Method for predicting hfpef in myocardial infarction patients based on epicardial fat volume

Abstract

This invention belongs to the field of intelligent healthcare, specifically relating to a method for predicting the risk of HFpEF in myocardial infarction patients based on epicardial fat volume. The method includes acquiring CMR images of myocardial infarction patients; extracting EAT entropy and left ventricular epicardial fat volume from the CMR images, wherein EAT entropy is the information entropy calculated based on the EAT region in the CMR images; and inputting the EAT entropy and left ventricular epicardial fat volume into a classifier to determine the risk of HFpEF in myocardial infarction patients. This application discovers that left ventricular epicardial volume and EAT entropy are independent predictors of HFpEF in myocardial infarction patients; and also discovers the predictive value of their combination for HFpEF.

Description

Technical Field

[0001] This invention relates to the field of intelligent healthcare, and more specifically, to a method, device, medium, and program product for predicting HFpEF in myocardial infarction patients based on epicardial fat volume. Background Technology

[0002] Pericardial adipose tissue, as an endocrine organ, is closely related to the occurrence and development of coronary artery disease (CAD). In particular, epicardial adipose tissue (EAT) can regulate perivascular inflammation and vascular remodeling through pro-inflammatory signals generated via paracrine pathways and trophoblast secretion mechanisms, thereby affecting the progression of coronary atherosclerosis and cardiac function. It is an important imaging indicator for cardiovascular risk stratification. Therefore, non-invasive imaging analysis of pericardial fat is helpful in identifying high-risk cardiovascular patients and is of great significance for the clinical diagnosis and treatment of CAD.

[0003] When studying epicardial adipose tissue on cardiac MRI, it is difficult to identify because the epicardial fat is thinly attached to the pericardium and difficult to discern on images. Therefore, it is often analyzed together with the pericardial fat that is close to the epicardium, referred to as pericardial fat. Currently, the study of pericardial fat generally involves the user manually outlining the pericardial contour and manually identifying high-signal fat areas within the defined pericardial contour by adjusting the image signal threshold.

[0004] However, EAT actually contains three fatty components, and different components have different effects on cardiovascular disease and will change differently in the course of the disease, thus bringing about epicardial heterogeneity. Among them, excessive activation of inflammatory cells may lead to chronic inflammation, promote atherosclerosis and heart disease, but its role in the development of heart failure with preserved ejection fraction (HFpEF) in patients with myocardial infarction (MI) is currently unclear. Summary of the Invention

[0005] In view of the above problems, the present invention provides a method for predicting the occurrence of HFpEF in myocardial infarction patients based on epicardial fat volume. The method uses cardiac magnetic resonance (CMR) to quantify the total and periventricular volume of epicardial fat (EAT) and assess its heterogeneity, and explores the heterogeneity of EAT and the predictive value of different EAT volumes combined with inflammatory cells for the occurrence of HFpEF in MI patients with normal left ventricular ejection fraction (LVEF).

[0006] This application (first aspect) discloses a method for predicting HFpEF in myocardial infarction patients based on epicardial fat volume, including:

[0007] Acquire CMR images of patients with myocardial infarction; extract EAT entropy and left ventricular epicardial fat volume based on the CMR images, wherein EAT entropy is the information entropy calculated based on the EAT region in the CMR images; input the EAT entropy and left ventricular epicardial fat volume into a classifier to obtain the risk of myocardial infarction patients developing HFpEF.

[0008] Furthermore, the EAT entropy is the Shannon entropy calculated based on the distribution of pixel values ​​in the EAT region after segmenting the CMR image into EAT regions.

[0009] Furthermore, the method for calculating the EAT entropy is as follows:

[0010]

[0011] Where i∈[0,255] represents the pixel gray value, and p(i) represents the probability of pixel value i appearing in the EAT region.

[0012] Furthermore, the frame that best reflects the target area in the image is called the keyframe, and the EAT entropy is the information entropy calculated based on the EAT region in the keyframe.

[0013] Furthermore, the EAT entropy of each frame in the CMR image is calculated first, and the average value of the EAT entropy of all frames is taken to obtain the EAT entropy.

[0014] Furthermore, the left ventricular epicardial fat volume is the left ventricular epicardial fat volume at the interventricular septum boundary.

