Method and system for evaluating lower extremity arterial surgical intervention based on image data

Abstract

The present application relates to the technical field of machine learning, more particularly to a lower extremity arterial surgery intervention evaluation method and system based on image data. The method comprises the following steps: S1: obtaining complete blood vessel contours and a blood vessel three-dimensional model based on blood flow information changes; S2: obtaining a tree structure of lower extremity arteries, dividing the blood vessel three-dimensional model into multiple blood vessel branches and blood vessel sections, extracting feature values of each blood vessel branch and blood vessel section, and obtaining a target blood vessel based on the feature values; S3: simulating multiple surgical plans and operation data corresponding to each surgical plan based on the target blood vessel and the corresponding feature values through a simulation unit; and S4: inputting basic information of a patient, feature values of the target blood vessel, three-dimensional images, and operation data corresponding to each surgical plan into a prediction model, and taking the surgical plan with the highest success rate as a target surgical plan. The present application solves the problem of inaccurate lower extremity arterial surgery intervention evaluation and improves the evaluation accuracy.

Description

Technical Field

[0001] This invention relates to the field of machine learning technology, and more specifically to a method and system for evaluating lower extremity arterial surgical interventions based on image data. Background Technology

[0002] With the rapid development of science and technology, modern medical diagnostic technology has also made significant progress. Through deep learning of medical imaging data and basic patient information, high-precision prediction of preoperative risks can be achieved. A similar prior art is Chinese Patent Publication No. CN113611390A, which proposes a risk prediction model and construction system for the conversion of thoracoscopic lung cancer resection to open thoracotomy. The construction system includes: a data acquisition module, a risk diagnosis model construction module, a data preprocessing module, a data fusion module, a dataset training module, a model validation and optimization module, a central control module, a risk diagnosis prediction module, a data storage module, and an update and display module. Based on tens of thousands of cases of data from the Department of Thoracic Surgery at West China Hospital, a risk prediction model for the conversion of thoracoscopic lung cancer resection to open thoracotomy is established using classic machine learning algorithms such as logistic regression, artificial neural networks, and random forests. Based on the established web-based nomogram, after inputting some indicators, the risk of conversion to open thoracotomy during thoracoscopic lung cancer resection can be predicted online. Furthermore, a similar prior art exists in Korean Patent Publication No. KR20190110381A, which discloses a clinical trial outcome prediction device and method. This method uses machine learning algorithms to predict the success probability of clinical trials for new treatments. It includes an input unit for inputting relevant clinical trial information and learning through processing clinical trial cases; a data processing unit for generating data; and a predictive model for predicting clinical trial outcomes using the learned data through machine learning. The generated predictive model is then used to predict the clinical trial outcome based on the relevant clinical trial information. Both of these patent documents achieve risk prediction for surgery or outcomes before surgery or clinical trials, but the accuracy is insufficient, and they cannot accurately predict the probability of surgical success. Summary of the Invention

[0003] To better address the above problems, this invention provides a method for evaluating lower extremity arterial surgical intervention based on imaging data, the method comprising the following steps:

[0004] Step S1: Acquire lower limb arterial angiography images of the patient through the camera unit, obtain blood flow information based on the pixel values ​​corresponding to the angiography area in the angiography image, and obtain a complete vascular contour and 3D vascular model based on the changes in the blood flow information;

[0005] Step S2: Obtain the tree structure of the lower limb arteries based on the three-dimensional vascular model, divide the three-dimensional vascular model into multiple vascular branches and vascular segments from top to bottom according to the tree structure, extract the feature values ​​of each vascular branch and vascular segment, and obtain the target blood vessel based on the feature values;

[0006] Step S3: Obtain a three-dimensional image of the target blood vessel through the three-dimensional blood vessel model; obtain multiple surgical options from the database based on the three-dimensional image of the target blood vessel and the feature values; and obtain corresponding simulated surgical operation data by simulating each of the surgical options through a simulation unit.

[0007] Step S4: Also obtain the patient's basic information, input the patient's basic information, the feature value of the target blood vessel, the three-dimensional image, and the simulated surgical operation data corresponding to each of the selected surgical options into the prediction model, and obtain and use the selected surgical option with the highest success rate as the target surgical option.

[0008] As a preferred embodiment of the present invention, step S1 includes:

[0009] Step S11: Obtain lower limb arterial angiography images through the camera unit, and obtain blood flow information based on the pixel values ​​corresponding to the angiography area in the angiography image. The blood flow information includes the blood flow path and the width of the blood flow path.

[0010] Step S12: Based on the blood flow information, obtain the target location where the blood flow path is broken or the width of the blood flow path is narrowed, and obtain the blood vessel contour corresponding to the target location based on the center line of the blood vessel before and after the target location, the blood vessel trend, the blood vessel contour and the blood flow information;

[0011] Step S13: Based on the blood flow information and the corresponding vascular contour in each angiographic image of the lower limb artery, construct the three-dimensional vascular model of the lower limb artery.

[0012] As a preferred embodiment of the present invention, step S2 includes:

[0013] Step S21: Obtain the tree structure of the lower limb artery based on the three-dimensional vascular model, and divide the lower limb artery into multiple vascular branches and vascular segments according to the top-to-bottom direction of the tree structure;

[0014] Step S22: Obtain the vessel diameter based on the vessel contour of each vessel branch or vessel segment, calculate the ratio of the blood flow path width to the vessel diameter in the corresponding blood flow information, and also calculate the number of pixels on the vessel surface of the vessel branch or vessel segment whose pixel value is greater than a set threshold, and use the ratio and the number of pixels as the feature value of the vessel branch or vessel segment;

[0015] Step S23: When the ratio is less than or equal to a first preset value, the vascular branch or the vascular segment is designated as a first segment, wherein the first segment is an occluded segment, and the length of the occluded segment is obtained; when the ratio is greater than the first preset value and less than or equal to a second preset value, the vascular branch or the vascular segment is designated as a second segment, and the size of the plaque in the second segment is calculated based on the length of the second segment; when the number of pixels is greater than a third preset value, the vascular branch or the vascular segment is designated as a third segment.

