Patient classification method, apparatus, device, medium based on magnetic resonance parameters

Abstract

The application discloses a patient classification method and device based on magnetic resonance parameters, equipment, medium, the method comprises the following steps: obtaining the target magnetic resonance parameter of the target patient; the target magnetic resonance parameter is selected from a plurality of preset magnetic resonance parameters based on the cardiac resynchronization label of the sample patient and the true result of the occurrence of the preset endpoint event; the target magnetic resonance parameter comprises a myocardial strain parameter, a gadolinium contrast agent delayed enhancement result and a left ventricular torsion parameter; the target classification tree is input into the target patient, and the target patient is classified by using the target classification tree, and the target classification result corresponding to the target patient is obtained; the target classification tree is obtained by extending the preset classification tree based on the cut-off value of each target magnetic resonance parameter and the influence of meeting each cut-off value on cardiac resynchronization; and the target classification result is output. The application can classify patients more accurately.

Description

Technical Field

[0001] This application belongs to the field of intelligent medical care, and in particular relates to a patient classification method, device, equipment, computer-readable storage medium and computer program product based on magnetic resonance parameters. Background Technology

[0002] Cardiac resynchronization therapy (CRT) is an important device-based treatment for dilated cardiomyopathy (DCM) with intraventricular conduction block and heart failure.

[0003] Currently, the primary recommendation for CRT is based on guidelines issued by the American College of Cardiology and the European Society of Cardiology, using the duration and morphological characteristics of the QRS complex on a 12-lead electrocardiogram. However, QRS duration and morphological characteristics are too simplistic and cannot comprehensively assess a patient's suitability for CRT. Furthermore, in clinical practice, some patients recommended for CRT have experienced poor response to the treatment.

[0004] In summary, current CRT recommendations are not accurate enough. Summary of the Invention

[0005] This application provides a patient classification method, apparatus, device, computer-readable storage medium, and computer program product based on magnetic resonance parameters, which can classify patients more accurately.

[0006] On one hand, embodiments of this application provide a patient classification method based on magnetic resonance imaging parameters, the method comprising:

[0007] Obtain target magnetic resonance parameters for the target patients; the target magnetic resonance parameters are selected from multiple preset magnetic resonance parameters of the sample patients based on the cardiac resynchronization labels and the actual results of the occurrence of the preset endpoint events; the target magnetic resonance parameters include at least one of the following: myocardial strain parameters, gadolinium contrast agent delayed enhancement results, and left ventricular torsion parameters;

[0008] The target MRI parameters of the target patient are input into the target classification tree, and the target classification tree is used to classify the target patient to obtain the target classification result corresponding to the target patient. The target classification tree is obtained by extending the preset classification tree based on the cutoff values ​​of each target MRI parameter and the influence of satisfying each cutoff value on cardiac resynchronization.

[0009] Output the target classification result.

[0010] In some embodiments, before obtaining the target magnetic resonance parameters of the target patient, the method further includes:

[0011] Based on the cardiac resynchronization tags of the sample patients, control patients were identified; the cardiac resynchronization tags of the control patients were not implanted.

[0012] Based on the actual results of the control patients when the preset endpoint events occurred, the cutoff values ​​of each preset magnetic resonance imaging parameter were determined;

[0013] Based on the cutoff values, the cardiac resynchronization labels of the sample patients, and the actual results of the occurrence of the preset endpoint events, target magnetic resonance parameters are selected from multiple preset magnetic resonance parameters of the sample patients.

[0014] In some embodiments, cutoff values ​​for each preset magnetic resonance imaging parameter are determined based on the actual results of a preset endpoint event in a control patient, including:

[0015] Based on the actual results of the control patients when the pre-specified endpoint events occurred, plot the receiver operating characteristic curves between each pre-specified magnetic resonance parameter and the actual results of the control patients when the pre-specified endpoint events occurred;

[0016] The cutoff values ​​for each preset magnetic resonance parameter were determined based on the operating characteristic curves of each subject.

[0017] In some embodiments, target magnetic resonance parameters are selected from multiple preset magnetic resonance parameters of the sample patients based on cutoff values, cardiac resynchronization labels of the sample patients, and actual results of the occurrence of preset endpoint events, including:

[0018] Based on the cardiac resynchronization tag of the sample patients and whether each preset magnetic resonance parameter meets the corresponding cutoff value, the first type of patients and the second type of patients corresponding to each preset magnetic resonance parameter are determined from the sample patients; the first type of patients meet the cutoff value of the preset magnetic resonance parameter and the cardiac resynchronization tag is implanted; the second type of patients meet the cutoff value of the preset magnetic resonance parameter and the cardiac resynchronization tag is not implanted.

[0019] Determine whether there is a significant difference in the actual results of the preset endpoint events for the first type of patients and the second type of patients corresponding to each of the preset magnetic resonance parameters;

[0020] The preset magnetic resonance parameters that show a significant difference in the actual results of the preset endpoint events for the corresponding first-type patients and second-type patients are determined as the target magnetic resonance parameters.

[0021] In some embodiments, determining whether there is a significant difference in the actual results of the preset endpoint event for the first type of patients and the second type of patients corresponding to each of the preset magnetic resonance parameters includes:

[0022] Based on the actual results of the preset endpoint events for the first and second types of patients corresponding to each preset magnetic resonance parameter, the significance values ​​of the Kaplan-Meier survival curves for the first and second types of patients corresponding to each preset magnetic resonance parameter are determined.

[0023] Patients in categories I and II, whose Kaplan-Meier survival curve significance values ​​were less than a pre-defined threshold, were identified as having a significant difference in the true outcome of the pre-defined endpoint event.

[0024] In some embodiments, before inputting the target patient's target magnetic resonance parameters into the target classification tree, the method further includes:

[0025] Obtain a preset classification tree; the preset classification tree is used to determine the initial classification result corresponding to the target patient based on the target patient's clinical parameters; the clinical parameters include at least one of the following: left ventricular ejection fraction (LVEF), QRS duration, QRS morphology characteristics, and cardiac function classification; the initial classification result includes a first initial classification and a second initial classification.

[0026] The initial classification result is determined as the initial intermediate node of the target classification tree; the branches of the initial intermediate node include the first initial classification and the second initial classification.

[0027] The cutoff value of the gadolinium contrast agent delayed enhancement result is determined as the first intermediate node of the first initial classification branch, and whether the cutoff value of the gadolinium contrast agent delayed enhancement result is satisfied is determined as the branch of the first intermediate node;

[0028] The cutoff values ​​of the ventricular strain parameter and the left ventricular torsion parameter are determined as the second intermediate node connecting the second initial classification branch, and whether the first preset condition is met is determined as the branch of the second intermediate node; the first preset condition is to meet the cutoff value of the myocardial strain parameter or the cutoff value of the left ventricular torsion parameter.

[0029] Based on the impact of satisfying each cutoff value on cardiac resynchronization, the branches of the first intermediate node and the branches of the second intermediate node are connected to the first classification result or the second classification result to obtain the target classification tree.

[0030] In some embodiments, connecting the branches of the first intermediate node and the branches of the second intermediate node to the first classification result or the second classification result based on the impact of satisfying each cutoff value on cardiac resynchronization includes:

[0031] The branch that has a positive impact on cardiac resynchronization is connected to one of the first and second classification results;

[0032] Branches that do not have a positive impact on cardiac resynchronization are connected to another branch in either the first or second classification result.

[0033] In some embodiments, the cardiac function classification includes a first to a fourth grade with progressively increasing levels, and the grade is positively correlated with the degree of patient activity limitation; the QRS complex morphological characteristics are left bundle branch block (LBBB) or non-LBBB; the first initial classification satisfies LVEF less than or equal to a first threshold, QRS complex duration greater than or equal to a second threshold, LBBB, and cardiac function classification of the second to fourth grades; the second initial classification satisfies LVEF less than or equal to a first threshold, QRS complex duration greater than or equal to a second threshold, non-LBBB, and cardiac function classification of the second to fourth grades, or satisfies LVEF less than or equal to a first threshold, QRS complex duration less than a second threshold and greater than a third threshold, LBBB, and cardiac function classification of the second to fourth grades;

[0034] The target patient is classified using a target classification tree to obtain the target classification results corresponding to the target patient, including:

[0035] If the initial classification result of the target patient is the first initial classification, then determine whether the gadolinium contrast agent delayed enhancement result meets the cutoff value; the cutoff value for meeting the gadolinium contrast agent delayed enhancement result includes: the gadolinium contrast agent delayed enhancement result is negative, the gadolinium contrast agent delayed enhancement result is positive and the enhancement percentage is ∈ [0, 8%).

[0036] If the gadolinium contrast agent delayed enhancement result is determined to meet the cutoff value, then the target classification result is determined to be one of the first classification result and the second classification result;

[0037] If it is determined that the gadolinium contrast agent delayed enhancement result does not meet the cutoff value, then the target classification result is determined to be the other of the first classification result and the second classification result;

[0038] If the initial classification result of the target patient is the second initial classification, then determine whether the first preset condition is met;

[0039] If the first preset condition is met, the target classification result is determined to be one of the first classification result and the second classification result;

[0040] If it is determined that the first preset condition is not met, then the target classification result is determined to be the other of the first classification result and the second classification result.

[0041] In some embodiments, the output target classification result includes at least one of the following:

[0042] The target classification results are displayed visually.

[0043] Generate and output a cardiac resynchronization therapy indication recommendation report based on the target classification results;

[0044] A pre-set artificial intelligence prediction model is used to generate a preoperative assessment report for cardiac resynchronization surgery, and the report is then output.

[0045] On the other hand, embodiments of this application provide a patient classification device based on magnetic resonance parameters. The device includes: an acquisition module for acquiring target magnetic resonance parameters of a target patient; the target magnetic resonance parameters are selected from multiple preset magnetic resonance parameters of the sample patient based on the cardiac resynchronization label of the sample patient and the actual results of the occurrence of a preset endpoint event; the target magnetic resonance parameters include at least one of the following: myocardial strain parameters, gadolinium contrast agent delayed enhancement percentage, and left ventricular torsion parameters;

[0046] The classification module is used to input the target magnetic resonance parameters of the target patient into the target classification tree, and to classify the target patient using the target classification tree to obtain the target classification result corresponding to the target patient. The target classification tree is obtained by extending the preset classification tree based on the cutoff values ​​of each target magnetic resonance parameter and the impact of satisfying each cutoff value on cardiac resynchronization.

[0047] The output module is used to output the target classification results.

[0048] In another aspect, embodiments of this application provide a patient classification device based on magnetic resonance parameters, the patient classification device based on magnetic resonance parameters includes: a processor and a memory storing computer program instructions;

[0049] When the processor executes computer program instructions, it implements a patient classification method based on magnetic resonance parameters, as described above.

[0050] In another aspect, embodiments of this application provide a computer-readable storage medium storing computer program instructions, which, when executed by a processor, implement the patient classification method based on magnetic resonance parameters as described above.

[0051] In another aspect, embodiments of this application provide a computer program product in which instructions, when executed by a processor of an electronic device, cause the electronic device to perform a patient classification method based on magnetic resonance parameters as described above.

