Method for constructing arteritis aorta vascular remodeling prediction model

Abstract

The invention discloses a method for constructing a takayasu arteritis vascular remodeling prediction model, and the method comprises the steps: S1, injecting a tracer agent, and carrying out PET-MRI combined imaging and CE-MRA imaging; s2, multi-modal scoring is carried out, and each MRA scoring result is obtained; s3, comparing with a comparison item to obtain each PET-VAS score; and S4, determining each prediction factor and each beta coefficient, and constructing a DEPM prediction model. After the < 18 > F-FDG tracer agent is injected, a systemic artery related image of a human body is obtained through PET-MRI combined imaging and CE-MRA imaging, and analysis of takayasu arteritis is facilitated; the liver is used as a comparison item, so that the comparison effect is more obvious; and for the plurality of prediction factors, a joyon index is obtained by using an ROC curve to calculate a cutoff value, then single-factor analysis and Cox multi-factor analysis are carried out, optimal analysis data of the plurality of prediction factors are determined, and the reliability of the model is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of Takayasu arteritis vascular remodeling prediction, and in particular to a method for constructing a Takayasu arteritis vascular remodeling prediction model. Background Art

[0002] Takayasu's arteritis, or TAK for short, is a chronic, progressive vascular inflammatory disease that primarily affects the aorta and its main branches. In the late stage, it can lead to changes in vascular structure, such as vascular stenosis, occlusion, and aneurysm formation, i.e., vascular remodeling. Takayasu's arteritis is one of the diseases that seriously affects the life and health of young people. The main patient population is young women, and about 90% of patients develop the disease before the age of 30.

[0003] Vascular remodeling is a key factor leading to poor prognosis in patients with Takayasu arteritis. Therefore, accurately predicting the risk factors for vascular remodeling in Takayasu arteritis is crucial for early identification of high-risk patients, optimization of treatment strategies, and improvement of patient prognosis. The discovery of vascular wall inflammation and structural damage depends on vascular imaging examinations. At present, there is no unified standard for vascular imaging evaluation methods for Takayasu arteritis at home and abroad. The reason is that patients with Takayasu arteritis may have early wall inflammation and late vascular structural changes in multiple large blood vessels throughout the body at the same time, and the commonly used vascular imaging methods in clinical practice cannot provide an integrated assessment of large blood vessels throughout the body at different stages of the disease course. For example, vascular ultrasound can only evaluate superficial blood vessels in segments, and the exploration of the descending thoracic aorta and distal aortic branches is limited, and it is highly dependent on the operator's technology; CTA is computed tomography angiography, which requires segmented detection of large blood vessels throughout the body. Patients must undergo multiple injections of contrast agents, which increases the burden on the kidneys and is not suitable for routine follow-up; MRA is magnetic resonance angiography, which can show vascular wall thickening, edema, and changes in lumen morphology, but similar to CTA, it requires segmented scanning.

[0004] In recent years, PET-CT has emerged. The full name is Positron Emission Tomography - Computed Tomography, which is positron emission tomography-computed tomography. This imaging combines the two technologies of PET and CT. PET measures the metabolic activity of cells in the body by injecting radioactive tracers into the body, while CT provides detailed images of anatomical structures. PET-CT is useful for evaluating vascular inflammation because it can detect abnormal metabolic activity at an early stage, which may be a manifestation of inflammation. However, for the assessment of specific vascular structural damage, the resolution of PET-CT is not as good as CTA and MRA; when PET-CT uses CT, it uses X-rays, which involve ionizing radiation exposure and pose a certain degree of harm to the human body. Frequent use will also increase the risk of cancer.