[0015] Furthermore, clinical data of myocardial infarction patients are simultaneously acquired, including one or more of the following: whether they have diabetes, BMI; the clinical data, EAT entropy, and left ventricular epicardial fat volume are input into a classifier to obtain the risk of myocardial infarction patients developing HFpEF.

[0016] Furthermore, the method for extracting EAT entropy and left ventricular epicardial fat volume based on the CMR images is as follows:

[0017] Step 1: Input the CMR image into the epicardial fat segmentation model and output the first segmentation result. The first segmentation result includes the total epicardial fat region, the left ventricular epicardial fat region, and the right ventricular epicardial fat region.

[0018] Step 2: Morphological processing is used to eliminate the segmentation error region in the first segmentation result to obtain the second segmentation result. The morphological processing includes closing operation and connected component analysis.

[0019] Step 3: Use a conditional random field to optimize the segmentation boundary of the second segmentation result to obtain the third segmentation result;

[0020] Step 4: Based on the third segmentation result, perform three-dimensional reconstruction to obtain three-dimensional epicardial fat;

[0021] Step 5: Calculate the EAT entropy and EAT volume of the epicardial fat based on the three-dimensional epicardial fat.

[0022] Furthermore, the left ventricular ejection fraction (LVEF) of the patients with myocardial infarction was normal.

[0023] Furthermore, the method for constructing the classifier includes:

[0024] The training set is obtained by acquiring the EAT volume, EAT entropy, and labels of myocardial infarction patients, wherein the labels are those that have occurred HFpEF and those that have not occurred HFpEF.

[0025] The EAT volume and EAT entropy are input into the machine learning model to output the predicted label. The predicted label is compared with the label, and the machine learning model is optimized and iterated until the stopping condition is met to obtain the classifier.

[0026] Furthermore, the machine learning model includes any one or more of the following: support vector machine, logistic regression, and XGBoost.

[0027] Furthermore, the method for constructing the epicardial fat segmentation model is as follows:

[0028] Obtain CMR images and annotations for the training set, including: epicardial fat, left ventricular epicardial fat, and right ventricular epicardial fat;

[0029] The epicardial fat segmentation model is obtained by inputting the CMR images and annotations from the training set into the neural network model and iteratively training it.

[0030] Furthermore, the neural network model includes any one or more of the following: U-Net, SwinTransformer, LTSM.

[0031] The second aspect of this application discloses a predictive system for HFpEF in myocardial infarction patients based on epicardial fat volume, comprising:

[0032] Acquisition module: Used to acquire CMR images of patients with myocardial infarction;

[0033] Feature extraction module: used to extract EAT entropy and left ventricular epicardial fat volume based on the CMR image, wherein the EAT entropy is the information entropy calculated based on the EAT region in the CMR image;

[0034] Prediction module: Used to input EAT entropy and left ventricular epicardial fat volume into the classifier to obtain the risk of HFpEF in patients with myocardial infarction.

[0035] A third aspect of this application discloses a computer device, the device comprising: a memory and a processor; the memory being used to store program instructions; the processor being used to invoke the program instructions, which, when executed, are used to perform the steps of the method described above.

[0036] The fourth aspect of this application discloses a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described above.

[0037] The fifth aspect of this application discloses a computer program product, including a computer program that, when executed by a processor, implements the steps of the above-described method.

[0038] This application has the following beneficial effects:

[0039] (1) This application is the first to finely segment and study epicardial fat, and found that left ventricular epicardial volume is an independent predictor of HFpEF in patients with myocardial infarction.

[0040] (2) This application incorporates Shannon entropy into the calculation of the EAT region and finds that EAT entropy is an independent predictor of HFpEF in patients with myocardial infarction.

[0041] (3) The prediction effect is better when the left ventricular epicardial volume and EAT entropy are combined, as discovered in this application;

[0042] (4) This application proposes a model and method for automatically segmenting CMR and calculating left ventricular epicardial volume and EAT entropy through artificial intelligence model, optimizes the calculation process of left ventricular epicardial volume and EAT entropy, firstly proposes to use artificial intelligence deep learning algorithm model to realize automatic segmentation and automatic reconstruction of pericardial fat three-dimensional model, and analyzes the volume, energy, entropy and other parameters of the segmented epicardial fat, which provides intuitive quantitative parameter support for analyzing epicardial fat and its impact on heart disease. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1 This is a schematic diagram of the method flow provided in the first aspect of the present invention;

[0045] Figure 2 This is a schematic diagram of a program product provided in the second aspect of the present invention;