[0052] The patient classification method, apparatus, device, and computer-readable storage medium based on magnetic resonance parameters of this application obtain target magnetic resonance parameters of the target patient. Since the target magnetic resonance parameters are selected from multiple preset magnetic resonance parameters of the sample patient based on the cardiac resynchronization label and the actual results of the occurrence of a preset endpoint event, the target magnetic resonance parameters include at least one of the following: myocardial strain parameters, gadolinium contrast agent delayed enhancement results, and left ventricular torsion parameters. Therefore, the target magnetic resonance parameters can comprehensively assess whether the patient's cardiac physiological state is suitable for cardiac resynchronization. The target magnetic resonance parameters of the target patient are input into a target classification tree, and the target patient is classified using the target classification tree to obtain the target classification result corresponding to the target patient. Since the target classification tree is obtained by extending the preset classification tree based on the cutoff values ​​of each target magnetic resonance parameter and the impact of satisfying each cutoff value on cardiac resynchronization, the target classification tree can accurately classify patients according to the effect of cardiac resynchronization under different target magnetic resonance parameters. Thus, the target classification result is output, achieving accurate classification of the target patient. Attached Figure Description

[0053] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0054] Figure 1 This is a schematic flowchart of a patient classification method based on magnetic resonance parameters provided in one embodiment of this application;

[0055] Figure 2 This is a flowchart illustrating a patient classification method based on magnetic resonance parameters provided in another embodiment of this application;

[0056] Figure 3 and Figure 4 This is a Kaplan-Meier survival curve for a COR class I patient in yet another embodiment of this application;

[0057] Figure 5 This is a Kaplan-Meier survival curve for COR class IIa and COR class IIb patients in yet another embodiment of this application;

[0058] Figure 6 This is a Kaplan-Meier survival curve for a COR IIa patient in yet another embodiment of this application;

[0059] Figure 7 This is a schematic diagram illustrating the probability of different types of patients experiencing a preset endpoint event in yet another embodiment of this application;

[0060] Figure 8 This is a logical diagram illustrating the classification of target patients by a target classification tree in another embodiment of this application;

[0061] Figure 9 A schematic diagram of the structure of a patient classification device based on magnetic resonance parameters provided in another embodiment of this application;

[0062] Figure 10 This is a schematic diagram of the hardware structure of a patient classification device based on magnetic resonance parameters provided in another embodiment of this application. Detailed Implementation

[0063] The features and exemplary embodiments of various aspects of this application will be described in detail below. To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain this application and not to limit it. For those skilled in the art, this application can be implemented without some of these specific details. The following description of the embodiments is merely to provide a better understanding of this application by illustrating examples.

[0064] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes said element.

[0065] All data involved in this application was obtained through legal channels, and explicit informed consent has been obtained from the relevant users / patients, and anonymization has been performed.

[0066] First, let me introduce the technical terms used in this application.

[0067] Cardiac resynchronization is an intervention strategy that uses electrophysiological modulation techniques to correct asynchronous ventricular contractions.

[0068] The QRS complex, or QRS wave complex, is a waveform in an electrocardiogram that represents ventricular depolarization, i.e., systolic electrical activity. It consists of the Q wave, R wave, and S wave.

[0069] QRS duration refers to the length of time the QRS complex lasts on an electrocardiogram, usually measured in milliseconds, and is used to assess ventricular conduction abnormalities.

[0070] Left bundle branch block (LBBB) is a cardiac electrophysiological abnormality in which the left bundle branch of the heart cannot effectively conduct electrical signals, resulting in asynchronous electrical activity between the left and right ventricles.

[0071] Left ventricular ejection fraction (LVEF) is the percentage of blood ejected from the left ventricle during each systole relative to the end-diastolic volume, and is used to measure cardiac pumping function.

[0072] Right ventricular ejection fraction (RVEF) is the percentage of blood ejected from the right ventricle during each systole relative to the end-diastolic volume, and is used to measure cardiac pumping function.

[0073] Myocardial strain parameters, also known as ventricular strain parameters, are indices of myocardial deformation measured using imaging techniques, reflecting ventricular systolic function.

[0074] Global Longitudinal Strain (GLS) is a myocardial strain parameter that represents the degree of overall deformation of the myocardium in the longitudinal direction, i.e., along the long axis of the heart.

[0075] Global Radial Strain (GRS) is a myocardial strain parameter that represents the degree of overall deformation of the myocardium in the radial direction, which is perpendicular to the long axis of the heart.

[0076] Global Circumferential Strain (GCS) is a myocardial strain parameter that represents the degree of overall deformation of the myocardium in the circumferential direction, i.e., around the ventricular cavity.

[0077] Late Gadolinium Enhancement (LGE) is a technique used in cardiac magnetic resonance imaging (MRI) involving delayed scanning after the application of a gadolinium contrast agent. It is used to detect areas of myocardial fibrosis or scarring.

[0078] Gadolinium contrast agent delayed enhancement percentage is the percentage of the LGE region relative to the total mass of the left ventricular myocardium, used to quantify the degree of myocardial damage.

[0079] Cardiac torsion is a ventricular mechanical parameter that represents the difference in the relative rotation angle between the apex and base of the heart during systole, reflecting the mechanical synergy of the helical structure of myocardial fibers.

[0080] The left ventricular torsion parameter is a ventricular mechanical parameter used to assess the mechanical properties of the helical structure of the left ventricular myocardial fibers. In this application, it is described in detail as the torsion parameter.

[0081] N-terminal pro-B-type natriuretic peptide (NT-proBNP) is a key biomarker for assessing cardiac function. It is secreted by ventricular myocytes in response to increased cardiac pressure or myocardial injury. NT-proBNP levels reflect cardiac function status and are commonly expressed as lgNT-proBNP.

[0082] The NYHA Cardiac Function Classification is a standard developed by the New York Heart Association to classify the cardiac function of heart failure patients into classes I to IV based on the ease with which daily activities can trigger symptoms, ranging from asymptomatic to symptomatic at rest.

[0083] The receiver operating characteristic curve, or ROC curve for short, is a curve that measures the operating characteristics of a receiver.

[0084] The Kaplan-Meier survival curve, or KM curve for short, is also known as the Kaplan-Meier survival curve.

[0085] The univariate Cox proportional hazards regression model is a regression model used to analyze the risk impact of a single predictor variable on the timing of an event.

[0086] The multivariate Cox proportional hazards regression model is a regression model used to simultaneously analyze the risk impact of multiple predictor variables on the timing of an event, while controlling for confounding factors.

[0087] Cardiac magnetic resonance (CMR) imaging is a non-invasive medical examination that uses magnetic fields, radio waves, and computer technology to generate detailed images of the heart and its blood vessels.

[0088] The prior art involved in this application will now be described in detail.

[0089] Cardiac resynchronization therapy (CRT) is an important device-based treatment for patients with dilated cardiomyopathy (DCM) and heart failure who also have intraventricular conduction block, such as left bundle branch block. Effective CRT can improve ventricular synchrony, enhance cardiac pumping function, and significantly improve patient symptoms, quality of life, and prognosis.

[0090] Currently, clinical recommendations regarding patient suitability for CRT treatment primarily rely on clinical guidelines published by authoritative organizations such as the American College of Cardiology (ACC), the American Heart Association (AHA), or the European Society of Cardiology (ESC). The core recommendations of these guidelines heavily depend on the analysis of standard 12-lead electrocardiograms, particularly the QRS duration and morphological characteristics, typically using a QRS duration ≥120ms or ≥150ms as key thresholds. Based on these relatively singular characteristics, the guidelines categorize patients into those recommended for CRT implantation, those who may benefit, or those not recommended.

[0091] The inventors' in-depth research revealed that existing guideline-based CRT recommendation methods have significant accuracy issues. This is because the QRS duration and morphological characteristics are too simplistic and cannot comprehensively reflect a patient's cardiac physiological state. Furthermore, in clinical practice, some patients recommended for CRT implantation according to current guideline standards have not achieved the expected benefits.

[0092] In summary, existing guideline-based CRT recommendation methods cannot accurately classify patients, and the classification results provide limited technical assistance to doctors.

[0093] To address the aforementioned problems in the prior art, embodiments of this application provide a patient classification method, apparatus, device, computer-readable storage medium, and computer program product based on magnetic resonance imaging (MRI) parameters. The patient classification method based on MRI parameters provided in this application will be described first below.

[0094] Figure 1 This is a schematic flowchart of a patient classification method based on magnetic resonance parameters provided in one embodiment of this application. Figure 1 As shown, an embodiment of this application provides a patient classification method based on magnetic resonance parameters, which is executed by a patient classification device based on magnetic resonance parameters, including steps 101 to 103.

[0095] Step 101: Obtain the target magnetic resonance parameters of the target patient; the target magnetic resonance parameters are selected from multiple preset magnetic resonance parameters of the sample patient based on the cardiac resynchronization label and the actual results of the occurrence of the preset endpoint event; the target magnetic resonance parameters include at least one of the following: myocardial strain parameters, gadolinium contrast agent delayed enhancement results, and left ventricular torsion parameters.

[0096] In this embodiment, the cardiac resynchronization tag is used to indicate whether the patient has undergone cardiac resynchronization. For example, the cardiac resynchronization tag can be either implanted or not implanted. Implanted indicates that the patient has had a cardiac pacing device implanted for cardiac resynchronization, while not implanted indicates that the patient has not had a cardiac pacing device implanted for cardiac resynchronization.

[0097] Obtaining target magnetic resonance parameters can be achieved by receiving target magnetic resonance parameters input by the user, by analyzing and calculating the target patient's magnetic resonance images, or by retrieving target magnetic resonance parameters from a preset database based on the target patient's identifier.

[0098] Preset MRI parameters include: left ventricular ejection fraction (LVEF), right ventricular ejection fraction (RVEF), myocardial strain parameters, gadolinium-contrast delayed enhancement (LGE) results, and left ventricular torsion (Torsion) parameters. LGE results are categorized as LGE negative or LGE positive; negative LGE indicates the absence of LGE, while positive LGE indicates its presence. When LGE is positive, the percentage of LGE is provided. Myocardial strain parameters include: global longitudinal strain (GLS), global circumferential strain (GCS), and global radial strain (GRS).

[0099] Preset magnetic resonance parameters can be obtained by analyzing and calculating cardiac magnetic resonance images of sample patients. Cardiac magnetic resonance images are obtained by performing magnetic resonance scanning of the heart using a magnetic resonance scanner. Cardiac magnetic resonance images can include the following three types: conventional localization images, conventional cine sequences, and delayed enhancement sequences.

[0100] By using specialized cardiac function analysis software to analyze and calculate the short-axis cine of the left ventricle, parameters such as left ventricular ejection fraction, right ventricular ejection fraction, and left ventricular mass can be obtained.

[0101] The full width at half height (WHM) method was used to analyze delayed enhancement in the short-axis section of the left ventricle. Abnormal delayed enhancement was defined as a threshold of more than half the maximum signal intensity within the myocardial scar in the region of interest. The area or volume of the region of interest and its relative percentage to the mass of the left ventricle could be calculated.

[0102] The endocardial and epicardial boundaries are identified in short-axis, two-chamber, three-chamber, and four-chamber cine sequences at end-diastole, enabling boundary tracking throughout the entire cardiac cycle and calculation of myocardial strain parameters. Long-axis cine sequences are used to calculate global longitudinal strain, while short-axis cine sequences are used to calculate global radial strain, global circumferential strain, and ventricular torsional mechanical parameters.

[0103] The ventricular torsion value is calculated using a standardized formula: torsion = (peak apical rotation angle - peak basal rotation angle) / ventricular long axis length. The quantification of ventricular torsion is based on the characteristic differences in horizontal rotational motion between the apex and base: when viewed from the apex, the apex exhibits counterclockwise rotation, while the base exhibits clockwise rotation.