[0016] Step S24: The first segment, the second segment, and the third segment are taken as the target blood vessel, wherein the first segment, the second segment, or the third segment may overlap, and the feature values ​​of the target blood vessel further include the length of the occluded segment in the first segment, the size of the plaque in the second segment, the number of pixels in the third segment, and the position and shape of the target blood vessel.

[0017] As a preferred embodiment of the present invention, step S3 includes:

[0018] Step S31: Obtain the three-dimensional image of the target blood vessel based on its location and the three-dimensional model of the blood vessel;

[0019] Step S32: Based on the three-dimensional image of the target blood vessel and the feature value corresponding to the target blood vessel, select a matching candidate surgical plan from the lower limb artery surgical plan template stored in the database, and simulate the surgical operation according to the feature value and the three-dimensional image through the simulation unit based on the candidate surgical plan, and obtain simulated surgical operation data, wherein the simulated surgical operation data includes surgical operation process data, surgical operation location and surgical tool information.

[0020] As a preferred embodiment of the present invention, step S4 includes:

[0021] The basic information of the patient is obtained, and the patient's basic information, the feature value of the target blood vessel, the three-dimensional image, and the simulated surgical operation data corresponding to each of the selected surgical options are input into the prediction model to obtain the success rate of each selected surgical option. The selected surgical option with the highest success rate is taken as the target surgical option, and the target surgical option and the surgical tools in the target surgical option are used as the reference for actual surgery. The patient's basic information includes the patient's age, gender, blood pressure, blood glucose, blood lipids, high homocysteine ​​level, renal function, and inflammatory markers.

[0022] As a preferred embodiment of the present invention, the training of the prediction model includes:

[0023] Step S310: Obtain all lower extremity arterial surgery data from the database. The surgical data includes the patient's basic information, a first angiography three-dimensional image, the feature value of the first angiography three-dimensional image, first surgical operation data, and first annotation information. The surgical data is used as the first sample data, wherein the first annotation information is the surgical success result corresponding to the first sample data.

[0024] Step S311: Cluster the surgical data according to the basic information of the corresponding patients and obtain multiple clusters. Divide the first three-dimensional contrast imaging image of the first sample data in each cluster into multiple sub-images according to the corresponding feature values. Flip and combine any two or more sub-images to obtain multiple second three-dimensional contrast imaging images. Combine the sub-images in the second three-dimensional contrast imaging images with the sub-operation data in the first surgical operation data to obtain second surgical operation data. Determine the compatibility between the corresponding sub-operation data in the second surgical operation data.

[0025] Step S312: When the sub-operation data in the second surgical operation data are compatible, the mean value of the patient information corresponding to the sub-image is used as the new patient information, and the new patient information, the second angiography data, the feature value of the second angiography data, the second surgical operation data and the second annotation information are used as the second sample data, wherein the second annotation information indicates that the surgery was successful. The first sample data and the second sample data are used as training data to train the prediction model.

[0026] As a preferred embodiment of the present invention, step S4 further includes: using the patient's basic information, the feature value of the target blood vessel, the three-dimensional image of the target blood vessel, and the surgical operation data corresponding to the target surgical plan as input data for the prediction model; using items other than age and gender in the patient's basic information in the input data as adjustment items; obtaining a first value and a second value for each adjustment item within a first range and a second range respectively; replacing the corresponding values ​​in the patient's basic information in the input data with the adjusted first value and second value respectively; inputting the replaced input data into the prediction model; obtaining a first success rate and a second success rate respectively; and obtaining the impact of each adjustment item on the success rate based on the difference between the first success rate and the second success rate, wherein the first range is greater than the value of the adjustment item in the patient's basic information, the second range is less than the value of the adjustment item in the patient's basic information, and the difference between the first value and the second value and the value of the adjustment item in the patient's basic information is equal to the difference between the first value and the second value.

[0027] As a preferred embodiment of the present invention, step S5 is further included after step S4:

[0028] The imaging unit acquires data on whether the catheter passes through the true lumen of the target blood vessel during the procedure, and obtains a first score. If the catheter passes through the true lumen, the first score is positive; if it does not pass through the true lumen, the first score is negative. The system also acquires data on whether there is dissection before the placement of the stent or drug-coated balloon, and obtains a second score. If there is dissection, the second score is negative; if there is no dissection, the second score is positive. The final success rate is the sum of the product of the first score and its corresponding weight, the product of the second score and its corresponding weight, and the success rate.

[0029] As a preferred embodiment of the present invention, the prediction model is a random forest model.

[0030] The present invention also provides a lower extremity arterial surgical intervention assessment system based on image data, the system being used to implement the above-described method, the system comprising:

[0031] The camera unit is used to acquire lower limb arterial angiography images of the patient, obtain blood flow information based on the pixel values ​​corresponding to the angiography area in the angiography image, and obtain a complete vascular contour and 3D vascular model based on the changes in the blood flow information.