[0005] In other words, there are at least two problems in the current scanning assessment measures for Takayasu arteritis: 1) There is a lack of comprehensive assessment methods for the whole body. Different methods need to be used for different parts of the human body to perform segmented scanning and judgment, which is a cumbersome process; 2) Some technologies may cause harm when scanning the human body, especially those involving CT technology, which have greater risks when used frequently. Summary of the invention

[0006] In view of the above two problems, the purpose of the present invention is to propose a method for constructing a prediction model of vascular remodeling in large arteritis. After injecting 18F-FDG tracer, PET-MRI combined imaging and CE-MRA imaging can effectively obtain images related to the human body's systemic arteries, which is convenient for subsequent large arteritis analysis; at the same time, the selected multimodality is the artery that is prone to large arteritis, and the MRA scoring results and PET-VAS vascular activity scores are obtained respectively, which can be effectively used to assist in judging the condition and subsequently constructing a prediction model; the liver, which is metabolically active and not affected by large arteritis, is used as a comparison item in the scoring, so that the comparison effect is more obvious; in addition, the Youden index is taken from the ROC curve to calculate the cutoff value for multiple predictive factors, and then univariate analysis and Cox multivariate analysis are performed, so that the best analysis data for multiple predictive factors can be determined and the stability and reliability of the model can be improved.

[0007] This is achieved through the following technical solutions: A method for constructing a prediction model for vascular remodeling in Takayasu arteritis comprises the following steps: S1. Inject tracer, perform PET-MRI combined imaging and CE-MRA imaging: first, inject 18F-FDG tracer into multiple patients, perform PET-MRI combined imaging on the whole body of each patient, and obtain each corresponding first image; then, inject gadolinium contrast agent into each patient, perform CE-MRA imaging, and obtain each corresponding second image; S2. Perform multimodal scoring to obtain each MRA scoring result: Based on the vascular wall thickness, vascular wall edema and vascular wall enhancement, perform multimodal scoring on each second image in step S1 to obtain each corresponding MRA scoring result; wherein the multiple modalities of the multimodal evaluation include: bilateral common carotid arteries and their branches, bilateral vertebral arteries, bilateral axillary arteries, bilateral subclavian arteries, brachiocephalic trunk, ascending aorta, aortic arch, descending aorta, abdominal aorta, celiac trunk, mesenteric artery, bilateral renal arteries and bilateral common iliac arteries and their branches; S3, comparing each modality with the comparison item to obtain each PET-VAS vascular activity score: obtaining the uptake value of the 18F-FDG tracer by the liver of each patient and setting it as the corresponding comparison item; for each modality in step S2, using the uptake value of the 18F-FDG tracer by the vessel wall of each modality as a judgment basis and comparing and evaluating it with each corresponding comparison item to obtain the PET-VAS vascular activity score of each first image; S4. Determine each predictive factor and each β coefficient, and construct a DEPM prediction model: obtain the ESR sedimentation rate and delayed diagnosis time of each patient, use each ESR sedimentation rate, each delayed diagnosis time, each PET-VAS vascular activity score and each MRA score result of each patient as predictive factors, set each corresponding β coefficient for weighting for each predictive factor, weight each predictive factor and each β coefficient, and establish a DEPM prediction model for vascular remodeling in large artery inflammation.

[0008] After the injection of 18F-FDG tracer, the combined PET-MRI imaging and CE-MRA imaging can be used to effectively obtain images related to the human body's systemic arteries, which is convenient for subsequent analysis of large arteritis. At the same time, the selected multimodalities are arteries that are prone to large arteritis. The MRA scoring results and PET-VAS vascular activity scores are obtained respectively, which can be effectively used to assist in the judgment of the disease and the subsequent construction of a prediction model. The liver, which is metabolically active and not affected by large arteritis, is also used as a comparison item in the scoring, making the comparison effect more obvious.

[0009] Preferably, in step S1, the injection amount of 18F-FDG tracer for any patient is 0.12 mCi / kg, and the injection amount of gadolinium contrast agent for any patient is 0.4 ml / kg. The injection amounts of 18F-FDG tracer and gadolinium contrast agent both effectively ensure the imaging effect, while also avoiding over-injection that may cause adverse effects on the patient's body.