[0104] The magnetic resonance scanner has a main magnetic field strength of 1.5T and features vector ECG gating, multi-channel coil combination, and MRI-compatible high-pressure injector. During scanning, it can adopt various scanning methods such as head-first supine scanning, wireless vector ECG single-gated scanning, and respiratory-guided ECG gating scanning.

[0105] The sample number of patients is multiple. For sample patients whose cardiac resynchronization is labeled as implanted, the preset magnetic resonance parameters refer to the parameters of the sample patients before the implantation of the cardiac pacing device used for cardiac resynchronization.

[0106] Preset endpoint events include: death from heart failure, sudden cardiac death (SCD), heart transplantation, and implantation of a left ventricular assist device.

[0107] The true outcome of a sample patient experiencing a pre-defined endpoint event refers to whether the sample patient experienced the pre-defined endpoint event within a pre-defined time period. In other words, the true outcome of a sample patient experiencing a pre-defined endpoint event can be either that the event occurred or that it did not occur.

[0108] The target magnetic resonance parameters can be parameters that are significantly correlated with the occurrence of the preset endpoint event, parameters that have a positive impact on cardiac resynchronization, parameters that are significantly correlated with the occurrence of the preset endpoint event or have a positive impact on cardiac resynchronization, or parameters that are significantly correlated with the occurrence of the preset endpoint event and have a positive impact on cardiac resynchronization.

[0109] In one embodiment of this application, the target magnetic resonance parameter is selected from the preset magnetic resonance parameters that have a positive impact on cardiac resynchronization.

[0110] In one embodiment of this application, a univariate Cox proportional hazards regression model and a multivariate Cox proportional hazards regression model can be used to determine whether each preset magnetic resonance parameter is significantly correlated with the occurrence of a preset endpoint event. For ease of description, the univariate Cox proportional hazards regression model will be referred to as the univariate regression model, and the multivariate Cox proportional hazards regression model will be referred to as the multivariate regression model.

[0111] For example, if the significance value p of a univariate regression model is less than a preset significance threshold, then the preset magnetic resonance parameters corresponding to that univariate regression model are significantly correlated with the occurrence of a preset endpoint event. The preset significance threshold can be 0.05. In other words, the preset magnetic resonance parameters corresponding to a univariate regression model with p < 0.05 are significantly correlated with the occurrence of a preset endpoint event.

[0112] In one embodiment of this application, Kaplan-Meier survival curves can be used to determine the effect of preset magnetic resonance parameters on cardiac resynchronization, as described in step 2033, which will not be repeated here.

[0113] In one embodiment of this application, the specific method for determining whether there is a significant correlation between each preset magnetic resonance parameter and the occurrence of a preset endpoint event can be as follows.

[0114] First, the cutoff values ​​for each predicted MRI parameter were determined. Then, a univariate regression model was used to perform regression analysis on the correlation between the cutoff values ​​of each preset MRI parameter and the occurrence of the preset endpoint event, as well as the correlation between multiple clinical parameters and the occurrence of the preset endpoint event. This is hereinafter referred to as univariate regression analysis, and the results are shown in Table 1 below. The method for determining the cutoff values ​​of each preset MRI parameter can be found in the description in step 202, and will not be repeated here.

[0115] The preset MRI parameters were: left ventricular ejection fraction, right ventricular ejection fraction, LGE tertile, GLS ≥ -5.9%, GCS ≥ -6.3%, GRS ≤ 8.0%, and Torsion ≤ 0.34° / cm. Multiple clinical parameters included: gender, age, QRS waveform, systolic blood pressure, NYHA functional class, NT-proBNP level, and creatinine level. The QRS waveform refers to the morphological characteristics of the QRS complex, including left bundle branch block, right bundle branch block, and nonspecific conduction delay.

[0116] Table 1 Results of univariate regression analysis

[0117]

[0118] According to the univariate regression analysis results shown in Table 1, there were significant correlations between the occurrence of the pre-specified endpoint event and QRS waveform, systolic blood pressure, NYHA functional classification, NT-proBNP level, creatinine level, biventricular ejection fraction, presence of LGE, LGE tertiles, GLS≥-5.9%, GCS≥-6.3%, GRS≤8.0%, and torsion≤0.34° / cm.

[0119] In one embodiment of this application, in order to more accurately determine the parameters that are significantly correlated with the occurrence of the preset endpoint event, the parameters with p < 0.05 in the univariate regression analysis results, namely QRS waveform, systolic blood pressure, NYHA functional classification, NT-proBNP level, creatinine level, RVEF, LGE tertiles, LVEF, GLS, GCS, GRS and torsion, are included in a multivariate Cox regression model, and the forward stepwise method is used for parameter screening, hereinafter referred to as multivariate regression analysis.

[0120] Before performing multivariate regression analysis, Pearson correlation analysis was used to calculate the correlation between the parameters. The correlation coefficients between LVEF, GLS, GCS and GRS were found to be greater than 0.6, which indicates that there is a significant correlation between LVEF, GLS, GCS and GRS.

[0121] Therefore, multivariate regression analysis requires the establishment of multiple independent multivariate regression models. Specifically, LVEF, GLS, GCS, and GRS are used to establish multivariate regression models with other parameters from the univariate regression analysis results that have p < 0.05, and the significance values ​​of each multivariate regression model are determined.

[0122] Multivariate regression analysis was performed, and the Harrell's C-statistic was used to evaluate the discriminative power of each multivariate regression model. The Harrell's C-statistic ranged from 0.5 to 1.0. A higher value indicates a stronger discriminative ability of the model. A Harrell's C-statistic of 0.5 indicates that the model has no discriminative ability, and a Harrell's C-statistic of 1.0 indicates that the model has complete discriminative ability.

[0123] A multivariate Cox proportional hazards regression model was constructed, named Model 1, relating LVEF to QRS waveform, systolic blood pressure, NYHA functional classification, NT-proBNP level, creatinine level, RVEF, LGE tertile variables, and torsion. The parameters were selected using the forward stepwise method.

[0124] A multivariate Cox proportional hazards regression model was constructed between GLS and QRS waveform, systolic blood pressure, NYHA functional classification, NT-proBNP level, creatinine level, RVEF, LGE tertile variables and torsion, named Model 2, and the parameters were screened using the forward stepwise method.

[0125] A multivariate Cox proportional hazards regression model was constructed, named Model 3, relating GCS to QRS waveform, systolic blood pressure, NYHA functional classification, NT-proBNP level, creatinine level, RVEF, LGE tertile variables, and torsion. The parameters were selected using the forward stepwise method.

[0126] A multivariate Cox proportional hazards regression model was constructed, named Model 4, relating GRS and QRS waveforms, systolic blood pressure, NYHA functional classification, NT-proBNP level, creatinine level, RVEF, LGE tertile variables, and torsion. The parameters were selected using the forward stepwise method.

[0127] The results of the multivariate regression analysis for Model 1 and Model 2 are shown in Table 2 below.

[0128] Table 2. Results of multivariate regression analysis for Model 1 and Model 2

[0129]

[0130] The results of the multivariate regression analysis for Model 3 and Model 4 are shown in Table 3 below.

[0131] Table 3. Results of multivariate regression analysis for Model 3 and Model 4

[0132]

[0133] The results of the multivariate regression analysis showed that creatinine, RVEF, and Torsion ≤0.34° / cm were not significantly associated with the occurrence of the pre-specified endpoint event. The parameters that were significantly associated with the occurrence of the pre-specified endpoint event were QRS waveform, systolic blood pressure, NYHA functional class, lgNT-proBNP, LGE tertile, LVEF ≤24.6%, GLS ≥-5.9%, GCS ≥-6.3%, and GRS ≤8.0%.

[0134] Step 102: Input the target magnetic resonance parameters of the target patient into the target classification tree, and use the target classification tree to classify the target patient to obtain the target classification result corresponding to the target patient; the target classification tree is obtained by extending the preset classification tree based on the cutoff values ​​of each target magnetic resonance parameter and the influence of satisfying each cutoff value on cardiac resynchronization.

[0135] In one embodiment of this application, a pre-constructed classification tree is used for initial classification of target patients. The pre-constructed classification tree can be constructed based on patient classifications in clinical guidelines published by the ACC, AHA, or ESC. The pre-constructed classification tree can also be constructed recursively by segmenting clinical parameters of sample patients, such as QRS duration, QRS morphology, systolic blood pressure, cardiac function classification, lgNT-proBNP level, and creatinine level.

[0136] The target classification tree is obtained by extending the preset classification tree. For example, based on the cutoff values ​​of the target magnetic resonance parameters, the next-level node is constructed on the initial classification results of the sample patients in the preset classification tree, and branches are extended on the next-level node and connected to the leaf nodes according to the impact of the cutoff values ​​of each target magnetic resonance parameter on cardiac resynchronization.

[0137] In one embodiment of this application, before or simultaneously with inputting the target magnetic resonance parameters into the target classification tree, the method further includes inputting clinical parameters of the target patient, such as QRS duration, QRS morphology, systolic blood pressure, cardiac function classification, lgNT-proBNP, and creatinine, into the target classification tree so that the target classification tree can determine the initial classification result of the target patient.

[0138] Step 103: Output the target classification result.

[0139] In this embodiment, the methods for outputting the target classification results may include: visual display, voice broadcast, printing, sending to the target device, etc.

[0140] The patient classification method based on magnetic resonance parameters provided in this embodiment obtains the target magnetic resonance parameters of the target patient. Since the target magnetic resonance parameters are selected from multiple preset magnetic resonance parameters of the sample patient based on the cardiac resynchronization label and the actual results of the occurrence of a preset endpoint event, the target magnetic resonance parameters include at least one of the following: myocardial strain parameters, gadolinium contrast agent delayed enhancement results, and left ventricular torsion parameters. Therefore, the target magnetic resonance parameters can comprehensively assess whether the patient's cardiac physiological state is suitable for cardiac resynchronization. The target magnetic resonance parameters of the target patient are input into a target classification tree, and the target classification tree is used to classify the target patient to obtain the target classification result corresponding to the target patient. Since the target classification tree is obtained by extending the preset classification tree based on the cutoff values ​​of each target magnetic resonance parameter and the impact of satisfying each cutoff value on cardiac resynchronization, the target classification tree can accurately classify patients according to the effect of cardiac resynchronization under different target magnetic resonance parameters. Thus, the target classification result is output, achieving accurate classification of the target patient.

[0141] Figure 2 This is a flowchart illustrating a patient classification method based on magnetic resonance parameters provided in another embodiment of this application. Figure 2 As shown, in order to accurately classify the target patients, steps 201 to 203 are included before step 101.

[0142] Step 201: Identify control patients based on the cardiac resynchronization tags of the sample patients; the cardiac resynchronization tags of the control patients are not implanted.

[0143] In one embodiment of this application, in order to determine the cutoff values ​​for each preset magnetic resonance imaging (MRI) parameter, sample patients tagged as not having cardiac resynchronization implanted are designated as control patients. Since the control patients do not have a cardiac pacing device implanted for cardiac resynchronization, the actual results of the preset endpoint event in the control patients can be used to determine the correlation between the preset MRI parameters and the occurrence of the preset endpoint event, thereby determining the cutoff values ​​for the preset MRI parameters.

[0144] Step 202: Determine the cutoff values ​​for each preset magnetic resonance imaging parameter based on the actual results of the control patients experiencing the preset endpoint event.