[0046] Figure 3 This is a schematic diagram of a computer device provided in an embodiment of the present invention;

[0047] Figure 4 This is a schematic diagram of the architecture of an exemplary computing device provided in an embodiment of the present invention;

[0048] Figure 5 This is a schematic diagram of the storage medium provided in an embodiment of the present invention;

[0049] Figure 6 This is a schematic diagram of a method for labeling or measuring overall and local EAT provided in an embodiment of the present invention;

[0050] Figure 7 This is a schematic diagram of an EAT entropy measurement method provided in an embodiment of the present invention;

[0051] Figure 8 This is a flowchart of a model training method provided in an embodiment of the present invention;

[0052] Figure 9 This is a schematic diagram illustrating the comparison of EAT parameters between the No-HFpEF group and the HFpEF group provided in an embodiment of the present invention;

[0053] Figure 10 This is a schematic diagram illustrating the correlation analysis between EAT and inflammatory cells provided in an embodiment of the present invention;

[0054] Figure 11 This is a schematic diagram of ROC curve analysis for predicting HFpEF in MI using EAT parameters, provided in an embodiment of the present invention.

[0055] Figure 12 This is a schematic diagram of a Kaplan-Meier survival curve provided in an embodiment of the present invention. Detailed Implementation

[0056] To enable those skilled in the art to better understand the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings.

[0057] In some of the processes described in the specification, claims, and accompanying drawings of this invention, multiple operations appearing in a specific order are included. However, it should be clearly understood that these operations may not be executed in the order they appear herein, or may be executed in parallel. The operation numbers, such as S101, S102, etc., are merely used to distinguish different operations and do not represent any execution order. Furthermore, these processes may include more or fewer operations, and these operations may be executed sequentially or in parallel. It should be noted that the descriptions such as "first," "second," etc., in this document are used to distinguish different messages, devices, modules, etc., and do not represent a sequential order, nor do they limit "first" and "second" to different types.

[0058] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0059] Figure 1 This is a schematic flowchart of a method for predicting HFpEF in myocardial infarction patients based on the assessment of epicardial fat volume, provided by an embodiment of the present invention. Specifically, the method includes the following steps:

[0060] S101 acquires CMR images of patients with myocardial infarction;

[0061] S102: Extract EAT entropy and left ventricular epicardial fat volume based on the CMR image, wherein the EAT entropy is the information entropy calculated based on the EAT region in the CMR image;

[0062] S103: Input the EAT entropy and left ventricular epicardial fat volume into the classifier to obtain the risk of HFpEF in patients with myocardial infarction. Since this application requires extracting the volume of different components of epicardial fat, an automatic segmentation calculation model for epicardial fat volume is first established by training the model.

[0063] I. Model Construction

[0064] This application first manually annotates cardiac MRI images and then trains an artificial intelligence deep learning algorithm model to automatically segment and reconstruct epicardial fat on short-axis cine sequences of cardiac MRI images. The regions of epicardial fat are marked with color on the original short-axis cine sequence image, a 3D model of epicardial fat that can be viewed by 360-degree rotation is reconstructed, and parameters such as the volume of epicardial fat are analyzed.

[0065] This invention provides an automated segmentation method for epicardial fat in CMR (cardiac magnetic resonance imaging). It employs deep learning combined with medical image processing to improve segmentation accuracy and automation. Traditional manual segmentation methods require physicians to draw lines frame by frame, which is not only time-consuming but also susceptible to subjective biases, making it difficult to guarantee consistency. This invention, based on an improved U-Net / Transformer segmentation network and combined with morphological post-processing and 3D reconstruction algorithms, achieves efficient and accurate automated segmentation of pericardial fat and provides quantitative fat analysis.

[0066] The automatic pericardial fat segmentation method used in this invention is as follows: Figure 8 As shown, it includes the following steps:

[0067] Step 1: Obtain CMR image dataset

[0068] In some embodiments, the image dataset includes a dataset of healthy individuals.

[0069] In some embodiments, the image dataset also includes an image dataset of patients with myocardial infarction.

[0070] Image preprocessing: Due to the low contrast of CMR images, the gray values ​​of epicardial fat are close to those of surrounding tissues, which can easily lead to missegmentation. Therefore, the following preprocessing is required: Normalization: The gray values ​​of CMR images are normalized to [0,1] to reduce the influence of different scanning parameters; Non-local mean filtering (NLM) is used to remove noise while preserving fat boundary features; Adaptive Gaussian filtering is used to reduce artifact interference and improve the quality of model input.