[0010] Preferably, in step S1, when CE-MRA imaging is performed, the imaging range includes the neck vessels and extends to the common iliac artery branches and intracranial blood vessels. The neck vessels and extend to the common iliac artery branches and intracranial blood vessels cover the main arteries of the human body, ensuring the effectiveness and comprehensiveness of the imaging.

[0011] Preferably, in step S2, when multimodal scoring is performed, a blood vessel wall thickness of >3 mm corresponds to 1 point, a blood vessel wall thickness of ≤3 mm corresponds to 0 point; the presence of vessel wall edema corresponds to 1 point, and the absence of vessel wall edema corresponds to 0 point. Scoring based on the vessel wall thickness and the presence of edema can effectively determine the inflammatory condition of the vessel wall.

[0012] Preferably, when performing comparative evaluation in step S3, when the vessel wall of any modality has no uptake of 18F-FDG tracer, it is scored as 0 points; when the vessel wall of any modality has uptake of 18F-FDG tracer but the uptake value is less than the corresponding comparison item, it is scored as 1 point; when the uptake value of 18F-FDG tracer by the vessel wall of any modality is equal to the corresponding comparison item, it is scored as 2 points; when the uptake value of 18F-FDG tracer by the vessel wall of any modality is greater than the corresponding comparison item, it is scored as 3 points; each score of multiple modalities corresponding to each first image after the comparison and evaluation is completed is added together as the PET-VAS vascular activity score corresponding to each first image. Comparative scoring based on the uptake value of the vessel wall and the corresponding comparison item can accurately and objectively judge the vascular activity of the vessel wall, thereby improving accuracy and reliability.

[0013] Preferably, when setting each β coefficient in step S4, the weight of each prediction factor is obtained by the maximum likelihood estimation method, and the size of each weight is the corresponding β coefficient. The maximum likelihood method can effectively predict each optimal regression coefficient, that is, the β coefficient, and is widely used and easy to use.

[0014] Preferably, when performing comparative evaluation in step S3, the maximum standardized uptake value SUVmax1 of the vessel wall of any modality for the 18F-FDG tracer is first obtained, and then the maximum standardized uptake value SUVmax2 of the corresponding liver for the 18F-FDG tracer is obtained, and a score is performed according to the size of SUVmax1 and SUVmax2. The maximum standardized uptake value can effectively reflect the absorption of the tracer, and thus can be used for evaluation.

[0015] Preferably, before setting each β coefficient in step S4, a univariate analysis is first performed on each predictor to screen out multiple core factors, and then a Cox multivariate analysis is performed on each core factor screened out by the univariate analysis to determine each β coefficient of each core factor. Performing a univariate analysis first can effectively screen out the core factors related to the patient's large arteritis, and then performing a Cox multivariate analysis on the core factors can improve the accuracy of the model.

[0016] Preferably, before performing univariate analysis, the ESR sedimentation rate, PET-VAS vascular activity score and MRA score results of each patient are respectively calculated using the Youden index through the ROC curve to calculate the cutoff value, and a univariate analysis is performed on each cutoff value. Calculating the cutoff value by taking the Youden index through the ROC curve can make the data more representative and accurate.

[0017] Preferably, in step S4, the predictive factors further include gender, age and C-reactive protein. Gender, age and C-reactive protein are also parameters related to Takayasu arteritis and can be used for auxiliary verification.