[0145] In one embodiment of this application, the actual results of the control patient experiencing a preset endpoint event can be obtained from a preset database based on the control patient's identifier. The actual results of the control patient experiencing the preset endpoint event in the preset database can be obtained and written into the database by relevant personnel through methods such as calling the patient or their family, reviewing medical records, or consulting the attending physician.

[0146] In one embodiment of this application, in order to accurately determine the cutoff value of the preset magnetic resonance parameters, step 202 includes steps 2021 to 2022.

[0147] Step 2021: Based on the actual results of the control patients experiencing the preset endpoint events, plot the receiver operating characteristic curves between each preset magnetic resonance imaging parameter and the actual results of the control patients experiencing the preset endpoint events.

[0148] In one embodiment of this application, the steps for plotting the ROC curve for a certain preset magnetic resonance parameter can be as follows.

[0149] The preset magnetic resonance parameters are used as thresholds in descending order of their values.

[0150] Control patients whose preset MRI parameters were greater than a threshold were considered to have experienced the preset endpoint event, while those whose parameters were less than the threshold were considered not to have experienced the preset endpoint event. Control patients whose preset MRI parameters were equal to the threshold were considered either not to have experienced the preset endpoint event or to have experienced the endpoint event.

[0151] Calculate the true positive rate (TPR) and false positive rate (FDR) for each threshold. TPR, also known as sensitivity, represents the proportion of control patients whose true outcome of the pre-specified endpoint was correctly predicted when the event occurred. FDR represents the proportion of control patients whose true outcome of the pre-specified endpoint was incorrectly predicted when the event did not occur.

[0152] A coordinate system was constructed with FPR as the x-axis and TPR as the y-axis. Points (0,0), (1,1), and the points corresponding to the calculated FPR and TPR for each threshold were plotted on the coordinate system and connected by a balancing line segment. This yielded the ROC curve between the preset MRI parameters and the actual results of the control patients experiencing the preset endpoint event. Point (0,0) represents the case where all control patients were judged not to have experienced the preset endpoint event, and point (1,1) represents the case where all control patients were judged to have experienced the preset endpoint event.

[0153] Step 2022: Determine the cutoff values ​​of each preset magnetic resonance parameter based on the working characteristic curve of each subject.

[0154] In one embodiment of this application, the accuracy of the ROC curve is evaluated by the area under the curve (AUC), and the cutoff value is determined by the Youden index.

[0155] Specifically, the Youden index is calculated for each threshold, and the threshold with the highest Youden index is determined as the cutoff value. Youden index = Sensitivity + Specificity - 1. Specificity = 1 - FPR.

[0156] The patient classification method based on magnetic resonance parameters provided in this embodiment can determine the true positive rate (TDR) and false positive rate (FDR) at different thresholds, and the Youden index at different thresholds, by plotting the receiver operating characteristic (ROC) curves between each preset magnetic resonance parameter and the actual results of the preset endpoint events in control patients, based on the actual results of the preset endpoint events in control patients. This allows for accurate determination of the cutoff values ​​of each preset magnetic resonance parameter, thereby achieving accurate classification of target patients.

[0157] In one embodiment of this application, by plotting receiver operating characteristic (ROC) curves between the actual outcomes of each preset magnetic resonance imaging (MRI) parameter and the occurrence of preset endpoint events in control patients, it was found that the AUC for 1-LVEF was 0.68, with a 95% confidence interval (CI) of 0.64–0.73 (p < 0.001); the AUC for the presence of LGE was 0.63, with a 95% confidence interval (CI) of 0.59–0.68 (p < 0.001); and the AUC for the percentage of LGE was 0.69, with a 95% confidence interval (CI) of 0.64–0.73. The AUC for GLS was 0.68, with a 95% confidence interval CI of 0.63–0.73, p < 0.001; the AUC for GCS was 0.68, with a 95% confidence interval CI of 0.63–0.72, p < 0.001; the AUC for 1-GRS was 0.69, with a 95% confidence interval CI of 0.65–0.74, p < 0.001; and the AUC for 1-torsion was 0.53, with a 95% confidence interval CI of 0.48–0.58, p = 0.27. Therefore, the cutoff values ​​for LVEF, GLS, GCS, GRS, and Torsion were determined to be 24.6%, -5.9%, -6.3%, 8.0%, and 0.34° / cm, respectively. The LGE results were ternary variables: LGE negative, LGE ∈ [0–8.0%), and LGE ≥ 8.0%.

[0158] Step 203: Based on each cutoff value, the cardiac resynchronization label of the sample patients, and the actual results of the occurrence of the preset endpoint event, select the target magnetic resonance parameters from multiple preset magnetic resonance parameters of the sample patients.

[0159] In some embodiments, step 203 includes steps 2031 to 2033.

[0160] Step 2031: Based on the cardiac resynchronization tag of the sample patients and whether each preset magnetic resonance parameter meets the corresponding cutoff value, determine the first type of patients and the second type of patients corresponding to each preset magnetic resonance parameter from the sample patients; the first type of patients meets the cutoff value of the preset magnetic resonance parameter and the cardiac resynchronization tag is implanted; the second type of patients meets the cutoff value of the preset magnetic resonance parameter and the cardiac resynchronization tag is not implanted.

[0161] In this embodiment, the cutoff value for the preset magnetic resonance parameters refers to the fact that, in the ROC curve of the cutoff value, the preset magnetic resonance parameters are within the range where the preset endpoint event is judged not to have occurred. For example, for LGE enhancement results, satisfying the cutoff value means LGE is negative or LGE∈[0-8.0%), and not satisfying the cutoff value means LGE≥8.0%; for LVEF, satisfying the cutoff value means LVEF≤24.6%, and not satisfying the cutoff value means LVEF>24.6%.

[0162] In one embodiment of this application, patients whose cardiac resynchronization therapy (CRT) is labeled as implanted can be identified as experimental patients, and patients whose CRT is labeled as not implanted can be identified as control patients. Furthermore, a first category of patients corresponding to each preset magnetic resonance imaging (MRI) parameter can be determined from the experimental patients, and a second category of patients corresponding to each preset MRI parameter can be determined from the control patients.

[0163] In one embodiment of this application, the cardiac resynchronization tag is a sample patient who has been implanted and meets the following parameters: LVEF≤35%; left ventricular end-diastolic diameter or volume greater than 2 standard deviations of the normal reference value after correction for body surface area, sex and / or age; QRS duration≥120ms; QRS morphological characteristics of complete left bundle branch block, complete right bundle branch block or nonspecific intraventricular conduction delay.

[0164] The end-diastolic diameter or volume of the left ventricle can be calculated using the two-plane method. The maximum left atrial volume index = maximum left atrial volume / body surface area. Maximum left atrial volume = 0.85 × (left atrial area in two-chamber plane × left atrial area in four-chamber plane) / shorter length of the left atrium (two-chamber or four-chamber plane: from the midpoint of the mitral annulus plane to the upper surface of the left atrium).

[0165] Step 2032: Determine whether there is a significant difference in the actual results of the preset endpoint events for the first type of patients and the second type of patients corresponding to each of the preset magnetic resonance parameters.

[0166] Step 2033: The preset magnetic resonance parameters for which the actual results of the preset endpoint events for the corresponding first-type patients and second-type patients are significantly different are determined as the target magnetic resonance parameters.

[0167] In this embodiment, when there is a significant difference in the actual outcomes of the preset endpoint event for the first and second types of patients corresponding to the preset magnetic resonance parameters, it indicates that cardiac resynchronization is significantly correlated with the occurrence of the preset endpoint event when the preset magnetic resonance parameter cutoff value is met, and the cutoff value of the preset magnetic resonance parameters has a positive impact on cardiac resynchronization. Conversely, when there is no significant difference in the actual outcomes of the preset endpoint event for the first and second types of patients corresponding to the preset magnetic resonance parameters, it indicates that cardiac resynchronization is not significantly correlated with the occurrence of the preset endpoint event when the preset magnetic resonance parameter cutoff value is met, and the cutoff value of the preset magnetic resonance parameters has no positive impact on cardiac resynchronization. Therefore, in order to accurately classify patients based on target magnetic resonance parameters, the preset magnetic resonance parameters for which there is a significant difference in the actual outcomes of the preset endpoint event for the corresponding first and second types of patients are determined as target magnetic resonance parameters.

[0168] The patient classification method based on magnetic resonance parameters provided in this embodiment determines the first and second categories of patients corresponding to each preset magnetic resonance parameter by using the cardiac resynchronization label of sample patients and whether each preset magnetic resonance parameter meets the corresponding cutoff value. This method identifies patients who meet each cutoff value and then classifies them into first and second categories based on the cardiac resynchronization label. It also determines whether there is a significant difference in the actual results of the preset endpoint event for the first and second categories of patients corresponding to each preset magnetic resonance parameter. Furthermore, it determines the relationship between cardiac resynchronization and the occurrence of the preset endpoint event when the cutoff value corresponding to the preset magnetic resonance parameter is met, accurately identifying the target magnetic resonance parameter that affects cardiac resynchronization, thus enabling accurate classification of the target patients.

[0169] In one embodiment of this application, to verify the accuracy of the target magnetic resonance parameters, third and fourth categories of patients corresponding to each preset magnetic resonance parameter can be determined from the sample patients based on the cardiac resynchronization tag of the sample patients and whether each preset magnetic resonance parameter meets the corresponding cutoff value. Third category patients do not meet the cutoff value of the preset magnetic resonance parameters, but the cardiac resynchronization tag is implanted; fourth category patients do not meet the cutoff value of the preset magnetic resonance parameters, but the cardiac resynchronization tag is not implanted. Based on the actual results of the preset endpoint events for each third category patient and each fourth category patient, the survival rate corresponding to each third category patient and each fourth category patient is calculated respectively. Based on the survival rates corresponding to each third category patient and each fourth category patient, the difference in control survival rate for each preset magnetic resonance parameter is calculated. The control survival rate difference can be the difference in survival rates corresponding to each third category patient and each fourth category patient. If the actual survival rate difference corresponding to the preset magnetic resonance parameter is greater than the preset difference, and the corresponding control survival rate difference is less than the preset difference, it can be determined that meeting the cutoff value of the target magnetic resonance parameter has an impact on cardiac resynchronization, and the target magnetic resonance parameter is accurate.

[0170] In one embodiment of this application, step 2032 includes steps 501 to 502.

[0171] Step 501: Based on the actual results of the preset endpoint events for the first and second types of patients corresponding to each preset magnetic resonance parameter, determine the significance value of the Kaplan-Meier survival curves for the first and second types of patients corresponding to each preset magnetic resonance parameter.

[0172] In some embodiments, the actual results of the first and second types of patients experiencing a preset endpoint event can be input into preset analysis software, and the preset analysis software can output the significance values ​​of the Kaplan-Meier survival curves for the first and second types of patients.

[0173] In some embodiments, the significance value of the Kaplan-Meier survival curve can be determined by plotting Kaplan-Meier survival curves for patients in the first and second categories.

[0174] Step 502: The first and second categories of patients whose Kaplan-Meier survival curves have significance values ​​less than a preset threshold are identified as having a significant difference in the true outcome of the preset endpoint event.

[0175] In some embodiments, the preset threshold can be 0.05. When the significance value p < 0.05 of the Kaplan-Meier survival curves for patients in category I and category II is less than 0.05, it indicates that there is a significant difference between the survival curves for patients in category I and category II. This significant difference can be statistically significant.