[0032] The extraction unit is used to obtain the tree structure of the lower limb arteries based on the three-dimensional vascular model, divide the three-dimensional vascular model into multiple vascular branches and vascular segments from top to bottom according to the tree structure, extract the feature value of each vascular branch and vascular segment, and obtain the target vascular based on the feature value;

[0033] The simulation unit is used to obtain a three-dimensional image of the target blood vessel through the three-dimensional blood vessel model, obtain multiple surgical options to be selected from the database based on the three-dimensional image of the target blood vessel and the feature values, and obtain corresponding simulated surgical operation data by simulating each of the surgical options to be selected through the simulation unit.

[0034] The prediction unit is used to acquire the patient's basic information, input the patient's basic information, the feature value of the target blood vessel, the three-dimensional image, and the simulated surgical operation data corresponding to each of the proposed surgical options into the prediction model, and acquire and select the proposed surgical option with the highest success rate as the target surgical option.

[0035] Compared with the prior art, the beneficial effects of the present invention are at least as follows:

[0036] This invention obtains blood flow information from the pixel values ​​of the angiography area in the aforementioned angiography image, and obtains the blood flow path and width based on the blood flow information to identify the target location where the blood flow is interrupted or the blood flow path width is narrowed. It extends the blood vessel's centerline, trend, and contour before and after the target location, and obtains the blood vessel contour and three-dimensional model of the target location based on the blood flow path of the target blood vessel. Furthermore, it divides the lower limb arteries into multiple blood vessel branches and segments in a top-to-bottom direction using a tree structure, determines whether the blood vessel branches or segments are blocked or have plaques, and the number of pixels in calcified areas, and obtains the feature values ​​of the target blood vessel. It also extracts the feature values ​​of the target blood vessel from the aforementioned data. The system selects matching candidate surgical plans from the lower extremity arterial surgery plan templates stored in the database. A simulation unit then simulates the surgical operation data of each candidate plan on the 3D image of the target blood vessel based on feature values. The system uses not only the feature values ​​of the target blood vessel, the 3D image of the target blood vessel, and the simulated surgical data corresponding to each candidate plan as input data to the prediction model, but also the patient's basic information. The prediction model outputs the success rate of each candidate plan, and the candidate plan with the highest success rate is selected as the target surgical plan. Through the synergy of these technical solutions, the accuracy of predicting the success rate of lower extremity arterial surgery is improved, and the target surgical plan with the highest success rate is obtained. Attached Figure Description

[0037] Figure 1 This is a flowchart of the lower extremity arterial surgical intervention assessment method based on image data according to the present invention;

[0038] Figure 2 This is a structural diagram of the lower limb arterial surgical intervention assessment system based on image data according to the present invention. Detailed Implementation

[0039] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0040] This invention provides a method for evaluating lower extremity arterial surgical intervention based on imaging data, such as... Figure 1 As shown, the method includes the following steps:

[0041] Step S1: Acquire lower limb arterial angiography images of the patient through the camera unit, obtain blood flow information based on the pixel values ​​corresponding to the angiography area in the angiography image, and obtain a complete vascular contour and 3D vascular model based on the changes in the blood flow information;

[0042] Specifically, the aforementioned camera unit acquires the aforementioned lower limb arterial angiography images, and further acquires blood flow information based on the pixel values ​​of the angiography area in the aforementioned angiography images. Based on the aforementioned blood flow information, the blood flow path and blood flow path width are acquired to determine the target location where the blood flow is interrupted or the blood flow path width narrows. The centerline, blood vessel trend, and blood vessel contour of the aforementioned blood vessel before and after the target location are extended, and the blood vessel contour of the aforementioned target location is acquired based on the blood flow path of the aforementioned target blood vessel. Similarly, the blood vessel contour at each location is acquired, and the aforementioned three-dimensional blood vessel model is constructed based on the aforementioned blood flow information and blood vessel contour in each of the aforementioned lower limb arterial angiography images. Through the aforementioned technical solution, even when blood vessels are narrowed or blocked, accurate blood vessel contours and three-dimensional blood vessel models can be acquired, laying the foundation for facilitating the calculation of feature values ​​of blood vessel segments or branches through the aforementioned three-dimensional blood vessel model.

[0043] Step S2: Obtain the tree structure of the lower limb arteries based on the three-dimensional vascular model, divide the three-dimensional vascular model into multiple vascular branches and vascular segments from top to bottom according to the tree structure, extract the feature values ​​of each vascular branch and vascular segment, and obtain the target blood vessel based on the feature values;

[0044] Specifically, the tree-like structure of the lower limb arteries is obtained through the aforementioned three-dimensional vascular model. This tree-like structure is then used to divide the lower limb arteries into multiple branches and segments from top to bottom, facilitating the analysis of the characteristics of each branch or segment. By calculating the ratio of the blood flow path width to the vessel diameter for each branch or segment, it is determined whether the branch or segment is blocked or has plaque. Furthermore, the number of pixels on the vessel surface corresponding to the branch or segment with pixel values ​​greater than a set threshold (i.e., the number of pixels in calcified areas) is calculated. This ratio and the number of pixels are used as feature values ​​for analyzing the branch or segment. When the ratio is less than or equal to a first preset value, the segment is considered an occluded segment and designated as the first segment. The length of the occluded segment is then calculated. When the ratio is greater than the first preset value and less than or equal to the second preset value, the blood vessel contains plaque and is designated as the second segment. The plaque size is also calculated, and the cross-sectional area of ​​the blood vessel is the product of the ratio and the cross-sectional area of ​​the plaque. When the ratio is greater than the second preset value, the blood vessel is considered a normal blood vessel. When the number of pixels is greater than the third preset value, the blood vessel is designated as the third segment, i.e., the calcified segment. The first, second, and third segments are designated as the target blood vessel. The occlusion length in the first segment, the plaque size in the second segment, or the number of pixels in the third segment are added to the feature values ​​of the target blood vessel. The position and shape of the target blood vessel are also added to the feature values. Through the above technical solution, the feature values ​​of the target blood vessel can be accurately obtained, providing data basis for accurately predicting the success rate of surgery before operation.