[0018] Compared with the prior art, the present invention has the following beneficial effects: The technical solution of the present invention can effectively obtain images related to the arteries of the human body systemically through PET-MRI combined imaging and CE-MRA imaging after injecting 18F-FDG tracer, which is convenient for subsequent analysis of large arteritis; at the same time, the selected multi-modality is the artery related to the disease prone to large arteritis, and the MRA scoring result and PET-VAS vascular activity score are obtained respectively, which can be effectively used to assist in judging the disease and subsequently constructing a prediction model; the liver with active metabolism and not affected by large arteritis is used as a comparison item in the scoring, so that the comparison effect is more obvious; in addition, the Youden index is taken from the ROC curve to calculate the cutoff value for multiple prediction factors, and then the univariate analysis and Cox multivariate analysis are performed, so that the best analysis data of multiple prediction factors can be determined and the stability and reliability of the model can be improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 Flow chart of a method for constructing a prediction model for vascular remodeling in large arteritis. DETAILED DESCRIPTION

[0020] The following will be combined with the attached embodiment of the present invention Figure 1 , the technical solutions in the embodiments of the present invention are described in detail.

[0021] like Figure 1 As shown, it is a flowchart of a method for constructing a prediction model of vascular remodeling in Takayasu arteritis. The prediction model of vascular damage in Takayasu arteritis is constructed based on multimodal hybrid imaging. Through PET-MRI combined imaging and CE-MRA imaging, multimodal hybrid imaging is performed on the arterial areas of the human body that may have Takayasu arteritis, which can effectively provide the data basis of the model, thereby establishing a DEPM prediction model with clear scoring standards. The method specifically includes the following steps: S1. Inject tracers and perform PET-MRI combined imaging and CE-MRA imaging: First, inject 18F-FDG tracers into multiple patients, perform PET-MRI combined imaging on the whole body of each patient, and obtain each corresponding first image; then inject gadolinium contrast agent into each patient, perform CE-MRA imaging, and obtain each corresponding second image. PET-MRI combined imaging is a combination of positron emission tomography and magnetic resonance imaging, which is suitable for situations where metabolic activity and anatomical structure need to be evaluated simultaneously; CE-MRA imaging is contrast-enhanced magnetic resonance angiography, which is a vascular imaging technology and is suitable for evaluating vascular diseases and other scenarios; based on these two types of imaging, different imaging information can be effectively obtained separately in the future, thereby obtaining comprehensive evaluation and prediction results.

[0022] In this embodiment, in step S1, the injection amount of 18F-FDG tracer for any patient is 0.12mCi / kg, and the injection amount of gadolinium contrast agent for any patient is 0.4ml / kg. Strictly limiting the injection amount of 18F-FDG tracer and gadolinium contrast agent effectively ensures the imaging effect, while also avoiding excessive injection that may cause adverse effects on the patient's body.

[0023] In this embodiment, in step S1, when CE-MRA imaging is performed, the imaging range includes the neck blood vessels and extends to the common iliac artery branches and intracranial blood vessels. The neck blood vessels and extend to the common iliac artery branches and intracranial blood vessels, covering the main arteries of the human body, ensuring the effectiveness and comprehensiveness of the imaging, and avoiding omissions when analyzing large arteritis.

[0024] S2. Perform multimodal scoring to obtain each MRA scoring result: Based on the vascular wall thickness and the edema of the vascular wall, perform multimodal scoring on each second image in step S1 to obtain each corresponding MRA scoring result; wherein the multiple modalities of the multimodal evaluation include: bilateral common carotid arteries and their branches, bilateral vertebral arteries, bilateral axillary arteries, bilateral subclavian arteries, brachiocephalic trunk, ascending aorta, aortic arch, descending aorta, abdominal aorta, celiac trunk, mesenteric artery, bilateral renal arteries and bilateral common iliac arteries and their branches.

[0025] In this embodiment, in step S2, when multimodal scoring is performed, a blood vessel wall thickness of >3 mm corresponds to 1 point, a blood vessel wall thickness of ≤3 mm corresponds to 0 point; the presence of vessel wall edema corresponds to 1 point, and the absence of vessel wall edema corresponds to 0 point. Scoring based on the vessel wall thickness and the presence of edema can effectively determine the inflammatory condition of the vessel wall.