[0176] If there is a significant difference between the survival curves of patients in category I and category II, it indicates that cardiac resynchronization is significantly correlated with the occurrence of the preset endpoint event when the preset MRI parameter cutoff value is met, and the corresponding preset MRI parameter is determined as the target MRI parameter. Conversely, if there is no significant difference between the survival curves of patients in category I and category II, such as if they almost overlap, it indicates that cardiac resynchronization is not significantly correlated with the occurrence of the preset endpoint event when the preset MRI parameter cutoff value is met, and the corresponding preset MRI parameter is not determined as the target MRI parameter.

[0177] In some embodiments, the actual outcomes of a patient experiencing a pre-defined endpoint event include: the follow-up duration, and the outcome of the pre-defined endpoint event occurring within the follow-up duration. The outcome of the pre-defined endpoint event occurring within the follow-up duration includes: no occurrence of the pre-defined endpoint event within the follow-up duration, and occurrence of the pre-defined endpoint event at a certain time point within the follow-up duration. The actual outcomes of different types of patients experiencing pre-defined endpoint events are presented below using Kaplan-Meier survival curves.

[0178] Figure 3 and Figure 4 This is a Kaplan-Meier survival curve for a COR I patient in yet another embodiment of this application.

[0179] exist Figure 3 and Figure 4 In the corresponding embodiment, COR Class I patients are those with a QRS duration ≥150ms and QRS morphological characteristics of left bundle branch block. Among COR Class I patients, those with implanted cardiac resynchronization technology (CRT) were classified as the COR Class I CRT group, and those with non-implanted CRT were classified as the COR Class I non-CRT group. Figure 3 and Figure 4 In the Kaplan-Meier survival curves shown, the number of patients followed up represents the number of patients in that group whose follow-up period has not yet ended as the follow-up time increases. Each point decrease in the curve indicates that one patient in that group experienced the pre-specified endpoint event at that corresponding follow-up time point. Each cross on the curve represents that one patient in that group had their corresponding follow-up time end at that time point, and that patient did not experience the pre-specified endpoint event during the corresponding follow-up period. It can be understood that the magnitude of the decrease in the Kaplan-Meier survival curve indicates the number of patients who experienced the pre-specified endpoint event at that corresponding follow-up time point.

[0180] like Figure 3 and Figure 4The Kaplan-Meier survival curves shown indicate that, compared to the non-CRT group in the COR I category, patients in the COR I CRT group had a significantly lower risk of the pre-specified endpoint event and a significantly higher survival rate (p=0.04). When LGE, GLS, GCS, GRS, and torsion met the cutoff values ​​(i.e., LGE negative, LGE between 0-8.0%, GLS ≥ -5.9%, GCS ≥ -6.3%, GRS ≤ 8.0%, and torsion ≤ 0.34° / cm), compared to the non-CRT group in the COR I category, patients in the COR I CRT group had a significantly lower risk of the pre-specified endpoint event and a significantly higher survival rate (p=0.02, p=0.04, p<0.001, p<0.001, p<0.001, p<0.001, p<0.001, p<0.001, respectively). When LVEF met the cutoff value (LVEF ≤ 24.6%), the risk of pre-specified endpoint events was not significantly reduced or survival was not significantly improved in the COR Class I CRT group compared to the non-CRT group (p = 0.009). When LVEF, LGE, GLS, GCS, GRS, and torsion did not meet the cutoff values, the event-free survival of patients in the COR Class I CRT group was similar to that in the COR Class I non-CRT group (p = 0.934, p = 0.60, p = 0.79, p = 0.76, p = 0.94, and p = 0.70, respectively). This indirectly validated that meeting the cutoff values ​​for GLE, GLS, GCS, GRS, and torsion has a positive impact on cardiac resynchronization, while meeting the cutoff value for LVEF does not have a positive impact on cardiac resynchronization, demonstrating the accuracy of the target MRI parameters for patient classification.

[0181] Figure 5 This is a Kaplan-Meier survival curve for COR IIa and COR IIb patients in another embodiment of this application.

[0182] exist Figure 5In the corresponding embodiments, COR Class IIa patients are those with a QRS duration between 120-150 ms and QRS morphological characteristics of left bundle branch block, or those with a QRS duration ≥150 ms and QRS morphological characteristics of non-left bundle branch block. COR Class IIb patients are those with a QRS duration between 120-150 ms and QRS morphological characteristics of non-left bundle branch block. Among COR Class IIa patients, those with implanted cardiac resynchronization technology (CRT) are classified as the COR Class IIa CRT group, and those with non-implanted CRT are classified as the COR Class IIa non-CRT group. Similarly, among COR Class IIb patients, those with implanted CRT are classified as the COR Class IIb CRT group, and those with non-implanted CRT are classified as the COR Class IIb non-CRT group. Figure 5 In the Kaplan-Meier survival curves shown, each point the curve drops represents the occurrence of the pre-defined endpoint event in one person in the group, and the number of survivors represents the number of people remaining in the group as the follow-up time increases.

[0183] like Figure 5 The Kaplan-Meier survival curves shown indicate that, compared to the COR IIa non-CRT group, the event-free survival rate of the COR IIa CRT group was similar (p=0.07); compared to the COR IIb non-CRT group, the event-free survival rate of the COR IIb CRT group was similar (p=0.93). When the level of left ventricular dysplasia (LGE) was high (LGE≥8.0%), or when left ventricular strain and torsion were mild, CRT implantation did not significantly improve clinical outcomes. The event-free survival rates of the COR IIa CRT group and the COR IIa non-CRT group were similar, as were the event-free survival rates of the COR IIb CRT group and the COR IIb non-CRT group.

[0184] Figure 6 This is a Kaplan-Meier survival curve for a COR IIa patient in yet another embodiment of this application.

[0185] exist Figure 6 In the corresponding embodiment, the same as Figure 5 In the corresponding embodiments, the same indicators are used to identify COR IIa patients, and COR IIa patients are divided into a COR IIa non-CRT group and a COR IIa CRT group. For example... Figure 6As shown, when GLS≥-5.9%, GCS≥-6.3%, GRS≤8.0%, or torsion≤0.34° / cm, the event-free survival rate of the COR IIa CRT group was significantly improved compared with the non-CRT group, with significance values ​​p=0.005, p=0.01, p=0.01, and p=0.02, respectively.

[0186] Figure 7 This is a schematic diagram illustrating the probability of different types of patients experiencing a preset endpoint event in another embodiment of this application.

[0187] exist Figure 7 In the corresponding embodiment, the same as Figure 3 and Figure 4 In the corresponding embodiments, the same indicators were used to identify COR Class I patients, and COR Class I patients were divided into a COR Class I non-CRT group and a COR Class I CRT group, using the same... Figure 5 In the corresponding embodiments, the same indicators are used to identify COR IIa patients, and COR IIa patients are divided into a COR IIa non-CRT group and a COR IIa CRT group. For example... Figure 7 As shown, there were 241 patients in COR class I and 255 patients in COR class IIa. The probability of the pre-specified endpoint event occurring in COR class I patients with negative LGE was 2.2%.

[0188] When LGE was negative, the COR Class I CRT group had a lower probability of the pre-specified endpoint event and a higher event-free survival rate compared to the COR Class I non-CRT group. The COR Class IIa CRT group also had a lower probability of the pre-specified endpoint event and a higher event-free survival rate compared to the COR Class IIa non-CRT group. When LGE ≥ 8.0%, GLS < -5.9%, GCS < -6.3%, GRS > 8.0%, and torsion > 0.34° / cm, the COR Class I CRT group had similar probabilities of the pre-specified endpoint event and similar event-free survival rates compared to the COR Class I non-CRT group, and the COR Class IIa CRT group had similar probabilities of the pre-specified endpoint event compared to the COR Class IIa non-CRT group. Event-free survival rate is the ratio of the number of patients who did not experience the pre-specified endpoint event to the total number of patients in this category.

[0189] In summary, when the cutoff values ​​for LGE, GLS, GCS, GRS, and torsion are met, cardiac resynchronization significantly reduces the probability of the pre-defined endpoint event; however, when the cutoff value for LVEF is met, cardiac resynchronization does not significantly reduce the probability of the pre-defined endpoint event. In other words, meeting the cutoff values ​​for LGE, GLS, GCS, GRS, and torsion has a positive impact on cardiac resynchronization, while meeting the cutoff value for LVEF does not have a positive impact. Figure 7 The probability of different types of patients experiencing a preset endpoint event, as shown in another embodiment of this application, further verifies the accuracy of the target magnetic resonance parameters determined in this application for patient classification.

[0190] In some embodiments, in order to accurately classify the target patient, steps 501 to 505 are included before step 103, “inputting the target magnetic resonance parameters of the target patient into the target classification tree”.

[0191] Step 501: Obtain a preset classification tree; the preset classification tree is used to determine the initial classification result corresponding to the target patient based on the clinical parameters of the target patient; the clinical parameters include at least one of the following: left ventricular ejection fraction (LVEF), QRS duration, QRS morphological characteristics, and cardiac function classification; the initial classification result includes a first initial classification and a second initial classification.

[0192] In this embodiment, the preset classification tree can divide the target patient into a first initial category and a second initial category based on the QRS duration and QRS morphological characteristics of the target patient.

[0193] In some embodiments, the preset classification tree assigns patients meeting the Class I recommendation criteria from the American College of Cardiology guidelines to the first initial classification, and patients meeting the Class IIa and Class IIb recommendation criteria to the second initial classification.

[0194] For example, the preset classification tree classifies target patients with QRS duration ≥150ms and left bundle branch block into a first initial category; classifies target patients with QRS duration between 120-150ms and left bundle branch block, and target patients with QRS duration ≥150ms but not left bundle branch block into a second initial category; and classifies target patients with QRS duration between 120-150ms and not left bundle branch block into a second initial category.

[0195] Step 502: Determine the initial classification result as the initial intermediate node of the target classification tree; the branches of the initial intermediate node include the first initial classification and the second initial classification.

[0196] In this embodiment, the classification result of the preset classification tree for the target patient includes two cases: the target patient is in the first initial classification and the target patient is in the second initial classification. Therefore, the branches of the initial intermediate node include the first initial classification and the second initial classification.

[0197] Step 503: Determine the cutoff value of the gadolinium contrast agent delayed enhancement result as the first intermediate node connecting the first initial classification branch, and determine whether the cutoff value of the gadolinium contrast agent delayed enhancement result is satisfied as the branch of the first intermediate node.

[0198] In this embodiment, for patients in the first initial classification, since cardiac resynchronization can reduce the probability of a preset endpoint event when the gadolinium contrast agent delayed enhancement result is met, the cutoff value of the gadolinium contrast agent delayed enhancement result is determined as the first intermediate node connecting the first initial classification branch, and whether the cutoff value of the gadolinium contrast agent delayed enhancement result is met is determined as the branch of the first intermediate node.

[0199] The first intermediate node has three branches: LGE negative, LGE∈[0-8.0%), and LGE≥8.0%.

[0200] Step 504: Determine the cutoff values ​​of the ventricular strain parameters and the left ventricular torsion parameters as the second intermediate nodes connecting the second initial classification branches, and determine whether the first preset condition is met as the branch of the second intermediate node; the first preset condition is to meet the cutoff values ​​of the myocardial strain parameters or the left ventricular torsion parameters.