[0071] Step 2: Label the training set

[0072] To ensure that the deep learning model can accurately identify pericardial fat, a high-quality training dataset needs to be constructed. The process is as follows:

[0073] Doctor's note: The CMR images were manually delineated by an experienced cardiovascular imaging specialist to determine the precise boundaries of the fat surrounding the heart.

[0074] The annotations are as follows Figure 6 As shown, Figure 6 -A is the original CMR image. Figure 6 In -B, the red circle delineates the visceral epicardium, the green circle delineates the parietal fat layer, and the pale yellow area between the green and red circles represents the total epicardial fat.

[0075] Figure 6 The gray circle in -C delineates the total epicardial fat excluding the left ventricular epicardial fat. The pale yellow area between the green and red circles, minus the gray circle, represents the right ventricular epicardial fat.

[0076] Figure 6 The gray circle in -D delineates the total epicardial fat, excluding the epicardial fat of the ventricles. The pale yellow area between the green and red circles, minus the portion within the gray circle, represents the epicardial fat of the left ventricle.

[0077] The three types of EAT identified by this labeling are: ① total epicardial fat, ② right ventricular epicardial fat, and ③ left ventricular epicardial fat.

[0078] The above annotations on the CMR images in the training set serve as supervision during model training.

[0079] Semi-automatic annotation tools can be used to assist doctors during annotation, improving annotation efficiency and reducing subjective errors.

[0080] Step 3: Preliminary segmentation (deep learning model)

[0081] In some embodiments, the U-Net network is selected as the backbone network for segmentation: it is suitable for local feature learning and can handle regularly shaped fat regions well.

[0082] In some embodiments, the segmented backbone network selects a Transformer (Swin Transformer) to capture long-range dependencies and enhance the ability to identify complex fat regions.

[0083] In some embodiments, a hybrid loss function is used during segmentation model training: Dice Loss (to improve the ability to recognize small objects) + Focal Loss (to reduce the impact of hard-to-segment regions).

[0084] In some embodiments, the training process uses a two-stage training: Stage 1 (coarse segmentation): This stage utilizes global contextual information to locate the approximate region of pericardial fat. Stage 2 (fine segmentation): This stage uses high-resolution feature maps to refine the boundaries and improve segmentation accuracy.

[0085] Step 4: Error optimization and post-processing after model segmentation

[0086] The segmentation results may contain errors, such as missegmentation (non-fat areas are mistakenly identified as fat) and omissions (some fat is not identified).

[0087] To address this, the following optimization strategies were adopted: 1. Morphological processing, such as closing operations: filling small holes to improve the coherence of fat regions; connected component analysis: removing isolated small regions to ensure that the final result only includes fat around the heart. 2. CRF (Conditional Random Field) optimization: combining pixel grayscale information with spatial neighborhood information to optimize segmentation boundaries and improve accuracy.

[0088] Step 5: 3D Reconstruction

[0089] To provide a more intuitive analysis, the Marching Cubes algorithm is used to reconstruct the segmentation results in 3D: Input: Post-processed 2D fat mask image; Output: 3D pericardial fat / 3D epicardial fat model. This allows doctors to observe fat distribution in a 3D environment and can be used for clinical applications such as surgical planning and disease assessment.

[0090] Step 6: Quantitative analysis of pericardial fat

[0091] After segmentation, the system automatically calculates the following quantitative indicators of fat:

[0092] ① Total fat volume (EAT Volume): The total volume of epicardial fat;

[0093] ②LV EAT: Left ventricular epicardial volume;

[0094] ③RV EAT: Right ventricular epicardial volume;

[0095] ④ Fat density (HU value): Calculate the average gray value of fat to analyze its biological activity.

[0096] ⑤ Fat entropy value: reflects the uniformity of fat distribution and can be used to identify abnormal adipose tissue.

[0097] ⑥ Cardiac function parameters: Combined with left ventricular ejection fraction (LVEF), the effect of pericardial fat on cardiac function was analyzed.