[0026] S3. Compare each modality with the comparison item to obtain each PET-VAS vascular activity score: obtain the liver uptake value of each patient for 18F-FDG tracer and set it as the corresponding comparison item; for each modality in step S2, use the vessel wall uptake value of each modality for 18F-FDG tracer as a judgment basis and compare and evaluate it with each corresponding comparison item to obtain the PET-VAS vascular activity score of each first image.

[0027] In this embodiment, when performing comparative evaluation in step S3, scoring can be performed based on the uptake of the tracer by the vessel wall. For example, when the vessel wall of any modality does not uptake the 18F-FDG tracer, it is scored as 0 points; when the vessel wall of any modality has uptake of the 18F-FDG tracer but the uptake value is less than the corresponding comparison item, it is scored as 1 point; when the uptake value of the vessel wall of any modality is equal to the corresponding comparison item, it is scored as 2 points; when the uptake value of the vessel wall of any modality is greater than the corresponding comparison item, it is scored as 3 points; each score of the multiple modalities corresponding to each first image after the comparison and evaluation is completed is added together as the PET-VAS vascular activity score corresponding to each first image. Comparative scoring based on the uptake value of the vessel wall and the corresponding comparison item can accurately and objectively judge the vascular activity of the vessel wall, thereby improving accuracy and reliability.

[0028] When performing comparative evaluation in step S3, the maximum standardized uptake value SUVmax1 of the vessel wall of any modality for the 18F-FDG tracer can be obtained first, and then the maximum standardized uptake value SUVmax2 of the corresponding liver for the 18F-FDG tracer can be obtained, and the score is performed according to the size of SUVmax1 and SUVmax2. The maximum standardized uptake value can effectively reflect the absorption of the tracer and can be used for evaluation.

[0029] S4. Determine each predictive factor and each β coefficient, and construct a DEPM prediction model: obtain the ESR sedimentation rate and delayed diagnosis time of each patient, use each ESR sedimentation rate, each delayed diagnosis time, each PET-VAS vascular activity score and each MRA score result of each patient as predictive factors, set each corresponding β coefficient for weighting for each predictive factor, weight each predictive factor and each β coefficient, and establish a DEPM prediction model for vascular damage in large artery inflammation.

[0030] In this embodiment, before setting each β coefficient, a univariate analysis is first performed on each predictor to screen out multiple core factors, and then a Cox multifactor analysis is performed on each predictor screened out by the univariate analysis to determine each β coefficient of each core factor. The univariate analysis can be performed first to effectively screen out the core factors related to the patient's large arteritis, and then the Cox multifactor analysis is performed on the core factors to improve the accuracy of the model. The above-mentioned univariate analysis and Cox multifactor analysis can both be analyzed by the maximum likelihood estimation method. First, the weight of each predictor when estimated using the maximum likelihood estimation method is obtained separately, and multiple core factors with large weights are screened out according to a specific threshold set by oneself; then multiple core factors are included in the Cox proportional hazard model, and the maximum likelihood estimation method is continued to be used to give the weight of each core factor at this time, that is, the regression coefficient, that is, the β coefficient.

[0031] In this embodiment, in step S4, the predictive factors may also include gender, age and C-reactive protein. C-reactive protein can reflect the inflammation level of the patient's serum, and the cutoff value can be calculated by taking the Youden index through the ROC curve, which is a widely recognized and commonly used statistical method. A representative cutoff value, i.e., a cutoff point value, can be selected, and then used as a factor in subsequent factor analysis, so that the results are more comprehensive and accurate.

[0032] In addition, before performing univariate analysis, the ESR sedimentation rate, PET-VAS vascular activity score, and MRA score of each patient can also be used to calculate the cutoff value by taking the Youden index through the ROC curve, and then perform univariate analysis on each cutoff value. Calculating the cutoff value by taking the Youden index through the ROC curve can make the data more representative and accurate.