[0201] In this embodiment, for patients in the second initial classification, since cardiac resynchronization can reduce the probability of a preset endpoint event when the cutoff values ​​of myocardial strain parameters and left ventricular torsion parameters are met, the cutoff values ​​of myocardial strain parameters and left ventricular torsion parameters are determined as the second intermediate node connecting the second initial classification branches, and whether the first preset condition is met is determined as the branch of the second intermediate node.

[0202] The second intermediate node has two branches: one that satisfies the first preset condition and one that does not. Satisfying the first preset condition has four possible outcomes: GLS ≥ -5.9%; GCS ≥ -6.3%; GRS ≤ 8.0%; and torsion ≤ 0.34° / cm. Not satisfying the first preset condition has one of the following possible outcomes: GLS > -5.9%, GCS < -6.3%, GRS > 8.0%, and torsion > 0.34° / cm.

[0203] Step 505: Based on the impact of satisfying each cutoff value on cardiac resynchronization, connect the branches of the first intermediate node and the branches of the second intermediate node to the first classification result or the second classification result to obtain the target classification tree.

[0204] In some embodiments, step 505, “connecting the branches of the first intermediate node and the branches of the second intermediate node to the first classification result or the second classification result based on the effect of satisfying each cutoff value on cardiac resynchronization,” includes steps 5051 to 5052.

[0205] Step 5051: Connect the branch that has a positive impact on cardiac resynchronization to one of the first and second classification results.

[0206] Step 5052: Connect the branch that does not have a positive impact on cardiac resynchronization to another branch in the first or second classification result.

[0207] In this embodiment, since satisfying the cutoff value for delayed enhancement of gadolinium contrast agent has a positive impact on cardiac resynchronization, LGE negative and LGE∈[0-8.0%) have a positive impact on cardiac resynchronization.

[0208] Since meeting the cutoff values ​​for myocardial strain parameters or left ventricular torsion parameters has a positive impact on cardiac resynchronization, meeting the first preset condition has a positive impact on cardiac resynchronization.

[0209] Therefore, branches with LGE negative, LGE ∈ [0-8.0%), and those satisfying the first preset condition are connected to the same classification result, while branches with LGE ≥ 8.0% and those not satisfying the first preset condition are connected to the same classification result. For example, branches with LGE negative, LGE ∈ [0-8.0%), and those satisfying the first preset condition are connected to the first classification result, while branches with LGE ≥ 8.0% and those not satisfying the first preset condition are connected to the second classification result.

[0210] The patient classification method based on magnetic resonance parameters provided in this embodiment extends a target classification tree on a preset classification tree according to the influence of the cutoff values ​​of each preset magnetic resonance parameter on cardiac resynchronization. The determined target classification tree can accurately classify the target patients.

[0211] Figure 8 This is a logical diagram illustrating how a target classification tree classifies target patients in another embodiment of this application. For example... Figure 8As shown, in another embodiment of this application, the cardiac function classification includes a first to a fourth grade with progressively increasing levels, and the grade is positively correlated with the degree of patient activity limitation; the morphological characteristics of the QRS complex are left bundle branch block (LBBB) or non-LBBB; the first initial classification satisfies LVEF less than or equal to a first threshold, QRS complex duration greater than or equal to a second threshold, LBBB, and cardiac function classification of the second to fourth grades; the second initial classification satisfies LVEF less than or equal to a first threshold, QRS complex duration greater than or equal to a second threshold, non-LBBB, and cardiac function classification of the second to fourth grades, or satisfies LVEF less than or equal to a first threshold, QRS complex duration less than a second threshold and greater than a third threshold, LBBB, and cardiac function classification of the second to fourth grades. Step 102, "classifying the target patient using a target classification tree to obtain the target classification result corresponding to the target patient," includes steps 1021 to 1026.

[0212] In this embodiment, the first threshold can be 35%, the second threshold can be 150ms, and the third threshold can be 120ms. Before inputting the target magnetic resonance parameters into the target classification tree, the initial classification result of the target patient is input into the target classification tree. The target classification tree first determines whether the initial classification result of the target patient is a first initial classification or a second initial classification.

[0213] Step 1021: If the initial classification result of the target patient is the first initial classification, determine whether the gadolinium contrast agent delayed enhancement result meets the cutoff value; the cutoff value for meeting the gadolinium contrast agent delayed enhancement result includes: the gadolinium contrast agent delayed enhancement result is negative, the gadolinium contrast agent delayed enhancement result is positive and the enhancement percentage is ∈ [0, 8%].

[0214] Step 1022: If the gadolinium contrast agent delayed enhancement result is determined to meet the cutoff value, then the target classification result is determined to be one of the first classification result and the second classification result.

[0215] Step 1023: If it is determined that the gadolinium contrast agent delayed enhancement result does not meet the cutoff value, then the target classification result is determined to be the other of the first classification result and the second classification result.

[0216] Step 1024: If the initial classification result of the target patient is the second initial classification, then determine whether the first preset condition is met.

[0217] Step 1025: If the first preset condition is met, the target classification result is determined to be one of the first classification result and the second classification result.

[0218] Step 1026: If it is determined that the first preset condition is not met, then the target classification result is determined to be the other of the first classification result and the second classification result.

[0219] The patient classification method based on magnetic resonance parameters provided in this embodiment can accurately classify patients.

[0220] In yet another embodiment of this application, step 103, “outputting target classification results”, includes at least one of steps 1031 to 1033.

[0221] Step 1031: Visualize the target classification results.

[0222] Electronic devices may include display devices, and the target classification results may be output in a way that visualizes the target classification results on the display device. Visualizing the target classification results can provide the results to doctors to assist them in developing treatment plans for the target patients.

[0223] Step 1032: Generate a cardiac resynchronization therapy indication recommendation report based on the target classification results, and output the cardiac resynchronization therapy indication recommendation report.

[0224] Electronic devices can be integrated into magnetic resonance imaging (MRI) scanners and can directly obtain target MRI parameters from the MRI scanner scan results. The electronic devices can pre-store templates for cardiac resynchronization therapy (CARS) indication recommendation reports. After classifying target patients according to target MRI parameters, CARS indication recommendation reports can be output based on the target classification results and the pre-stored templates for CARS indication recommendation reports.

[0225] Step 1033: Use a preset artificial intelligence prediction model to generate a preoperative assessment report for cardiac resynchronization surgery and output the preoperative assessment report for cardiac resynchronization surgery.

[0226] The electronic device can be configured with a preset artificial intelligence prediction model. After the target classification result corresponding to the target patient is determined by using the target classification tree, the target classification result is input into the preset artificial intelligence prediction model. The preset artificial intelligence prediction model is used to generate a preoperative assessment report for cardiac resynchronization surgery for the target patient and outputs the preoperative assessment report for cardiac resynchronization surgery.

[0227] The patient classification method based on magnetic resonance parameters provided in this embodiment outputs the target classification results of the target patient in multiple ways, which can be used in the cardiology or radiology departments of hospitals to assist doctors in formulating treatment plans for the target patient.

[0228] Based on the patient classification method based on magnetic resonance parameters provided in the above embodiments, this application also provides specific implementation methods of a patient classification device based on magnetic resonance parameters. Please refer to the following embodiments.

[0229] Figure 9 A schematic diagram of a patient classification device based on magnetic resonance parameters provided in another embodiment of this application is shown below. Figure 9 In yet another embodiment of this application, the patient classification device 60 based on magnetic resonance parameters includes the following units:

[0230] The first acquisition module 61 is used to acquire the target magnetic resonance parameters of the target patient; the target magnetic resonance parameters are selected from multiple preset magnetic resonance parameters of the sample patient based on the cardiac resynchronization label and the actual results of the occurrence of the preset endpoint event; the target magnetic resonance parameters include at least one of the following: myocardial strain parameters, gadolinium contrast agent delayed enhancement results, and left ventricular torsion parameters;

[0231] The classification module 62 is used to input the target magnetic resonance parameters of the target patient into the target classification tree, and to classify the target patient using the target classification tree to obtain the target classification result corresponding to the target patient; the target classification tree is obtained by extending the preset classification tree based on the cutoff values ​​of each target magnetic resonance parameter and the influence of satisfying each cutoff value on cardiac resynchronization.

[0232] Output module 63 is used to output the target classification results.

[0233] The patient classification device based on magnetic resonance parameters provided in this embodiment acquires the target magnetic resonance parameters of the target patient. Since the target magnetic resonance parameters are selected from multiple preset magnetic resonance parameters of the sample patient based on the cardiac resynchronization label and the actual results of the occurrence of a preset endpoint event, the target magnetic resonance parameters include at least one of the following: myocardial strain parameters, gadolinium contrast agent delayed enhancement results, and left ventricular torsion parameters. Therefore, the target magnetic resonance parameters can comprehensively assess whether the patient's cardiac physiological state is suitable for cardiac resynchronization. By inputting the target magnetic resonance parameters of the target patient into a target classification tree and classifying the target patient using the target classification tree, the target classification result corresponding to the target patient is obtained. Since the target classification tree is obtained by extending the preset classification tree based on the cutoff values ​​of each target magnetic resonance parameter and the impact of satisfying each cutoff value on cardiac resynchronization, the target classification tree can accurately classify patients according to the effect of cardiac resynchronization under different target magnetic resonance parameters, and then output the target classification result, thus achieving accurate classification of the target patient.

[0234] As one implementation of this application, in order to accurately classify target patients, the above-mentioned device may further include:

[0235] The first determination module is used to determine the control patients based on the cardiac resynchronization tags of the sample patients; the cardiac resynchronization tags of the control patients are not implanted.

[0236] The second determining module is used to determine the cutoff values ​​of each preset magnetic resonance parameter based on the actual results of the preset endpoint event in the control patients.

[0237] The screening module is used to screen target magnetic resonance parameters from multiple preset magnetic resonance parameters of sample patients based on various cutoff values, cardiac resynchronization labels of sample patients, and actual results of the occurrence of preset endpoint events.

[0238] As another implementation of this application, in order to accurately determine the cutoff values ​​of each preset magnetic resonance parameter, and thus achieve accurate classification of the target patient, the aforementioned second determining module can be specifically used for:

[0239] Based on the actual results of the control patients when the preset endpoint events occurred, receiver operating characteristic (ROC) curves were plotted between each preset MRI parameter and the actual results of the control patients when the preset endpoint events occurred; cutoff values ​​for each preset MRI parameter were determined based on each ROC curve.

[0240] As another implementation of this application, in order to accurately determine the target magnetic resonance parameters that have a positive impact on cardiac resynchronization of patients, so as to accurately classify the target patients, the above-mentioned screening module can be specifically used to: determine the first type of patients and the second type of patients corresponding to each preset magnetic resonance parameter from the sample patients based on the cardiac resynchronization label of the sample patients and whether each preset magnetic resonance parameter meets the corresponding cutoff value; the first type of patients meet the cutoff value of the preset magnetic resonance parameter and the cardiac resynchronization label is implanted; the second type of patients meet the cutoff value of the preset magnetic resonance parameter and the cardiac resynchronization label is not implanted; determine whether there is a significant difference in the actual results of the preset endpoint event for the first type of patients and the second type of patients corresponding to each preset magnetic resonance parameter; and determine the preset magnetic resonance parameters for which there is a significant difference in the actual results of the preset endpoint event for the first type of patients and the second type of patients as the target magnetic resonance parameters.