[0045] Step S3: Obtain a three-dimensional image of the target blood vessel through the three-dimensional blood vessel model; obtain multiple surgical options from the database based on the three-dimensional image of the target blood vessel and the feature values; and obtain corresponding simulated surgical operation data by simulating each of the surgical options through a simulation unit.

[0046] Specifically, the three-dimensional image of the target blood vessel can be obtained through its location and three-dimensional model. Based on the three-dimensional image and feature values ​​of the target blood vessel, a matching candidate surgical plan is selected from the lower limb artery surgical plan templates stored in the database. That is, the candidate surgical plan is selected based on the disease type, disease data, location, and shape of the target blood vessel. The simulation unit simulates the simulated surgical operation data of each candidate surgical plan on the three-dimensional image of the target blood vessel based on the feature values. Through the above technical solution, the simulated surgical operation data of each candidate surgical plan can be obtained, laying the foundation for selecting the target surgical plan from the candidate surgical plans.

[0047] Step S4: Also obtain the patient's basic information, input the patient's basic information, the feature value of the target blood vessel, the three-dimensional image, and the simulated surgical operation data corresponding to each of the selected surgical options into the prediction model, and obtain and use the selected surgical option with the highest success rate as the target surgical option.

[0048] Specifically, since the patient's basic information is a high-risk indicator affecting surgery for upper and lower extremity arteriosclerosis obliterans, in order to accurately predict the success rate of each surgical option for lower extremity arteriosclerosis obliterans, the prediction model is not only fed with the characteristic values ​​of the target vessel, the three-dimensional image of the target vessel, and the simulated surgical data corresponding to each surgical option, but also with the patient's basic information. The prediction model outputs the success rate of each surgical option and selects the option with the highest success rate as the target surgical option. This target surgical option is then used as a reference for actual surgery. Through this technical solution, a target surgical option with a high success rate suitable for the patient can be obtained, providing a reference for medical staff to perform surgery.

[0049] Further, step S1 includes:

[0050] Step S11: Obtain lower limb arterial angiography images through the camera unit, and obtain blood flow information based on the pixel values ​​corresponding to the angiography area in the angiography image. The blood flow information includes the blood flow path and the width of the blood flow path.

[0051] Step S12: Based on the blood flow information, obtain the target location where the blood flow path is broken or the width of the blood flow path is narrowed, and obtain the blood vessel contour corresponding to the target location based on the center line of the blood vessel before and after the target location, the blood vessel trend, the blood vessel contour and the blood flow information;

[0052] Step S13: Based on the blood flow information and the corresponding vascular contour in each angiographic image of the lower limb artery, construct the three-dimensional vascular model of the lower limb artery.

[0053] Specifically, the aforementioned imaging unit acquires the aforementioned lower limb arterial angiography images. Furthermore, based on the pixel values ​​of the angiography area in the angiography images, blood flow information is obtained, i.e., the pixel brightness value of the contrast agent in the blood as presented in the angiography images. Based on the aforementioned blood flow information, the blood flow path and width are obtained to identify target locations where blood flow is interrupted or the blood flow path width narrows, i.e., locations of vascular occlusion or stenosis. The centerline, trend, and contour of the blood vessel before and after the target location are extended, and the blood vessel contour at the target location is obtained based on the blood flow path of the target blood vessel. Similarly, the blood vessel contour at each location is obtained. Based on the aforementioned blood flow information and blood vessel contour in each of the aforementioned lower limb arterial angiography images, the aforementioned three-dimensional vascular model is constructed. Through this technical solution, even when blood vessels are narrowed or occluded, accurate blood vessel contours and three-dimensional vascular models can be obtained, laying the foundation for facilitating the calculation of feature values ​​of blood vessel segments or branches using the aforementioned three-dimensional vascular model.

[0054] Further, step S2 includes:

[0055] Step S21: Obtain the tree structure of the lower limb artery based on the three-dimensional vascular model, and divide the lower limb artery into multiple vascular branches and vascular segments according to the top-to-bottom direction of the tree structure;

[0056] Step S22: Obtain the vessel diameter based on the vessel contour of each vessel branch or vessel segment, calculate the ratio of the blood flow path width to the vessel diameter in the corresponding blood flow information, and also calculate the number of pixels on the vessel surface of the vessel branch or vessel segment whose pixel value is greater than a set threshold, and use the ratio and the number of pixels as the feature value of the vessel branch or vessel segment;

[0057] Specifically, the tree-like structure of the lower limb arteries is obtained through the aforementioned three-dimensional vascular model, representing the overall structure of the lower limb arteries. This tree-like structure is then used to divide the lower limb arteries into multiple branches and segments from top to bottom, facilitating the analysis of the characteristics of each branch or segment. By calculating the ratio of the blood flow path width to the vessel diameter for each branch or segment, it is determined whether the branch or segment is blocked or has plaque. Furthermore, the number of pixels on the vessel surface corresponding to each branch or segment with pixel values ​​greater than a set threshold (i.e., the number of pixels in calcified areas) is calculated. This ratio and the number of pixels are used as feature values ​​for analyzing the branch or segment. Through this technical solution, the acquisition of these feature values ​​lays the foundation for identifying target vessels.

[0058] Step S23: When the ratio is less than or equal to a first preset value, the vascular branch or the vascular segment is designated as a first segment, wherein the first segment is an occluded segment, and the length of the occluded segment is obtained; when the ratio is greater than the first preset value and less than or equal to a second preset value, the vascular branch or the vascular segment is designated as a second segment, and the size of the plaque in the second segment is calculated based on the length of the second segment; when the number of pixels is greater than a third preset value, the vascular branch or the vascular segment is designated as a third segment.