[0098] II. Research Methods:

[0099] 2.1 Research Subjects

[0100] This is a historical cohort study that collected patients diagnosed with MI by clinical and CMR at the Second Affiliated Hospital of Kunming Medical University between January 2015 and July 2023, but with normal LVEF. Follow-up was conducted with the occurrence of HFpEF as the endpoint event, and patients were divided into a no-HFpEF group and an HFpEF group.

[0101] 2.2 Image Processing:

[0102] For the acquired CMR images, parameters such as cardiac structure, function, EAT volume (including total epicardial fat volume, right ventricular epicardial fat, and left ventricular epicardial fat) and infarct volume were obtained using the CMR post-processing software CVI-42. EAT heterogeneity parameters were obtained using Python software.

[0103] EAT heterogeneity parameters include: LADSV, LAESV, EDVI, ESVI, CO, CI, infarct area, GRS, GCS, and GLS.

[0104] In some embodiments, the acquired CMR images are input into a pre-constructed model for automatic segmentation, post-processing optimization, and output parameters such as EAT volume (including total epicardial fat volume, ② right ventricular epicardial fat, ③ left ventricular epicardial fat).

[0105] The formula for calculating the EAT entropy of a single layer using Python software is as follows:

[0106]

[0107] For a grayscale image (8-bit, pixel value range 0~255), the formula for calculating the information entropy H is:

[0108]

[0109] Where p(i) represents the probability of pixel value i appearing in the image (i.e., the frequency of the pixel value divided by the total number of pixels).

[0110] The steps for calculating the EAT entropy in this application (Python example) are as follows:

[0111] Step 1: Calculate the pixel histogram

[0112] Calculate the EAT region segmented from the image ( Figure 7 -A, where Figure 7 -A is the corresponding Figure 6 The frequency of occurrence of each pixel value in the EAT region between the green and red lines is normalized to obtain the probability distribution p(i). Figure 7 -B is shown).

[0113] Step 2: Calculate entropy

[0114] Iterate through all possible pixel values ​​(0~255) and calculate the EAT entropy of the current slice according to the following formula.

[0115] In some embodiments, the frame that best represents the CMR of the heart region is selected as the key frame, and the EAT entropy in the key frame is calculated.

[0116] In some embodiments, the method for extracting EAT entropy is to simultaneously save the original and drawn PNG images of each layer, based on the original EAT collection. Importing the images of each layer into Python software can automatically quantify the entropy value of each layer, and finally take the average value.

[0117] In some embodiments, the outlined image is imported into Python. Python combines all pixel values ​​from each layer to calculate the frequency of each pixel value in the segmented EAT region of the image and normalizes it to obtain a probability distribution. p ( i ), and further based on EAT at all levels p ( i The EAT entropy is calculated.

[0118] In some embodiments, LV EAT is also based on the original EAT collection, with the interventricular septum as the boundary to remove RVEAT. The standard EAT collection method is as follows: in the short-axis sequence of the two chamber heart, the visceral epicardium and the parietal fat layer are manually marked at the end of diastole, and the high signal fat tissue between the two is marked with a signal intensity threshold, while avoiding the coronary arteries and pericardial fat.

[0119] 2.3 Statistical Analysis

[0120] Independent samples t-tests, nonparametric tests, and chi-square tests were used to analyze the differences in clinical baseline data and CMR indices between the two groups of patients. Spearman rank correlation analysis was used to analyze the correlation between EAT parameters and inflammatory cells. Univariate and multivariate Cox regression were further performed to analyze the predictive value of each index for HFpEF in MI patients. ROC curves were plotted to evaluate the predictive power of each parameter for HFpEF. Finally, based on the optimal cutoff value, Kaplan-Meier event survival curves were used to display the cumulative incidence rate curve.

[0121] 3 Results

[0122] 3.1 Baseline Data Comparison

[0123] In the HFpEF group, BMI was 24.40 (22.23, 26.73) vs 23.40 (21.60, 25.55) (kg / m2), diabetes n (%) was 29 (39.19) vs 27 (20.93), renal failure n (%) was 7 (9.46) vs 3 (2.33), white blood cell count was 7.38 (6.15, 9.02) vs 7.12 (5.96, 7.95) (10*9 / L), neutrophils were 4.58 (3.59, 5.77) vs 4.06 (3.22, 4.81) (10*9 / L), and monocyte count was 0.51 (0.38, 0.65) vs. The total EAT volume (0.45 (0.37, 0.53) (10*9 / L), total EAT volume (69.43±21.21 vs. 62.18±19.93 ml), lower venous endothelial (LV) EAT volume (24.81±7.39 vs. 18.45 (13.03, 25.30 ml), lower venous endothelial (RV) EAT volume (44.61±15.03 vs. 40.81±13.68 ml), and lower venous estrogen (LAESV) volume (56.18 (42.13, 73.00) vs. 51.24 (39.46, 64.44 ml)) were all higher than those in the group without HFpEF.