[0241] As another implementation of this application, in order to quickly determine whether there is a significant difference in the actual results of the preset endpoint events for the first type of patients and the second type of patients corresponding to each preset magnetic resonance parameter, the above screening module may be further used to: determine the significance value of the Kaplan-Meier survival curves of the first type of patients and the second type of patients corresponding to each preset magnetic resonance parameter based on the actual results of the preset endpoint events for the first type of patients and the second type of patients corresponding to each preset magnetic resonance parameter; and determine the first type of patients and the second type of patients whose Kaplan-Meier survival curve significance values ​​are less than a preset threshold as having a significant difference in the actual results of the preset endpoint events.

[0242] As another implementation of this application, in order to determine a target classification tree capable of accurately classifying the target patient, the above-mentioned apparatus may further include:

[0243] The second acquisition module is used to acquire a preset classification tree; the preset classification tree is used to determine the initial classification result corresponding to the target patient based on the clinical parameters of the target patient; the clinical parameters include at least one of the following: left ventricular ejection fraction (LVEF), QRS duration, QRS morphological characteristics, and cardiac function classification; the initial classification result includes a first initial classification and a second initial classification.

[0244] The third determining module is used to determine the initial classification result as the initial intermediate node of the target classification tree; the branches of the initial intermediate node include the first initial classification and the second initial classification.

[0245] The fourth determining module is used to determine the cutoff value of the gadolinium contrast agent delayed enhancement result as the first intermediate node connected by the first initial classification branch, and to determine whether the cutoff value of the gadolinium contrast agent delayed enhancement result is satisfied as the branch of the first intermediate node;

[0246] The fifth determining module is used to determine the cutoff values ​​of the ventricular strain parameters and the left ventricular torsion parameters as the second intermediate nodes connecting the second initial classification branches, and to determine whether the first preset condition is met as the branch of the second intermediate node; the first preset condition is to meet the cutoff value of the myocardial strain parameters or the cutoff value of the left ventricular torsion parameters.

[0247] The connection module is used to connect the branches of the first intermediate node and the branches of the second intermediate node to the first classification result or the second classification result based on the impact of satisfying each cutoff value on cardiac resynchronization, so as to obtain the target classification tree.

[0248] As another implementation of this application, in order to determine a target classification tree that can accurately classify the target patient, the above-mentioned connection module can be specifically used to: connect a branch that has a positive impact on cardiac resynchronization to one of the first classification result and the second classification result; and connect a branch that does not have a positive impact on cardiac resynchronization to the other of the first classification result or the second classification result.

[0249] As another implementation of this application, in order to accurately classify the target patient, the cardiac function classification includes grades one to four, which are sequentially increased in level, and the grade is positively correlated with the degree of activity limitation of the patient; the morphological characteristics of the QRS complex are left bundle branch block (LBBB) or not LBBB; the first initial classification satisfies LVEF less than or equal to the first threshold, QRS complex duration greater than or equal to the second threshold, LBBB, and cardiac function classification of grade two to four; the second initial classification satisfies LVEF less than or equal to the first threshold, QRS complex duration greater than or equal to the second threshold, not LBBB, and cardiac function classification of grade two to four, or satisfies LVEF less than or equal to the first threshold, QRS complex duration less than the second threshold and greater than the third threshold, LBBB, and cardiac function classification of grade two to four, and the above classification module is specifically used for: if the initial classification result of the target patient If the initial classification is the first category, then it is determined whether the gadolinium contrast agent delayed enhancement result meets the cutoff value; the cutoff value for meeting the gadolinium contrast agent delayed enhancement result includes: the gadolinium contrast agent delayed enhancement result is negative, the gadolinium contrast agent delayed enhancement result is positive and the enhancement percentage is ∈ [0, 8%); if the gadolinium contrast agent delayed enhancement result meets the cutoff value, then the target classification result is determined to be one of the first classification result and the second classification result; if the gadolinium contrast agent delayed enhancement result does not meet the cutoff value, then the target classification result is determined to be the other of the first classification result and the second classification result; if the initial classification result of the target patient is the second initial category, then it is determined whether the first preset condition is met; if the first preset condition is met, then the target classification result is determined to be one of the first classification result and the second classification result; if the first preset condition is not met, then the target classification result is determined to be the other of the first classification result and the second classification result.

[0250] As another implementation of this application, the output module is specifically used for at least one of the following: visually displaying the target classification results; generating a cardiac resynchronization therapy indication recommendation report based on the target classification results, and outputting the cardiac resynchronization therapy indication recommendation report; generating a preoperative cardiac resynchronization therapy assessment report using a preset artificial intelligence prediction model, and outputting the preoperative cardiac resynchronization therapy assessment report.

[0251] Figure 7 This is a schematic diagram of the hardware structure of a patient triage device based on magnetic resonance parameters according to another embodiment of this application. See also... Figure 7 In yet another embodiment of this application, the patient classification device based on magnetic resonance parameters may include a processor 71 and a memory 72 storing computer program instructions.

[0252] In yet another embodiment of this application, a patient classification device based on magnetic resonance parameters may include a processor 71 and a memory 72 storing computer program instructions.

[0253] Specifically, the processor 71 may include a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of this application.

[0254] Memory 72 may include mass storage for data or instructions. For example, and not limitingly, memory 72 may include a hard disk drive (HDD), floppy disk drive, flash memory, optical disk, magneto-optical disk, magnetic tape, or Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, memory 72 may include removable or non-removable (or fixed) media. Where appropriate, memory 72 may be internal or external to the integrated gateway disaster recovery device. In a particular embodiment, memory 72 is non-volatile solid-state memory.

[0255] In a particular embodiment, memory 72 includes read-only memory (ROM). Where appropriate, the ROM may be a mask-programmed ROM, a programmable ROM (PROM), an erasable PROM (EPROM), an electrically erasable PROM (EEPROM), an electrically rewritable ROM (EAROM), or flash memory, or a combination of two or more of these.

[0256] Memory may include read-only memory (ROM), random access memory (RAM), disk storage media devices, optical storage media devices, flash memory devices, and electrical, optical, or other physical / tangible memory storage devices. Therefore, typically, memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the methods according to one aspect of this disclosure.

[0257] The processor 71 reads and executes computer program instructions stored in the memory 72 to implement any of the patient classification methods based on magnetic resonance parameters in the above embodiments.

[0258] In one example, the patient triage device based on magnetic resonance imaging parameters may also include a communication interface 73 and a bus 74. Wherein, as Figure 7 As shown, the processor 71, memory 72, and communication interface 73 are connected through bus 74 and complete communication with each other.

[0259] Communication interface 73 is mainly used to realize communication between various modules, devices, units and / or equipment in the embodiments of this application.

[0260] Bus 74 includes hardware, software, or both, that couples components of an online data traffic metering device together. For example, and not limitingly, the bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an Infinite Bandwidth Interconnect, a Low Pin Count (LPC) bus, a memory bus, a Microchannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, or other suitable buses, or combinations of two or more of these. Where appropriate, bus 74 may include one or more buses. Although specific buses are described and illustrated in embodiments of this application, any suitable bus or interconnect is contemplated herein.

[0261] Furthermore, in conjunction with the magnetic resonance imaging-based patient classification methods in the above embodiments, this application embodiment can provide a computer-readable storage medium for implementation. This computer-readable storage medium stores computer program instructions; when executed by a processor, these computer program instructions implement any of the magnetic resonance imaging-based patient classification methods in the above embodiments.

[0262] This application also provides a computer program product, including a computer program that, when executed, implements any of the magnetic resonance-based patient classification methods described in the above embodiments.

[0263] It should be clarified that this application is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of this application is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of this application.

[0264] The functional modules shown in the above block diagram can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this application are programs or code segments used to perform the required tasks. Programs or code segments can be stored on a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried on a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, radio frequency (RF) links, etc. Code segments can be downloaded via computer networks such as the Internet, intranets, etc.

[0265] It should also be noted that the exemplary embodiments mentioned in this application describe methods or systems based on a series of steps or apparatus. However, this application is not limited to the order of the above steps; that is, the steps can be performed in the order mentioned in the embodiments, or in a different order, or several steps can be performed simultaneously.

[0266] The aspects of this disclosure have been described above with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this disclosure. It should be understood that each block in the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that these instructions, executable via the processor of the computer or other programmable data processing apparatus, enable the implementation of the functions / actions specified in one or more blocks of the flowchart illustrations and / or block diagrams. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field-programmable logic circuit. It is also understood that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can also be implemented by special-purpose hardware performing the specified functions or actions, or can be implemented by a combination of special-purpose hardware and computer instructions.

[0267] The above are merely specific embodiments of this application. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, modules, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. It should be understood that the protection scope of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the protection scope of this application.