[0059] Step S24: The first segment, the second segment, and the third segment are taken as the target blood vessel, wherein the first segment, the second segment, or the third segment may overlap, and the feature values ​​of the target blood vessel further include the length of the occluded segment in the first segment, the size of the plaque in the second segment, the number of pixels in the third segment, and the position and shape of the target blood vessel.

[0060] Specifically, when the ratio is less than or equal to the first preset value (i.e., when the ratio is very small), the vascular branch or vascular segment is designated as the first segment, i.e., the vascular occlusion segment, and the length of the occlusion segment is calculated. When the ratio is greater than the first preset value and less than or equal to the second preset value (i.e., plaque exists in the blood vessel), the vascular branch or vascular segment is designated as the second segment, and the plaque size is calculated based on the length of the second segment, where the length of the second segment is the length of the plaque, and the product of the ratio and the cross-sectional area of ​​the blood vessel is the cross-sectional area of ​​the plaque. When the ratio is greater than the second preset value, the blood vessel is considered a normal blood vessel, and the number of pixels is also greater than the third preset value. The aforementioned vascular branch or segment is designated as the third segment, i.e., the calcified segment. The first, second, and third segments are designated as the target vessel. The occlusion length in the first segment, the plaque size in the second segment, or the number of pixels in the third segment are added to the feature values ​​of the target vessel. The position and shape of the target vessel are also added to the feature values. Since the target vessel may have both plaque and calcification or both occlusion and calcification, the first and third segments or the second and third segments may overlap. Through the above technical solution, the feature values ​​of the target vessel can be accurately obtained, providing data for accurately predicting the success rate of surgery before operation.

[0061] Further, step S3 includes:

[0062] Step S31: Obtain the three-dimensional image of the target blood vessel based on its location and the three-dimensional model of the blood vessel;

[0063] Step S32: Based on the three-dimensional image of the target blood vessel and the feature value corresponding to the target blood vessel, select a matching candidate surgical plan from the lower limb artery surgical plan template stored in the database, and simulate the surgical operation according to the feature value and the three-dimensional image through the simulation unit based on the candidate surgical plan, and obtain simulated surgical operation data, wherein the simulated surgical operation data includes surgical operation process data, surgical operation location and surgical tool information.

[0064] Specifically, the three-dimensional image of the target blood vessel can be obtained through its location and three-dimensional model. Based on this image and its feature values, a matching surgical plan is selected from lower limb artery surgical plan templates stored in the database. This selection is based on the target blood vessel's disease type, disease data, location, and shape. Disease types include occlusion, plaque, and calcification. Disease data includes the occlusion length, plaque size, and calcification area size from the feature values. The surgical plan includes the implementation method of the surgery, such as forward recanalization, reverse recanalization, or bidirectional recanalization. A simulation unit simulates the surgical operation data for each selected surgical plan on the three-dimensional image of the target blood vessel based on the feature values. This technical solution allows for the acquisition of simulated surgical operation data for each selected surgical plan, laying the foundation for selecting a target surgical plan from the available options.

[0065] Further, step S4 includes:

[0066] The basic information of the patient is obtained, and the patient's basic information, the feature value of the target blood vessel, the three-dimensional image, and the simulated surgical operation data corresponding to each of the proposed surgical options are input into the prediction model to obtain the success rate of each proposed surgical option. The proposed surgical option with the highest success rate is selected as the target surgical option, and the target surgical option and the surgical tools in the target surgical option are used as the reference for actual surgery. The patient's basic information includes: the patient's age, gender, blood pressure, blood glucose, blood lipids, high homocysteine ​​level, renal function, and inflammatory markers.

[0067] Specifically, the success of lower extremity arteriosclerosis obliterans surgery depends not only on the surgical procedure but also on the patient's basic information. This information is a high-risk indicator for lower extremity arteriosclerosis obliterans surgery; exceeding these high-risk indicators can cause numerous complications, leading to poor postoperative rehabilitation and even surgical failure. To accurately predict the success rate of each potential surgical option for lower extremity arteriosclerosis obliterans surgery, the model uses not only the characteristic values ​​of the target vessel, its three-dimensional image, and simulated surgical data for each potential surgical option as input data, but also the patient's basic information. The model outputs the success rate of each potential surgical option and selects the option with the highest success rate as the target surgical option, which serves as a reference for actual surgery. The surgical tools include at least the type, thickness, and brand of guidewires and a first balloon catheter. This technical approach allows for the identification of a target surgical option with a high success rate suitable for the patient, providing a reference for medical personnel during surgery.

[0068] Furthermore, the training of the prediction model includes:

[0069] Step S310: Obtain all lower extremity arterial surgery data from the database. The surgical data includes the patient's basic information, a first angiography three-dimensional image, the feature value of the first angiography three-dimensional image, first surgical operation data, and first annotation information. The surgical data is used as the first sample data, wherein the first annotation information is the surgical success result corresponding to the first sample data.

[0070] Step S311: Cluster the surgical data according to the basic information of the corresponding patients and obtain multiple clusters. Divide the first three-dimensional contrast imaging image of the first sample data in each cluster into multiple sub-images according to the corresponding feature values. Flip and combine any two or more sub-images to obtain multiple second three-dimensional contrast imaging images. Combine the sub-images in the second three-dimensional contrast imaging images with the sub-operation data in the first surgical operation data to obtain second surgical operation data. Determine the compatibility between the corresponding sub-operation data in the second surgical operation data.