[0124] The EAT entropy of 6.40 (6.22, 6.86) vs 6.75 (6.26, 7.15) and TG of 1.40 (0.86, 2.05) vs 1.63 (1.20, 2.40) were lower than those of the group without HFpEF (p<0.05).

[0125] There were no statistically significant differences between the two groups in terms of gender, age, smoking history, history of hypertension, systolic blood pressure, diastolic blood pressure, pulse pressure, NYHA classification, LDL, HDL, TC, lymphocytes, eosinophils, basophils, erythrocytes, urea, creatinine, RV EAT, LVEF, LADSV, EDVI, ESVI, CO, CI, infarct area, overall LV strain, mitral regurgitation, tricuspid regurgitation, aortic regurgitation, pulmonary hypertension, diseased vessel branches, and diseased vessels (P > 0.05) (as shown in Table 1 and...). Figure 9 (As shown).

[0126] Table 1 Comparison of clinical baseline data between the two groups

[0127]

[0128]

[0129]

[0130] 3.2 Correlation analysis between EAT and inflammatory cells

[0131] Total EAT, LV EAT, and RV EAT all showed a mild to moderate positive correlation with leukocytes and monocytes. Among these, total EAT showed the strongest correlation with monocytes (r=0.658, p<0.001), while RV EAT showed the strongest correlation with leukocytes (r=0.469, p<0.001). (Table 2 and...) Figure 10 )

[0132] Table 2 Correlation analysis between EAT and inflammatory cells

[0133]

[0134] 3.3 Analysis of independent risk factors for HFpEF in MI patients

[0135] Indicators showing differences between the two groups and considered predictors of HFpEF were used as independent variables, and HFpEF was used as the dependent variable. Univariate Cox regression analysis showed that the risk factors for HFpEF were renal failure, diabetes, BMI, monocytes, total EAT, LV EAT, and EAT entropy (p < 0.05). In the multivariate Cox analysis model, LV EAT (HR: 1.094, 95% CI: 1.019–1.174), EAT entropy (HR: 0.398, 95% CI: 0.220–0.721), diabetes (HR: 2.320, 95% CI: 1.398–3.849), and BMI (HR: 1.093, 95% CI: 1.025–1.165) were independent predictors of HFpEF (Table 3). ROC analysis yielded the following results: AUC for BMI was 0.581, AUC for LV EAT was 0.628, and AUC for EAT entropy was 0.405. The joint predictive value of LV EAT and EAT entropy was the highest, with an AUC of 0.740 (Table 4 and). Figure 11 After a median follow-up of 27 years (range 2–120 months), Kaplan-Meier survival curves showed that lower ventricular efferent tachycardia (LV) greater than 16.42 ml was associated with the occurrence of high-frequency refractory embolism (HFpEF), while EAT entropy was not. Figure 12 ).

[0136] Table 3. Univariate and Multivariate Cox Regression Analysis

[0137]

[0138] Table 4 ROC Analysis

[0139]

[0140] Table Notes: * This indicates that p < 0.05.

[0141] A total of 203 eligible MI patients were included, of whom 74 developed HFpEF and 129 did not. There were no differences in age, sex, or infarct volume between the two groups; however, there were statistically significant differences in BMI, diabetes, renal failure, white blood cell count, neutrophil count, monocyte count, total EAT, EAT entropy, left ventricular paraventricular EAT (LV EAT), LAESV, and TG (P < 0.05). Both total and local EAT were positively correlated with white blood cell count and monocyte count. Univariate and multivariate Cox regression analyses showed that BMI, diabetes, LV EAT, and EAT entropy were independent risk factors for HFpEF. Further ROC analysis showed that the AUC for BMI was 0.581, for LV EAT was 0.628, and for EAT entropy was 0.405. The combined predictive value of LV EAT and EAT entropy was the highest, with an AUC of 0.740. After a median follow-up of 27 years (range 2-120 months), Kaplan-Meier survival curves showed that LV EAT greater than 16.42 ml was associated with the occurrence of HFpEF, but not with EAT entropy.