Claims

1. A patient classification method based on magnetic resonance imaging parameters, characterized in that, include: Obtain the target magnetic resonance parameters of the target patient; The target magnetic resonance parameters are selected from multiple preset magnetic resonance parameters of the sample patients based on the cardiac resynchronization labels and the actual results of the occurrence of preset endpoint events; the target magnetic resonance parameters include at least one of the following: myocardial strain parameters, gadolinium contrast agent delayed enhancement results, and left ventricular torsion parameters; The target magnetic resonance parameters of the target patient are input into the target classification tree, and the target classification tree is used to classify the target patient to obtain the target classification result corresponding to the target patient. The target classification tree is obtained by extending the preset classification tree based on the cutoff values ​​of each of the target magnetic resonance parameters and the influence of satisfying each of the cutoff values ​​on cardiac resynchronization. Output the target classification result; Before obtaining the target magnetic resonance parameters of the target patient, the process also includes: Based on the cardiac resynchronization tags of the sample patients, control patients were identified; the cardiac resynchronization tags of the control patients were not implanted. Based on the actual results of the control patients experiencing the preset endpoint event, the cutoff values ​​for each preset magnetic resonance imaging parameter were determined; Based on the cutoff values, the cardiac resynchronization labels of the sample patients, and the actual results of the occurrence of the preset endpoint events, target magnetic resonance parameters are selected from multiple preset magnetic resonance parameters of the sample patients. The determination of cutoff values ​​for each preset magnetic resonance imaging parameter based on the actual results of the preset endpoint event in the control patients includes: Based on the actual results of the control patients when the preset endpoint events occurred, plot the receiver operating characteristic curves between each of the preset magnetic resonance parameters and the actual results of the control patients when the preset endpoint events occurred; The cutoff values ​​for each preset magnetic resonance imaging parameter were determined based on the subject operating characteristic curves. For gadolinium contrast agent delayed enhancement (LGE), the cutoff value was 8.0%; for left ventricular torsion (torsion), the cutoff value was 0.34° / cm; and for myocardial strain parameters GLS, GCS, and GRS, the cutoff values ​​were -5.9%, -6.3%, and 8.0%, respectively. The process involves selecting target magnetic resonance parameters from multiple preset magnetic resonance parameters of the sample patients based on the cutoff values, the cardiac resynchronization labels of the sample patients, and the actual results of the occurrence of preset endpoint events. This includes: Based on the cardiac resynchronization tag of the sample patients and whether each preset magnetic resonance parameter meets the corresponding cutoff value, the first type of patients and the second type of patients corresponding to each preset magnetic resonance parameter are determined from the sample patients; the first type of patients meet the cutoff value of the preset magnetic resonance parameter and the cardiac resynchronization tag is implanted; the second type of patients meet the cutoff value of the preset magnetic resonance parameter and the cardiac resynchronization tag is not implanted. Determine whether there is a significant difference in the actual results of the preset endpoint events for the first type of patients and the second type of patients corresponding to each of the preset magnetic resonance parameters; The preset magnetic resonance parameters that show a significant difference in the actual results of the preset endpoint event for the corresponding first-type patients and second-type patients are determined as the target magnetic resonance parameters; The determination of whether there is a significant difference in the actual results of the first type of patients and the second type of patients corresponding to each of the preset magnetic resonance parameters for the occurrence of the preset endpoint event includes: Based on the actual results of the preset endpoint events for the first and second types of patients corresponding to each preset magnetic resonance parameter, the significance values ​​of the Kaplan-Meier survival curves for the first and second types of patients corresponding to each preset magnetic resonance parameter are determined. Patients in category 1 and category 2 whose Kaplan-Meier survival curve significance values ​​are less than a preset threshold are identified as having a significant difference in the true outcome of the preset endpoint event. In the Kaplan-Meier survival curve, the horizontal axis represents the number of patients followed up, and the vertical axis represents the event-free survival rate. The number of patients followed up represents the number of patients in the first and second patient groups whose follow-up period has not yet ended as the follow-up time increases. Each point decrease in the Kaplan-Meier survival curve indicates that one person in that group has experienced the pre-specified endpoint event at that time point of the corresponding follow-up period. Each cross in the Kaplan-Meier survival curve indicates that one person in that group has completed the corresponding follow-up period at that time point, and that person has not experienced the pre-specified endpoint event during the corresponding follow-up period. The step of classifying the target patient using the target classification tree to obtain the target classification result corresponding to the target patient includes: Input the initial classification results of the target patient into the target classification tree; If the initial classification result of the target patient is the first initial classification, then determine whether the gadolinium contrast agent delayed enhancement result meets the cutoff value; the cutoff value for meeting the gadolinium contrast agent delayed enhancement result includes: the gadolinium contrast agent delayed enhancement result is negative, the gadolinium contrast agent delayed enhancement result is positive and the enhancement percentage is ∈ [0, 8%). If the gadolinium contrast agent delayed enhancement result is determined to meet the cutoff value, then the target classification result is determined to be one of the first classification result and the second classification result; If it is determined that the gadolinium contrast agent delayed enhancement result does not meet the cutoff value, then the target classification result is determined to be the other of the first classification result and the second classification result; If the initial classification result of the target patient is the second initial classification, then it is determined whether the first preset condition is met; the first preset condition is to meet the cutoff value of the myocardial strain parameter or the cutoff value of the left ventricular torsion parameter. If it is determined that the first preset condition is met, then the target classification result is determined to be one of the first classification result and the second classification result; If it is determined that the first preset condition is not met, then the target classification result is determined to be the other of the first classification result and the second classification result; For the left ventricular torsion parameter, the cutoff value is defined as torsion ≤ 0.34° / cm, and the cutoff value is defined as torsion > 0.34° / cm. For the myocardial strain parameter, the cutoff value is defined as GLS ≥ -5.9%, GCS ≥ -6.3%, and GRS ≤ 8.0%, and the cutoff value is defined as GLS < -5.9%, GCS < -6.3%, or GRS > 8.0%.

2. The patient classification method based on magnetic resonance parameters according to claim 1, characterized in that, Before inputting the target patient's target magnetic resonance parameters into the target classification tree, the method further includes: A preset classification tree is obtained; the preset classification tree is used to determine the initial classification result corresponding to the target patient based on the clinical parameters of the target patient; the clinical parameters include at least one of the following: left ventricular ejection fraction (LVEF), QRS duration, QRS morphological characteristics, and cardiac function classification; the initial classification result includes a first initial classification and a second initial classification. The initial classification result is determined as the initial intermediate node of the target classification tree; the branches of the initial intermediate node include the first initial classification and the second initial classification; The cutoff value of the gadolinium contrast agent delayed enhancement result is determined as the first intermediate node of the first initial classification branch connection, and whether the cutoff value of the gadolinium contrast agent delayed enhancement result is satisfied is determined as the branch of the first intermediate node; The cutoff values ​​of the myocardial strain parameter and the left ventricular torsion parameter are determined as the second intermediate node connecting the second initial classification branch, and whether the first preset condition is met is determined as the branch of the second intermediate node; Based on the impact of satisfying each of the aforementioned cutoff values ​​on cardiac resynchronization, the branches of the first intermediate node and the branches of the second intermediate node are connected to the first classification result or the second classification result to obtain the target classification tree.

3. The patient classification method based on magnetic resonance parameters according to claim 2, characterized in that, The step of connecting the branches of the first intermediate node and the branches of the second intermediate node to the first classification result or the second classification result based on the impact of satisfying each of the cutoff values ​​on cardiac resynchronization includes: The branch that has a positive impact on cardiac resynchronization is connected to one of the first and second classification results; Branches that do not have a positive impact on cardiac resynchronization are connected to another branch in either the first or second classification result.

4. The patient classification method based on magnetic resonance parameters according to claim 2, characterized in that, The cardiac function classification includes grades one through four, with the grade increasing sequentially, and the grade is positively correlated with the degree of activity limitation of the patient; the morphological characteristics of the QRS complex are left bundle branch block (LBBB) or not LBBB; the first initial classification satisfies LVEF less than or equal to the first threshold, QRS complex duration greater than or equal to the second threshold, LBBB, and cardiac function classification of grade two through four; the second initial classification satisfies LVEF less than or equal to the first threshold, QRS complex duration greater than or equal to the second threshold, not LBBB, and cardiac function classification of grade two through four, or satisfies LVEF less than or equal to the first threshold, QRS complex duration less than the second threshold and greater than the third threshold, LBBB, and cardiac function classification of grade two through four.

5. The patient classification method based on magnetic resonance parameters according to claim 1, characterized in that, The output of the target classification result includes at least one of the following: The target classification results are displayed visually. A cardiac resynchronization therapy indication recommendation report is generated based on the target classification results, and the cardiac resynchronization therapy indication recommendation report is output. A pre-set artificial intelligence prediction model is used to generate a preoperative assessment report for cardiac resynchronization surgery, and the preoperative assessment report for cardiac resynchronization surgery is output.

6. A patient classification device based on magnetic resonance parameters, characterized in that, The device includes: The acquisition module is used to acquire target magnetic resonance parameters of the target patient; the target magnetic resonance parameters are selected from multiple preset magnetic resonance parameters of the sample patient based on the cardiac resynchronization label and the actual results of the occurrence of the preset endpoint event; the target magnetic resonance parameters include at least one of the following: myocardial strain parameters, gadolinium contrast agent delayed enhancement results, and left ventricular torsion parameters; The classification module is used to input the target magnetic resonance parameters of the target patient into the target classification tree, and to classify the target patient using the target classification tree to obtain the target classification result corresponding to the target patient; the target classification tree is obtained by extending the preset classification tree based on the cutoff values ​​of each target magnetic resonance parameter and the influence of satisfying each cutoff value on cardiac resynchronization. The output module is used to output the target classification result; The device further includes: The first determination module is used to determine the control patients based on the cardiac resynchronization tags of the sample patients; the cardiac resynchronization tags of the control patients are not implanted. The second determination module is used to determine the cutoff values ​​for each preset magnetic resonance imaging parameter based on the actual results of the preset endpoint events in the control patients; for the delayed enhancement result LGE of gadolinium contrast agent, the cutoff value is 8.0%; for the left ventricular torsion parameter torsion, the cutoff value is 0.34° / cm; and for the myocardial strain parameters GLS, GCS, and GRS, the cutoff values ​​are -5.9%, -6.3%, and 8.0%, respectively. The filtering module is used to filter target magnetic resonance parameters from multiple preset magnetic resonance parameters of sample patients based on various cutoff values, as well as the cardiac resynchronization labels of sample patients and the actual results of the occurrence of preset endpoint events. The screening module is also used to: determine, based on the cardiac resynchronization tag of the sample patients and whether each preset magnetic resonance parameter meets the corresponding cutoff value, identify the first type of patients and the second type of patients corresponding to each preset magnetic resonance parameter from the sample patients; the first type of patients meet the cutoff value of the preset magnetic resonance parameter and the cardiac resynchronization tag is implanted; the second type of patients meet the cutoff value of the preset magnetic resonance parameter and the cardiac resynchronization tag is not implanted; determine whether there is a significant difference in the actual results of the preset endpoint event for the first type of patients and the second type of patients corresponding to each preset magnetic resonance parameter; and determine the preset magnetic resonance parameter for which there is a significant difference in the actual results of the preset endpoint event for the first type of patients and the second type of patients as the target magnetic resonance parameter; The screening module is also used to: determine the significance value of the Kaplan-Meier survival curves of the first and second types of patients corresponding to each preset magnetic resonance parameters based on the actual results of the preset endpoint events; and determine the first and second types of patients whose Kaplan-Meier survival curve significance values ​​are less than a preset threshold as having a significant difference in the actual results of the preset endpoint events; wherein, the horizontal axis of the Kaplan-Meier survival curve represents the number of patients followed up, and the vertical axis represents the event-free survival rate, wherein the number of patients followed up represents the number of patients in the first and second patient groups whose follow-up period has not ended as the follow-up time increases, each point decrease in the Kaplan-Meier survival curve represents that one person in that group has experienced the preset endpoint event at that time point of the corresponding follow-up period, and each cross in the Kaplan-Meier survival curve represents that one person in the group whose corresponding follow-up period ended at that time point, and that person did not experience the preset endpoint event during the corresponding follow-up period; The classification module is also used for: inputting the initial classification result of the target patient into the target classification tree; if the initial classification result of the target patient is the first initial classification, determining whether the gadolinium contrast agent delayed enhancement result meets the cutoff value; the cutoff value for meeting the gadolinium contrast agent delayed enhancement result includes: the gadolinium contrast agent delayed enhancement result is negative, the gadolinium contrast agent delayed enhancement result is positive and the enhancement percentage is ∈ [0, 8%); if the gadolinium contrast agent delayed enhancement result meets the cutoff value, determining the target classification result is one of the first classification result and the second classification result; if the gadolinium contrast agent delayed enhancement result does not meet the cutoff value, determining the target classification result is the other of the first classification result and the second classification result; if the initial classification result of the target patient is the second initial classification, determining whether the first preset condition is met; the first preset condition is meeting the cutoff value of the myocardial strain parameter or the cutoff value of the left ventricular torsion parameter; if the first preset condition is met, determining the target classification result is one of the first classification result and the second classification result; if the first preset condition is not met, determining the target classification result is the other of the first classification result and the second classification result. For the left ventricular torsion parameter, the cutoff value is defined as torsion ≤ 0.34° / cm, and the cutoff value is defined as torsion > 0.34° / cm. For the myocardial strain parameter, the cutoff value is defined as GLS ≥ -5.9%, GCS ≥ -6.3%, and GRS ≤ 8.0%, and the cutoff value is defined as GLS < -5.9%, GCS < -6.3%, or GRS > 8.0%.

7. A patient classification device based on magnetic resonance parameters, characterized in that, The patient triage based on magnetic resonance parameters includes: a processor and a memory storing computer program instructions; When the processor executes the computer program instructions, it implements the patient classification method based on magnetic resonance parameters as described in any one of claims 1-5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer program instructions that, when executed by a processor, implement the patient classification method based on magnetic resonance parameters as described in any one of claims 1-5.

9. A computer program product, characterized in that, When the instructions in the computer program product are executed by the processor of the electronic device, the electronic device causes the electronic device to perform the patient classification method based on magnetic resonance parameters as described in any one of claims 1-5.

Images