[0071] Step S312: When the sub-operation data in the second surgical operation data are compatible, the mean value of the patient information corresponding to the sub-image is used as the new patient information, and the new patient information, the second angiography data, the feature value of the second angiography data, the second surgical operation data and the second annotation information are used as the second sample data, wherein the second annotation information indicates that the surgery was successful. The first sample data and the second sample data are used as training data to train the prediction model.

[0072] Specifically, during the training of the aforementioned prediction model, the more training data available, the higher the accuracy of the prediction model. Since the sample data in the database is limited, to obtain richer sample data, the lower limb arterial surgery data of all patients stored in the database, i.e., the first sample data, is further clustered according to the patient's basic information to obtain multiple clusters. Because different patients have different basic information (health indicators), different surgical plans and procedures, and different surgical outcomes, clustering is used to obtain clusters with similar patient information. The first angiography 3D image in each cluster is then divided into multiple sub-images according to the corresponding feature values. Any two or more sub-images are flipped and combined to obtain the corresponding second angiography 3D image. Furthermore, the sub-surgical operation data corresponding to the sub-images in the second angiography 3D image are combined to obtain the second surgical operation data. Based on the compatibility of the sub-surgical operation data in the second surgical operation data, the second sample data is obtained, for example: the first sub-image... When the first sub-image is combined with the second sub-image to form the second angiographic 3D image, the first sub-image is an image of an occluded vessel, and the second sub-image is an image of a plaque vessel. Both operations are unilateral patency opening. The positions and shapes of the vessels corresponding to the first and second sub-images do not conflict. The first sub-operation data corresponding to the first sub-image and the second sub-operation data corresponding to the second sub-image can be fused. That is, the first sub-operation data or the second sub-operation data can be used to complete the surgery of the disease type in the second angiographic 3D image. At this time, the self-operation data in the second surgical operation data are considered to be compatible, and the above-mentioned second annotation information is marked as successful surgery. The annotation information of the above-mentioned first sample data is also marked as successful surgery. The above-mentioned first sample data and the above-mentioned second sample data are used as training data to train the above-mentioned prediction model. When the above-mentioned sub-operation data are incompatible, since there may be other operation methods that are different from the above-mentioned first operation data and second operation data to successfully complete the above-mentioned surgery, it cannot be directly assumed that the surgery will definitely fail. Through the above technical solution, a more accurate prediction model can be obtained, thereby improving the prediction accuracy of the surgical success rate.

[0073] Furthermore, step S4 also includes:

[0074] The basic information of the patient, the feature values ​​of the target blood vessel, the three-dimensional image of the target blood vessel, and the surgical operation data corresponding to the target surgical plan are used as input data for the prediction model. Other items in the patient's basic information in the input data, excluding age and gender, are used as adjustment items. For each adjustment item, a first value and a second value are obtained within a first range and a second range, respectively. The adjusted first and second values ​​are then used to replace the corresponding values ​​in the patient's basic information in the input data. The replaced input data is then input into the prediction model to obtain a first success rate and a second success rate. The impact of each adjustment item on the success rate is obtained based on the difference between the first and second success rates. Specifically, the first range is greater than the value of the adjustment item in the patient's basic information, the second range is less than the value of the adjustment item in the patient's basic information, and the difference between the first and second values ​​and the value of the adjustment item in the patient's basic information is equal to the difference between them.

[0075] Specifically, to obtain more precise postoperative care points and prolong the patency time of stent placement during surgery, the values ​​of all adjustable items in the patient's basic information, excluding age and gender (which are not adjustable), are adjusted one by one. The first and second values ​​of each adjusted item are then used to replace the corresponding values ​​in the patient's basic information. The input data of the replaced adjusted item values ​​are then input into the prediction model to obtain a first success rate and a second success rate. The first and second success rates are compared with the first and second differences in the success rate corresponding to the target surgical plan obtained in step S4, respectively. A third difference between the first and second differences is used to determine the impact of the adjusted item on the surgical success rate. A larger third difference indicates a greater impact, and vice versa. Through this technical solution, the impact of each adjusted item on the success rate of lower extremity arterial surgery is obtained, thus identifying the factors influencing the surgical success rate. This information is used to guide postoperative rehabilitation care, thereby further improving the surgical success rate.

[0076] Furthermore, step S4 is followed by step S5:

[0077] The imaging unit acquires data on whether the catheter passes through the true lumen of the target blood vessel during the procedure, and obtains a first score. If the catheter passes through the true lumen, the first score is positive; if it does not pass through the true lumen, the first score is negative. The system also acquires data on whether there is dissection before the placement of the stent or drug-coated balloon, and obtains a second score. If there is dissection, the second score is negative; if there is no dissection, the second score is positive. The final success rate is the sum of the product of the first score and its corresponding weight, the product of the second score and its corresponding weight, and the success rate corresponding to the target surgical plan.

[0078] Specifically, during surgery, determining whether the guidewire passes through the true lumen is crucial, as it relates to the safety and effectiveness of the procedure. The true lumen refers to the original lumen of the blood vessel, while the false lumen is an abnormal cavity formed after damage to the vascular intima. Whether there is dissection before placing a stent or DCB (drug-coated balloon) is also an important indicator for evaluating the safety of the procedure. Therefore, the sum of the first score and the second score mentioned above, along with their corresponding weights and the success rate mentioned above, is used as the final success rate. Through the above technical solution, a more accurate success rate for the lower extremity arterial surgery can be obtained.

[0079] Furthermore, the prediction model is a random forest model.