[0142] Conclusion: In MI patients with normal LVEF, the occurrence of HFpEF was not correlated with MI volume. However, BMI, diabetes, LV EAT, and EAT entropy were independent risk factors for HFpEF and had good predictive value, with the combination of EAT entropy and LV EAT showing the highest predictive efficacy. Furthermore, both total and local EAT were moderately positively correlated with white blood cell and monocyte counts. This suggests that in addition to traditional indicators, clinicians should pay more attention to EAT heterogeneity and paraventricular EAT in MI patients with normal LVEF to improve the prediction of HFpEF risk.

[0143] In some embodiments, CMR images of the patient are acquired, and EAT entropy and LVEAT (volume) are extracted based on the steps in the first part. These are then input into a classifier model. Based on the data, the prediction result for the patient to develop HFpEF is 1, indicating that the risk of developing HFpEF is high and targeted prevention should be carried out as early as possible.

[0144] Figure 3 This is a schematic diagram of a computer device provided in an embodiment of the present invention, such as... Figure 3 As shown, the device 2000 may include: one or more processors 2010 and one or more memories 2020; wherein the memories store computer-readable code that, when run by the one or more processors, can perform the methods described above.

[0145] The processor in this embodiment can be an integrated circuit chip with signal processing capabilities. The processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), an off-the-shelf programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, operations, and logic block diagrams disclosed in this embodiment. The general-purpose processor can be a microprocessor or any conventional processor, and can be based on an x86 or ARM architecture.

[0146] In general, the various exemplary embodiments of this disclosure can be implemented in hardware or dedicated circuitry, software, firmware, logic, or any combination thereof. Some aspects can be implemented in hardware, while others can be implemented in firmware or software that can be executed by a controller, microprocessor, or other computing device. When aspects of embodiments of this disclosure are illustrated or described as block diagrams, flowcharts, or using some other graphical representation, it will be understood that the blocks, apparatuses, systems, techniques, or methods described herein can be implemented as non-limiting examples in hardware, software, firmware, dedicated circuitry or logic, general-purpose hardware or controllers or other computing devices, or some combination thereof.

[0147] For example, the method or apparatus according to embodiments of this disclosure can also be used by means of Figure 4 The architecture of the computing device 3000 shown is used for implementation. For example... Figure 4 As shown, the computing device 3000 may include a bus 3010, one or more CPUs 3020, a read-only memory (ROM) 3030, a random access memory (RAM) 3040, a communication port 3050 connected to a network, an input / output component 3060, a hard disk 3070, etc. The storage devices in the computing device 3000, such as the ROM 3030 or the hard disk 3070, may store various data or files used for processing and / or communication of the methods provided in this disclosure, as well as program instructions executed by the CPU. The computing device 3000 may also include a user interface 3080. Of course, Figure 4 The architecture shown is merely exemplary and can be omitted as needed when implementing different devices. Figure 4 One or more components in the computing device shown.

[0148] This invention also includes a computer-readable storage medium, such as... Figure 5The diagram illustrates a storage medium 4000 provided in an embodiment of the present invention. The computer storage medium 4020 stores computer-readable instructions 4010. When the computer-readable instructions 4010 are executed by a processor, the method described above according to embodiments of the present disclosure can be performed. The computer-readable storage medium in the embodiments of the present disclosure may be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. Non-volatile memory may be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory may be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), Synchronous Dynamic Random Access Memory (SDRAM), Double Data Rate Synchronous Dynamic Random Access Memory (DDRSDRAM), Enhanced Synchronous Dynamic Random Access Memory (ESDRAM), Synchronous Link Dynamic Random Access Memory (SLDRAM), and Direct Memory Bus Random Access Memory (DR RAM). It should be noted that the memory used in the methods described herein is intended to include, but is not limited to, these and any other suitable types of memory.

[0149] This disclosure also provides a computer program product or computer program that, when executed by a processor, implements the steps of the above-described method, such as... Figure 2 As shown, the computer program product or computer program includes:

[0150] Acquisition module 201: Used to acquire CMR images of patients with myocardial infarction;

[0151] Feature extraction module 202: used to extract EAT entropy and left ventricular epicardial fat volume based on the CMR image, wherein the EAT entropy is the information entropy calculated based on the EAT region in the CMR image;

[0152] Prediction module 203: Used to input EAT entropy and left ventricular epicardial fat volume into the classifier to obtain the risk of HFpEF in patients with myocardial infarction.

