaluate whether a diabetic patient has a risk of kidney fibrosis and generate a kidney fibrosis risk data report;

[0118] The second acquisition unit is used to acquire the blood glucose value XTZ, urine protein value NDB of the diabetic patient, and the diet characteristics of the patient;

[0119] A third calculation unit, configured to calculate and obtain a comprehensive risk coefficient ZHXS;

[0120] A third evaluation unit, configured to evaluate whether the daily diet of diabetic patients has an impact risk on kidney lesions and generate a diet impact risk data report;

[0121] A model construction unit, configured to construct a kidney lesion risk assessment model to predict the risk of kidney lesions in diabetic patients.

[0122] In this embodiment, through the combination of multimodal ultrasound image acquisition and intelligent algorithms, the kidney health status of diabetic patients can be comprehensively evaluated. The system conducts multi-dimensional comprehensive analysis from the kidney morphology, blood flow characteristics to the patient's blood glucose, urine protein and diet data, and provides an accurate kidney lesion risk assessment by calculating the atrophy, fibrosis and comprehensive risk coefficients. This technology can help doctors quickly and accurately identify the potential lesion risks of patients' kidneys, perform personalized interventions in a timely manner, and significantly improve the efficiency and accuracy of early diagnosis.

[0123] The setting of the threshold size is for the convenience of comparison. Regarding the threshold size, it depends on the amount of sample data and the base quantity set by those skilled in the art for each group of sample data; as long as it does not affect the proportional relationship between the parameters and the quantified values.

[0124] The above formulas are all obtained by collecting a large amount of data for software simulation and selecting a formula close to the true value. The coefficients in the formula are set by those skilled in the art according to the actual situation. The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, makes equivalent substitutions or changes, and should be covered by the protection scope of the present invention.

Claims

1. A method for detecting the risk of kidney disease in diabetic patients based on ultrasound, characterized in that: The following steps are involved: S1. A two-dimensional ultrasound probe, an elastic imaging probe and a Doppler probe are set in the patient's kidney area. The probes obtain the kidney tissue through continuous scanning, and collect the ultrasound images of the kidney morphology, structure, kidney blood flow characteristics and kidney cortex; S2. Preprocessing the ultrasound image using image processing technology. After the image processing, the ultrasound image is spatially registered and then fused, and the fused kidney area is automatically segmented into several areas; S3, by extracting the renal tissue area Ss of the i-th renal region i 、Renal cyst tissue area Snz i , kidney contour roundness Sly i , the smoothness of the kidney contour Slp i , the curvature of the kidney contour Sql i and the texture features of the kidney Swl i , calculate and obtain the kidney atrophy risk coefficient Wsfx, and compare and analyze it with the first standard threshold, evaluate whether the diabetic patient has the risk of kidney atrophy, and generate a kidney atrophy risk data report; S4, by extracting the renal cortical thickness Cor of the i-th renal region i 、The hardness of kidney tissue ZYD i , blood perfusion rate BLG i , calculate and obtain the renal fibrosis risk coefficient XWHfx, and compare and analyze it with the second standard threshold to evaluate whether diabetic patients have the risk of renal fibrosis and generate a renal fibrosis risk data report; S5. By collecting the blood sugar value XTZ, urine protein value NDB and dietary characteristics of diabetic patients, the comprehensive risk coefficient ZHXS is calculated and compared with the third standard threshold value to evaluate whether the daily diet of diabetic patients has an impact on the risk of kidney disease, and generate a diet impact risk data report; S6. Establish a kidney disease risk assessment model, train and test the model, and use machine learning algorithms to generate prediction functions to predict the risk of kidney disease in diabetic patients.

2. The method for detecting the risk of kidney disease in diabetic patients based on ultrasound according to claim 1, characterized in that: The step S1 is specifically as follows: In the kidney area of ​​diabetic patients, a two-dimensional ultrasound probe is set to collect ultrasound images of the basic morphology of the kidney and the internal tissue structure; an elastic imaging probe is set to collect ultrasound images of the hardness of the kidney tissue and the thickness of the kidney cortex; and a Doppler probe is set to collect ultrasound images of the blood flow characteristics of the kidney.

3. The method for detecting the risk of kidney disease in diabetic patients based on ultrasound according to claim 2, characterized in that: The step S2 comprises: S21, using wavelet denoising technology to denoise the collected ultrasound image, decomposing the data information in several tissue layer images and kidney structure layout images through wavelet transform, removing the noise component in the wavelet domain, and using histogram equalization to adjust the brightness, stretch the image range, and adjust the image contrast; S22. Perform spatial registration on the ultrasound images so that images of different modalities can be aligned in the same coordinate system for fusion, and automatically segment the fused kidney region into several regions, which are marked as regions L1, L2, L3, ..., Ln, respectively, where n represents the total number of regions.