[0080] This invention also provides a lower extremity arterial surgical intervention assessment system based on image data, the system being used to implement the above-described method, such as... Figure 2 As shown, the system includes:

[0081] The camera unit is used to acquire lower limb arterial angiography images of the patient, obtain blood flow information based on the pixel values ​​corresponding to the angiography area in the angiography image, and obtain a complete vascular contour and 3D vascular model based on the changes in the blood flow information.

[0082] The extraction unit is used to obtain the tree structure of the lower limb arteries based on the three-dimensional vascular model, divide the three-dimensional vascular model into multiple vascular branches and vascular segments from top to bottom according to the tree structure, extract the feature value of each vascular branch and vascular segment, and obtain the target vascular based on the feature value;

[0083] The simulation unit is used to obtain a three-dimensional image of the target blood vessel through the three-dimensional blood vessel model, obtain multiple surgical options to be selected from the database based on the three-dimensional image of the target blood vessel and the feature values, and obtain corresponding simulated surgical operation data by simulating each of the surgical options to be selected through the simulation unit.

[0084] The prediction unit is used to acquire the patient's basic information, input the patient's basic information, the feature value of the target blood vessel, the three-dimensional image, and the simulated surgical operation data corresponding to each of the proposed surgical options into the prediction model, and acquire and select the proposed surgical option with the highest success rate as the target surgical option.

[0085] In summary, this invention obtains blood flow information through the pixel values ​​of the angiography area in the aforementioned angiography image, and obtains the blood flow path and width based on the blood flow information to identify the target location where blood flow is interrupted or the blood flow path width narrows. It extends the blood vessel's centerline, trend, and contour before and after the target location, and obtains the blood vessel contour and three-dimensional model of the target location based on the blood flow path of the target vessel. Furthermore, it divides the lower limb arteries into multiple branches and segments using a tree-like structure from top to bottom, determines whether the branches or segments are blocked or have plaques, and identifies the number of pixels in calcified areas, and obtains the feature values ​​of the target vessel. It also uses the three-dimensional image of the target vessel and its feature values ​​to... The database stores lower extremity arterial surgery templates, from which matching candidate surgical plans are selected. A simulation unit then simulates the surgical operation data for each candidate plan on a 3D image of the target blood vessel based on feature values. The model uses not only the feature values ​​of the target blood vessel, its 3D image, and the simulated surgical data for each candidate plan as input data, but also the patient's basic information. The model outputs the success rate of each candidate plan and selects the plan with the highest success rate as the target plan. Through the combined use of these techniques, the accuracy of predicting the success rate of lower extremity arterial surgery is improved, and the target plan with the highest success rate is obtained.

[0086] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0087] The above-described embodiments are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

[0088] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for evaluating lower extremity arterial surgical intervention based on imaging data, characterized in that, The method includes the following steps: Step S1: Acquire lower limb arterial angiography images of the patient through the camera unit, obtain blood flow information based on the pixel values ​​corresponding to the angiography area in the angiography image, and obtain a complete vascular contour and 3D vascular model based on the changes in the blood flow information; Step S2: Obtain the tree structure of the lower limb arteries based on the three-dimensional vascular model, divide the three-dimensional vascular model into multiple vascular branches and vascular segments from top to bottom according to the tree structure, extract the feature values ​​of each vascular branch and vascular segment, and obtain the target blood vessel based on the feature values; Step S3: Obtain a three-dimensional image of the target blood vessel through the three-dimensional blood vessel model; obtain multiple surgical options from the database based on the three-dimensional image of the target blood vessel and the feature values; and obtain corresponding simulated surgical operation data by simulating each of the surgical options through a simulation unit. Step S4: Also obtain the patient's basic information, input the patient's basic information, the feature value of the target blood vessel, the three-dimensional image, and the simulated surgical operation data corresponding to each of the selected surgical options into the prediction model, and obtain and use the selected surgical option with the highest success rate as the target surgical option.

2. The method according to claim 1, characterized in that, Step S1 includes: Step S11: Obtain lower limb arterial angiography images through the camera unit, and obtain blood flow information based on the pixel values ​​corresponding to the angiography area in the angiography image. The blood flow information includes the blood flow path and the width of the blood flow path. Step S12: Based on the blood flow information, obtain the target location where the blood flow path is broken or the width of the blood flow path is narrowed, and obtain the blood vessel contour corresponding to the target location based on the center line of the blood vessel before and after the target location, the blood vessel trend, the blood vessel contour and the blood flow information; Step S13: Based on the blood flow information and the corresponding vascular contour in each angiographic image of the lower limb artery, construct the three-dimensional vascular model of the lower limb artery.

3. The method according to claim 1, characterized in that, Step S2 includes: Step S21: Obtain the tree structure of the lower limb artery based on the three-dimensional vascular model, and divide the lower limb artery into multiple vascular branches and vascular segments according to the top-to-bottom direction of the tree structure; Step S22: Obtain the vessel diameter based on the vessel contour of each vessel branch or vessel segment, calculate the ratio of the blood flow path width to the vessel diameter in the corresponding blood flow information, and also calculate the number of pixels on the vessel surface of the vessel branch or vessel segment whose pixel value is greater than a set threshold, and use the ratio and the number of pixels as the feature value of the vessel branch or vessel segment; Step S23: When the ratio is less than or equal to a first preset value, the vascular branch or the vascular segment is designated as a first segment, wherein the first segment is an occluded segment, and the length of the occluded segment is obtained; when the ratio is greater than the first preset value and less than or equal to a second preset value, the vascular branch or the vascular segment is designated as a second segment, and the size of the plaque in the second segment is calculated based on the length of the second segment; when the number of pixels is greater than a third preset value, the vascular branch or the vascular segment is designated as a third segment. Step S24: The first segment, the second segment, and the third segment are taken as the target blood vessel, wherein the first segment, the second segment, or the third segment may overlap, and the feature values ​​of the target blood vessel further include the length of the occluded segment in the first segment, the size of the plaque in the second segment, the number of pixels in the third segment, and the position and shape of the target blood vessel.