[0153] It should be noted that the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0154] In general, the various exemplary embodiments of this disclosure can be implemented in hardware or dedicated circuitry, software, firmware, logic, or any combination thereof. Some aspects can be implemented in hardware, while others can be implemented in firmware or software that can be executed by a controller, microprocessor, or other computing device. When aspects of embodiments of this disclosure are illustrated or described as block diagrams, flowcharts, or using some other graphical representation, it will be understood that the blocks, apparatuses, systems, techniques, or methods described herein can be implemented as non-limiting examples in hardware, software, firmware, dedicated circuitry or logic, general-purpose hardware or controllers or other computing devices, or some combination thereof.

[0155] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0156] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection between apparatuses or units through some interfaces, and may be electrical, mechanical, or other forms.

[0157] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0158] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0159] The exemplary embodiments of this disclosure described in detail above are merely illustrative and not restrictive. Those skilled in the art will understand that various modifications and combinations can be made to these embodiments or their features without departing from the principles and spirit of this disclosure, and such modifications should fall within the scope of this disclosure.

Claims

1. A method for predicting HFpEF in myocardial infarction patients based on epicardial fat volume, characterized in that, The method includes: Obtain CMR images of patients with myocardial infarction. The left ventricular ejection fraction of patients with myocardial infarction is normal. Image features are extracted based on the CMR images, including: EAT entropy calculated based on the three-dimensional reconstructed epicardial fat region and left ventricular epicardial fat volume at the interventricular septum boundary; the EAT entropy is the Shannon entropy calculated based on the distribution of pixel values ​​in the epicardial fat region after segmenting the epicardial fat region from the CMR images. The image features are input into a classifier to obtain the result of whether a myocardial infarction patient will develop HFpEF.

2. The method for predicting HFpEF in myocardial infarction patients based on epicardial fat volume according to claim 1, characterized in that, The method for calculating the EAT entropy is as follows: Where i∈[0,255] represents the pixel gray value, and p(i) represents the probability of pixel value i appearing in the epicardial fat region.

3. The method for predicting HFpEF in myocardial infarction patients based on epicardial fat volume according to claim 1, characterized in that, Simultaneously, clinical data of patients with myocardial infarction are acquired, including one or more of the following: whether they have diabetes, BMI; the clinical data and imaging features are input into a classifier to obtain the result of whether HFpEF will occur.

4. The method for predicting HFpEF in myocardial infarction patients based on epicardial fat volume according to claim 1, characterized in that, The method for extracting image features based on the CMR images is as follows: Step 1: Input the CMR image into the epicardial fat segmentation model and output the first segmentation result. The first segmentation result includes the total epicardial fat region, the left ventricular epicardial fat region, and the right ventricular epicardial fat region. Step 2: Morphological processing is used to eliminate the segmentation error region in the first segmentation result to obtain the second segmentation result. The morphological processing includes closing operation and connected component analysis. Step 3: Use a conditional random field to optimize the segmentation boundary of the second segmentation result to obtain the third segmentation result; Step 4: Based on the third segmentation result, perform three-dimensional reconstruction to obtain three-dimensional epicardial fat; Step 5: Calculate the EAT entropy and EAT volume of the epicardial fat based on the three-dimensional epicardial fat.

5. The method for predicting HFpEF in myocardial infarction patients based on epicardial fat volume according to claim 1, characterized in that, The method for constructing the classifier includes: The training set is obtained by acquiring the EAT volume, EAT entropy, and labels of myocardial infarction patients, wherein the labels are those that have occurred HFpEF and those that have not occurred HFpEF. The EAT volume and EAT entropy are input into the machine learning model to output the predicted label. The predicted label is compared with the label, and the machine learning model is optimized and iterated until the stopping condition is met to obtain the classifier.

6. A computer device, characterized in that, The device includes: a memory and a processor; the memory is used to store a computer program; the processor executes the computer program to implement the steps of the method according to any one of claims 1-5.

7. A computer-readable storage medium, characterized in that, It stores a computer program that, when executed by a processor, implements the steps of the method as described in any one of claims 1-5.

8. A computer program product, comprising a computer program, characterized in that, When executed by a processor, the computer program implements the steps of the method described in any one of claims 1-5.

Images