[0033] When there are many predictive factors, for example, including the above-mentioned gender, age, delayed diagnosis time, ESR sedimentation rate, PET-VAS vasoactivity score and MRA score results, first perform univariate analysis to screen out the most important core factors affecting large arteritis, and then perform Cox multivariate analysis, and set corresponding scores for different core factors to establish a DEPM prediction model. As shown in Table 1 below, it is a score table for the DEPM prediction model: Table 1: Scoring table of DEPM prediction model

[0034] To score in the scoring table, you need to first identify the one with the smallest β coefficient among multiple core factors as the basic item, and set the basic item score to 1; then, the β coefficients of other predictive factors are respectively compared with the β coefficient of the basic item, and the score N is given according to the size of the ratio N. For example: the β coefficient of delayed diagnosis time is 0.526, and the score is 1; the β coefficient of ESR is 1.4, then 1.4 / 0.526≈2, so the score of ESR is 2. Then, for any patient, the four relevant data of the patient can be judged respectively with the scoring criteria in Table 1 above, and then the scores of the four items are added together to obtain the score of the DEPM prediction model of the patient. It should be noted that the above-mentioned β coefficient is only an assumed data for illustrating the scoring of the DEPM prediction model; the full name of DEPM is Dynamic Enhanced Prediction Model, and the DEPM prediction model is the dynamic enhanced prediction model. DEPM is the abbreviation of the first letters of the four items of the risk model. The risk of large arteritis in patients is given by scoring. The maximum total score of the DEPM prediction model is limited to 10 points, among which low risk corresponds to 0-3 points, medium risk corresponds to 4-6 points, and high risk corresponds to 7-10 points.

[0035] In summary, after the injection of 18F-FDG tracer, the present invention can effectively obtain systemic artery-related images of the human body through PET-MRI combined imaging and CE-MRA imaging, which is convenient for subsequent analysis of large arteritis; at the same time, the selected multimodality is the artery that is prone to large arteritis, and the MRA scoring results and PET-VAS vascular activity scores are obtained respectively, which can be effectively used to assist in judging the condition and subsequently constructing a prediction model; the liver, which is metabolically active and not affected by large arteritis, is used as a comparison item when scoring, so that the comparison effect is more obvious; in addition, the ROC curve is used to take the Youden index to calculate the cutoff value for multiple predictive factors, and then univariate analysis and Cox multivariate analysis are performed, so that the best analysis data for multiple predictive factors can be determined and the stability and reliability of the model can be improved, which has significant progress.

[0036] The above embodiments are only for illustrating the technical idea of ​​the present invention, and cannot be used to limit the protection scope of the present invention. Any changes made on the basis of the technical solution in accordance with the technical idea proposed by the present invention shall fall within the protection scope of the present invention.

Claims

1. A method for constructing a prediction model for vascular remodeling in Takayasu arteritis, characterized in that: The steps include: S1. Inject tracer and perform PET-MRI combined imaging and CE-MRA imaging: First, 18F-FDG tracer is injected into multiple patients, and PET-MRI combined imaging is performed on the whole body of each patient to obtain each corresponding first image; then, gadolinium contrast agent is injected into each patient, and CE-MRA imaging is performed to obtain each corresponding second image; S2. Perform multimodal scoring to obtain the scoring results of each MRA: Based on the vascular wall thickness, the vascular wall edema and the vascular wall enhancement degree, a multimodal score is performed on each second image in step S1 to obtain each corresponding MRA score result; wherein the multiple modalities of the multimodal evaluation include: bilateral common carotid arteries and their branches, bilateral vertebral arteries, bilateral axillary arteries, bilateral subclavian arteries, brachiocephalic trunk, ascending aorta, aortic arch, descending aorta, abdominal aorta, celiac trunk, mesenteric artery, bilateral renal arteries and bilateral common iliac arteries and their branches; S3. Compare each modality with the comparison item to obtain each PET-VAS vascular activity score: Obtain the liver uptake value of 18F-FDG tracer for each patient and set it as the corresponding comparison item; For each modality in step S2, the uptake value of the vessel wall of each modality to the 18F-FDG tracer is used as a judgment basis and compared and evaluated with each corresponding comparison item to obtain a PET-VAS vascular activity score of each first image; S4. Determine each predictor and each β coefficient and construct the DEPM prediction model: The ESR sedimentation rate and delayed diagnosis time of each patient were obtained, and each ESR sedimentation rate, each delayed diagnosis time, each PET-VAS vascular activity score and each MRA score result of each patient were used as predictors. Each predictor was set with a corresponding β coefficient for weighting, and the DEPM prediction model of vascular damage in large arteritis was established based on weighting of each predictor and each β coefficient.