4. The method according to claim 1, characterized in that, Step S3 includes: Step S31: Obtain the three-dimensional image of the target blood vessel based on its location and the three-dimensional model of the blood vessel; Step S32: Based on the three-dimensional image of the target blood vessel and the feature value corresponding to the target blood vessel, select a matching candidate surgical plan from the lower limb artery surgical plan template stored in the database, and simulate the surgical operation according to the feature value and the three-dimensional image through the simulation unit based on the candidate surgical plan, and obtain simulated surgical operation data, wherein the simulated surgical operation data includes surgical operation process data, surgical operation location and surgical tool information.

5. The method according to claim 1, characterized in that, Step S4 includes: The basic information of the patient is obtained, and the patient's basic information, the feature value of the target blood vessel, the three-dimensional image, and the simulated surgical operation data corresponding to each of the selected surgical options are input into the prediction model to obtain the success rate of each selected surgical option. The selected surgical option with the highest success rate is taken as the target surgical option, and the target surgical option and the surgical tools in the target surgical option are used as the reference for actual surgery. The patient's basic information includes the patient's age, gender, blood pressure, blood glucose, blood lipids, high homocysteine ​​level, renal function, and inflammatory markers.

6. The method according to claim 1, characterized in that, The training of the prediction model includes: Step S310: Obtain all lower extremity arterial surgery data from the database. The surgical data includes the patient's basic information, a first angiography three-dimensional image, the feature value of the first angiography three-dimensional image, first surgical operation data, and first annotation information. The surgical data is used as the first sample data, wherein the first annotation information is the surgical success result corresponding to the first sample data. Step S311: Cluster the surgical data according to the basic information of the corresponding patients and obtain multiple clusters. Divide the first three-dimensional contrast imaging image of the first sample data in each cluster into multiple sub-images according to the corresponding feature values. Flip and combine any two or more sub-images to obtain multiple second three-dimensional contrast imaging images. Combine the sub-images in the second three-dimensional contrast imaging images with the sub-operation data in the first surgical operation data to obtain second surgical operation data. Determine the compatibility between the corresponding sub-operation data in the second surgical operation data. Step S312: When the sub-operation data in the second surgical operation data are compatible, the mean value of the patient information corresponding to the sub-image is used as the new patient information, and the new patient information, the second angiography data, the feature value of the second angiography data, the second surgical operation data and the second annotation information are used as the second sample data, wherein the second annotation information indicates that the surgery was successful. The first sample data and the second sample data are used as training data to train the prediction model.

7. The method according to claim 1, characterized in that, Step S4 further includes: using the patient's basic information, the feature value of the target blood vessel, the three-dimensional image of the target blood vessel, and the surgical operation data corresponding to the target surgical plan as input data for the prediction model; using items other than age and gender in the patient's basic information in the input data as adjustment items; obtaining a first value and a second value for each adjustment item within a first range and a second range; replacing the corresponding values ​​in the patient's basic information in the input data with the adjusted first value and second value; inputting the replaced input data into the prediction model; obtaining a first success rate and a second success rate; and obtaining the impact of each adjustment item on the success rate based on the difference between the first success rate and the second success rate, wherein the first range is greater than the value of the adjustment item in the patient's basic information, the second range is less than the value of the adjustment item in the patient's basic information, and the difference between the first value and the second value and the value of the adjustment item in the patient's basic information is equal to the difference between the first value and the second value.

8. The method according to claim 1, characterized in that, Step S4 is followed by step S5: The imaging unit acquires data on whether the catheter passes through the true lumen of the target blood vessel during the procedure, and obtains a first score. If the catheter passes through the true lumen, the first score is positive; if it does not pass through the true lumen, the first score is negative. The system also acquires data on whether there is dissection before the placement of the stent or drug-coated balloon, and obtains a second score. If there is dissection, the second score is negative; if there is no dissection, the second score is positive. The final success rate is the sum of the product of the first score and its corresponding weight, the product of the second score and its corresponding weight, and the success rate.

9. The method according to claim 1, characterized in that, The prediction model is a random forest model.

10. A lower extremity arterial surgical intervention assessment system based on image data, the system being used to implement the method as described in any one of claims 1-9, characterized in that, The system includes: The camera unit is used to acquire lower limb arterial angiography images of the patient, obtain blood flow information based on the pixel values ​​corresponding to the angiography area in the angiography image, and obtain a complete vascular contour and 3D vascular model based on the changes in the blood flow information. The extraction unit is used to obtain the tree structure of the lower limb arteries based on the three-dimensional vascular model, divide the three-dimensional vascular model into multiple vascular branches and vascular segments from top to bottom according to the tree structure, extract the feature value of each vascular branch and vascular segment, and obtain the target vascular based on the feature value; The simulation unit is used to obtain a three-dimensional image of the target blood vessel through the three-dimensional blood vessel model, obtain multiple surgical options to be selected from the database based on the three-dimensional image of the target blood vessel and the feature values, and obtain corresponding simulated surgical operation data by simulating each of the surgical options to be selected through the simulation unit. The prediction unit is used to acquire the patient's basic information, input the patient's basic information, the feature value of the target blood vessel, the three-dimensional image, and the simulated surgical operation data corresponding to each of the proposed surgical options into the prediction model, and acquire and select the proposed surgical option with the highest success rate as the target surgical option.

Images