	US 12,471,881 B1
	System for detecting myocardial damage based on imaging characteristics
Tongtao Cui, Guangzhou (CN); Siqi Wang, Guangzhou (CN); Tao Guo, Guangzhou (CN); Zeyang Fang, Guangzhou (CN); Youliang Wu, Guangzhou (CN); Yunyun Zhang, Guangzhou (CN); and Rongmao He, Guangzhou (CN)
Assigned to The First Affiliated Hospital of Guangzhou Medical University (Guangzhou Respiratory Center), Guangzhou (CN)
Filed by The First Affiliated Hospital of Guangzhou Medical University (Guangzhou Respiratory Center), Guangzhou (CN)
Filed on Jun. 10, 2025, as Appl. No. 19/233,793.
Application 19/233,793 is a continuation of application No. PCT/CN2025/087441, filed on Apr. 7, 2025.
Claims priority of application No. 202411188672.X (CN), filed on Aug. 28, 2024.
Int. Cl. A61B 5/00 (2006.01); A61B 5/055 (2006.01); A61B 8/00 (2006.01); A61B 8/08 (2006.01); G01R 33/48 (2006.01); G01R 33/56 (2006.01); G06T 7/00 (2017.01); G06T 7/12 (2017.01); G06T 7/136 (2017.01); G06V 10/40 (2022.01); G06V 10/74 (2022.01); G06V 10/82 (2022.01); G16H 10/60 (2018.01); G16H 30/20 (2018.01)

CPC A61B 8/0883 (2013.01) [A61B 5/0035 (2013.01); A61B 5/0044 (2013.01); A61B 5/055 (2013.01); A61B 5/7264 (2013.01); A61B 5/7425 (2013.01); A61B 8/463 (2013.01); A61B 8/488 (2013.01); A61B 8/5223 (2013.01); A61B 8/5261 (2013.01); G01R 33/4814 (2013.01); G01R 33/5608 (2013.01); G06T 7/0014 (2013.01); G06T 7/12 (2017.01); G06T 7/136 (2017.01); G06V 10/40 (2022.01); G06V 10/761 (2022.01); G06V 10/82 (2022.01); G16H 10/60 (2018.01); G16H 30/20 (2018.01); G06T 2207/10088 (2013.01); G06T 2207/20084 (2013.01); G06T 2207/20221 (2013.01); G06T 2207/30048 (2013.01); G06V 2201/031 (2022.01)]

2 Claims

1. A system for detecting myocardial damage based on imaging characteristics, characterized by comprising an image acquisition module, a feature extraction module, a data processing module, and a report generation module;

the image acquisition module is configured to acquire high-resolution cardiac imaging data; the feature extraction module is configured to extract cardiac structural and functional feature information from the acquired high-resolution cardiac imaging data; the data processing module is configured to analyze and identify myocardial damage based on the extracted feature information; the report generation module is configured to generate medical reports for physicians regarding myocardial damage;

the image acquisition module comprises an imaging acquisition unit and an imaging preprocessing unit; the image acquisition unit is configured to acquire the patient's cardiac imaging data by utilizing high-resolution medical imaging devices, comprising MRI image data and Doppler imaging data within multiple cardiac cycles; the imaging preprocessing unit is configured to preprocess the acquired high-resolution cardiac imaging data;

the feature extraction module comprises an anatomical feature extraction unit and a functional feature extraction unit; the anatomical feature extraction unit is configured to extract structural features of the patient's heart based on the preprocessed high-resolution cardiac imaging data; the functional feature extraction unit is configured to extract functional features of the patient's heart based on the preprocessed high-resolution cardiac imaging data;

the data processing module comprises a myocardial injury analysis unit and a deterioration severity assessment unit; the myocardial injury analysis unit analyzes the type and severity level of the patient's myocardial damage based on the structural features and functional features of the patient's heart; the deterioration severity assessment unit evaluates the deterioration trend of the patient's myocardial damage by incorporating analytical data on the patient's myocardial damage from historical data;

the anatomical feature extraction unit performs extraction of the structural features of the patient's heart through the following steps:

S11: acquire MRI image data from the preprocessed high-resolution cardiac imaging data; the MRI image data includes cardiac cross-sectional image data at multiple time points within multiple cardiac cycles;

S12: for all cross-sectional image data, perform threshold segmentation method to obtain cardiac boundary and internal contour information corresponding to each cardiac cross-sectional image;

S13: for the images at multiple time points on each cross-section, perform fusion based on the cardiac edge and internal contour information corresponding to each image to obtain standardized fusion images for each cross-section;

S14: input the standardized fused images on all cross-sections into a pre-trained CNN feature recognition model to obtain the structural features of the patient's heart;

Specifically, in the step S13, the standardized fused image of each cross-section is obtained through the following steps:

S131: for each cross-section, calculate a similarity value of the edge contour between each image in the cross-section and all other images in the same cross-section:

D(C_i,C_j)=max{sup_x∈C_{_i}inf_y∈C_{_j}∥x−y∥,sup_y∈C_{_j}∥x−y∥};

wherein, D (C_i, C_j) represents the similarity value of the edge contour between a certain image i and a certain image j on the same cross-section; C_irepresents the set of points of the cardiac edge and internal contour of a certain image i on the cross-section; C_jrepresents the set of points of the cardiac edge and internal contour points of a certain image j on the cross-section; x represents a certain point within the set C_j; y represents a certain point within the set C₁; ∥x−y∥ represents the Euclidean distance between point x and point y; inf represents the infimum operator, which indicates finding the minimum value among all values y belonging to the set C_j, and inf_y∈C_{_j}∥x−y∥ represents to find the value that minimizes ∥x−y∥ among all possible y ∈C_j; sup represents the supremum operator, which indicates finding the maximum value among all values x belonging to the set C_i, and sup_x∈C_{_i}inf_y∈C_{_j}∥x−y∥ represents to find the value that maximizes inf_y∈C_{_j}∥x −y∥ among all possible x ∈C_i;

S132: set a predetermined similarity threshold ε, and put all images with edge contour similarity less than the predetermined similarity threshold into the same group, and in each group, the edge contour similarities between any two images are all less than the predetermined similarity threshold E; suppose there are n groups in total;

S133: fuse the images in each group:

L_krepresents the fused image in the k-th group; km represents the total number of images in the k-th group; L_k,prepresents the p-th image in the k-th group;

S134: perform comprehensive fusion on the fused images of each group to obtain the standardized fused image of the cross-section:

L_brepresents the standardized fused image corresponding to cross-section b; β represents an amplification coefficient with a value greater than 1; M represents the total number of images on cross-section b;

the myocardial injury analysis unit performs a quantitative assessment of the type and severity of myocardial damage by integrating and concatenating the structural and functional features of the patient's heart into a pre-trained myocardial damage neural network recognition model.