2. The method for constructing a vascular remodeling prediction model for Takayasu arteritis according to claim 1, characterized in that: In step S1, the injection volume of 18F-FDG tracer for any patient is 0.12 mCi / kg, and the injection volume of gadolinium contrast agent for any patient is 0.4 ml / kg.

3. The method for constructing a prediction model for vascular remodeling in Takayasu arteritis according to claim 1, characterized in that: In step S1, when CE-MRA imaging is performed, the imaging range includes the neck vessels and extends to the common iliac artery branches and intracranial blood vessels.

4. The method for constructing a prediction model for vascular remodeling in Takayasu arteritis according to claim 1, characterized in that: In step S2, when performing multimodal scoring, a vascular wall thickness of >3 mm corresponds to 1 point, and a vascular wall thickness of ≤3 mm corresponds to 0 point; the presence of vascular wall edema corresponds to 1 point, and the absence of vascular wall edema corresponds to 0 point.

5. The method for constructing a prediction model for vascular remodeling in Takayasu arteritis according to claim 1, characterized in that: When comparing and evaluating in step S3, when the vessel wall of any modality has no uptake of 18F-FDG tracer, it is scored as 0 points; when the vessel wall of any modality has uptake of 18F-FDG tracer but the uptake value is less than the corresponding comparison item, it is scored as 1 point; when the uptake value of the vessel wall of any modality for 18F-FDG tracer is equal to the corresponding comparison item, it is scored as 2 points; when the uptake value of the vessel wall of any modality for 18F-FDG tracer is greater than the corresponding comparison item, it is scored as 3 points; Each score of the multiple modalities corresponding to each first image after comparison and evaluation is added together to obtain the PET-VAS vascular activity score corresponding to each first image.

6. The method for constructing a prediction model for vascular remodeling in Takayasu arteritis according to claim 1, characterized in that: When setting each β coefficient in step S4, the weight of each prediction factor is obtained by the maximum likelihood estimation method, and the size of each weight is the corresponding β coefficient.

7. The method for constructing a prediction model for vascular remodeling in Takayasu arteritis according to claim 5, characterized in that: When performing comparative evaluation in step S3, first obtain the maximum standardized uptake value SUVmax1 of the vessel wall of any mode for the 18F-FDG tracer, then obtain the corresponding maximum standardized uptake value SUVmax2 of the liver for the 18F-FDG tracer, and score according to the size of SUVmax1 and SUVmax2.

8. The method for constructing a prediction model for vascular remodeling in Takayasu arteritis according to claim 1, characterized in that: Before setting each β coefficient in step S4, a univariate analysis is first performed on each predictive factor to screen out multiple core factors, and then a Cox multivariate analysis is performed on each core factor screened out by the univariate analysis to determine each β coefficient of each core factor.

9. The method for constructing a prediction model for vascular remodeling in Takayasu arteritis according to claim 8, characterized in that: Before univariate analysis, the ESR sedimentation rate, PET-VAS vascular activity score and MRA score of each patient were respectively calculated using the ROC curve and the Youden index to calculate the cutoff value, and univariate analysis was performed on each cutoff value.

10. The method for constructing a prediction model for vascular remodeling in Takayasu arteritis according to claim 1, characterized in that: In step S4, the predictors also include gender, age, and C-reactive protein.

Images