4. The method for detecting the risk of kidney disease in diabetic patients based on ultrasound according to claim 3, characterized in that: The step S3 comprises: S31, by extracting the renal tissue area Ss of the i-th renal region i 、Renal cyst tissue area Snz i , kidney contour roundness Sly i , the smoothness of the kidney contour Slp i , the curvature of the kidney contour Sql i and the texture features of the kidney Swl i After dimensionless processing, the kidney atrophy risk coefficient Wsfx is calculated and obtained. The formula is as follows: Where n represents the total number of kidney regions, S normal represents the reference area of ​​a healthy kidney, R iderl represents the ideal kidney contour feature, Swl mean represents the mean texture inside the kidney, Swl total represents the texture mean inside a healthy kidney, w1, w2, and w3 represent weight coefficients; S32, by presetting the first standard threshold Q1 in advance, and comparing and analyzing the kidney atrophy risk coefficient Wsfx with the first standard threshold Q1, a first evaluation result is obtained, including: When the renal atrophy risk coefficient Wsfx is less than the first standard threshold Q1, it means that the diabetic patient has no risk of renal atrophy and needs to be continuously monitored; When the kidney atrophy risk coefficient Wsfx ≥ the first standard threshold value Q1, it indicates that the diabetic patient's kidney is at risk of atrophy, triggering the first warning instruction and generating a kidney atrophy risk data report.

5. The method for detecting the risk of kidney disease in diabetic patients based on ultrasound according to claim 3, characterized in that: The step S4 comprises: S41, by extracting the renal cortical thickness Cor of the i-th renal region i 、The hardness of kidney tissue ZYD i , blood perfusion rate BLG i After dimensionless processing, the renal fibrosis risk coefficient XWHfx is calculated and the formula is as follows: In the formula, Cor normal Represents the cortical thickness of a healthy kidney, ZYD iderl Indicates the hardness of ideal kidney tissue, BLG normal represents the blood perfusion rate of healthy kidney, w4, w5 and w6 represent weight coefficients; S42, by presetting the second standard threshold Q2 in advance, and comparing and analyzing the renal fibrosis risk coefficient XWHfx with the second standard threshold Q2, a second evaluation result is obtained, including: When the renal fibrosis risk coefficient XWHfx is less than the second standard threshold Q2, it means that the diabetic patient has no risk of renal fibrosis and needs to be continuously monitored; When the renal fibrosis risk coefficient XWHfx ≥ the second standard threshold Q2, it indicates that the diabetic patient has a risk of renal fibrosis, triggering the second warning instruction and generating a renal fibrosis risk data report.

6. The method for detecting the risk of kidney disease in diabetic patients based on ultrasound according to claim 1, characterized in that: The step S5 comprises: S51. By collecting the blood sugar value XTZ, urine protein value NDB and dietary characteristics of diabetic patients, after dimensionless processing, the comprehensive risk coefficient ZHXS is calculated, and the formula is as follows: Where XTZ min Indicates the minimum value of the blood sugar numerical variable, XTZ max Indicates the maximum value of the blood sugar numerical variable, NDB min Indicates the minimum value of urine protein numerical variable, NDB max represents the maximum value of the urine protein numerical variable, Tang represents the patient's daily sugar intake, Zhi represents the patient's daily fat intake, Tang normal Indicates the patient's standard daily sugar intake, normal represents the patient's standard daily fat intake, w7, w8 and w9 represent weight coefficients; S52, by presetting the third standard threshold Q3 in advance, and comparing and analyzing the comprehensive risk factor ZHXS with the third standard threshold Q3, a third assessment result is obtained, including: When the comprehensive risk coefficient ZHXS is less than the third standard threshold Q3, it means that the daily diet of diabetic patients has no risk of affecting kidney disease and needs to be continuously monitored; When the comprehensive risk coefficient ZHXS ≥ the third standard threshold Q3, it means that the daily diet of diabetic patients has a risk of affecting kidney disease, triggering the third warning instruction and generating a diet impact risk data report.

7. The method for detecting the risk of kidney disease in diabetic patients based on ultrasound according to claim 6, characterized in that: The step S6 comprises: S61. Use convolutional neural networks to construct a kidney disease risk assessment model, and use the comprehensive risk coefficient to train and test the initial convolutional neural network model. At the same time, use the kidney atrophy risk coefficient and kidney fibrosis risk coefficient to extract kidney disease features to identify feature information, and use the acquired feature information to train and test the kidney disease risk assessment model. Use machine learning algorithms to generate prediction functions to predict the risk of kidney disease in diabetic patients.

8. An ultrasound-based diabetic renal lesion risk detection system, applied to an ultrasound-based diabetic renal lesion risk detection method according to any one of claims 1 to 7, the detection system comprising: The first acquisition unit is used to acquire ultrasound images of kidney morphology, structure, kidney blood flow characteristics, and kidney cortex; An image processing unit is used to pre-process the ultrasound image, and after the image processing, perform spatial registration and fusion on the ultrasound image; A region segmentation unit is used to automatically segment the fused kidney region into several regions; The first calculation unit is used to calculate and obtain the kidney atrophy risk coefficient Wsfx; The first assessment unit is used to assess whether a diabetic patient has a risk of renal atrophy and generate a renal atrophy risk data report; The second calculation unit is used to calculate and obtain the renal fibrosis risk coefficient XWHfx; The second assessment unit is used to assess whether diabetic patients have a risk of renal fibrosis and generate a renal fibrosis risk data report; The second collection unit is used to collect the blood sugar value XTZ, urine protein value NDB and dietary characteristics of the diabetic patient; The third calculation unit is used to calculate and obtain the comprehensive risk coefficient ZHXS; The third assessment unit is used to assess whether the daily diet of diabetic patients has an impact on the risk of kidney disease and generate a data report on the impact of diet on risk; A model building unit is used to build a kidney disease risk assessment model to predict the risk of kidney disease in diabetic patients.

Images