TWM575602U - Automatic reporting system of cardiovascular calcium or stenosis degree - Google Patents

Abstract

The creation provides an automatic reporting system of cardiovascular calcium or stenosis degree, comprising a server saving a database having a number of cardiovascular images with various calcification or stenosis locations and degree; an image processing module; an artificial neural network learning module; a module for recognizing and marking cardiovascular calcium or stenosis; and a figure producing module.

Description

Automatic reporting system for cardiac vascular calcification or stenosis

The present invention relates to the field of detection of cardiac vascular calcification or stenosis. In particular, the present invention relates to an automated report generation system for cardiac vascular calcification or stenosis.

Cardiovascular disease occupies the top three causes of death in Taiwan. There are no obvious symptoms of cardiovascular disease. Ninety percent of patients died before hospital death. Most of the patients were found to have reduced blood volume due to cardiovascular stenosis. The cause of cardiovascular stenosis is vascular calcification. When there is excessive deposition of calcium on the blood vessel wall, it can be called vascular calcification, which causes blood vessel stenosis. Calcification occurs because the patient's blood lipids deposit on the walls of the coronary arteries, slowly forming fibrotic plaques, some of the smooth muscle cells are filled with fat, the core is gradually necrotic, and calcification thickens with time. Analysis of the calcification of cardiovascular vessels can predict the risk of stenosis and further prevent the risk of future myocardial infarction. Three high (high blood pressure, high blood fat, high blood sugar) patients, smokers, seniors and other high-risk groups need effective screening. In addition, calcification analysis can reflect the inflammation associated with long-term physical imbalances in patients, thereby presuming ongoing diseases, in addition to cardiovascular diseases, as well as diabetes, metabolic syndrome, and cancer. There is almost no risk of cardiovascular disease in the calcification index of 0. Compared with those without calcification, about three-quarters of those with a calcification index of 400 or more have coronary artery stenosis; those with a calcification index of more than 1,000 are significantly associated with cancer. Therefore, calcification analysis can be used as a basis for predicting risk estimates, and those at high risk should receive coronary CT angiography (CCTA) for developing agents. Calcification analysis quantified the values, post-processing measurements and output reports from computed tomography. CCTA refers to multi-section computed tomography images by professional physicians. It relies on the experience of doctors to perform manual calculation, labeling, and mapping of coronary artery stenosis distribution for CCTA images. Therefore, there is still a need to develop an automatic report generation system to reduce errors and improve efficiency due to differences in physician experience.

The present invention provides an automatic report generation system for the location and extent of cardiac vascular calcification or stenosis, comprising: a server having a database for storing vascular images of different angles and positions and degrees of calcification or stenosis; The group performs image processing on the vascular image file and intercepts the image file of only the blood vessel; a type of neural network learning module that uses the neural network training model for vascular calcification or stenosis; And a labeling module that intercepts a portion of the calcified or narrowed blood vessel, and uses different angles of blood vessels to determine the degree of calcification or stenosis, and then automatically displays the calcification or stenosis position and the degree of calcification or stenosis using the trained model. Or manually visualize the location of calcification or stenosis and calcification or stenosis with the naked eye and enter the results into the module; and a pattern generation module to establish a standard pattern of cardiovascular vessels and map the resulting calcification or stenosis to the standard pattern Corresponding position and automatically generate a detection result pattern marked with calcification or stenosis position and extent. In some embodiments, the server stores information about vessel angles at different angles and degrees of calcification or stenosis. These different angles of vascular mapping and information on the extent of calcification or stenosis are collected from multiple patients to establish big data. Coronary calcification is a marker of plaque on the blood vessels (fat deposits) or atherosclerosis. Soft plaques occur earlier and are less stable, and calcified plaques occur later. The higher the calcification index, the higher the degree of coronary artery stenosis. In some embodiments, calcification causes vascular stenosis, and vascular calcification or stenosis is classified as <25%, 25-49%, 50-69%, and >70% stenosis. The image processing module performs image processing on the blood vessel image file, and intercepts the image file of only the blood vessel. Next, the neural network learning module is used to train the model of the vascular calcification or stenosis with a neural network. The training model firstly performs supervised learning to identify whether it is a vascular calcification point. After a moderate number of learning, the neural network can learn the judgment rule of whether the blood vessel has calcification. A neural network is a mathematical model or computational model that mimics the biological neural network (the central nervous system of animals, especially the structure and function of the brain, for estimating or approximating functions. By supervised learning, The judgment rule of whether the blood vessel is calcified is summarized, thereby automatically achieving the judgment of whether the blood vessel is calcified or narrowed. In a specific embodiment, the neural network is a Convolutional Neural Networks (CNN) learning module. The vascular calcification or stenosis identification and labeling module intercepts a section of the calcified or stenotic vessel, and uses different angles of blood vessels to determine the degree of calcification or stenosis, and then automatically displays the calcification or stenosis position using the trained model. And the degree of calcification or stenosis, or manually visualize the location of calcification or stenosis and calcification or stenosis with the naked eye and enter the results into the module. The pattern generation module first establishes a standard pattern of cardiovascular and the resulting calcification or stenosis position Corresponds to the corresponding position of the standard pattern and automatically generates the position and degree marked with calcification or stenosis First, a pattern of detection results. First, establish a standard pattern of cardiac blood vessels, and then map the calcified or stenotic position and the degree of calcification or stenosis to the standard pattern with the trained model; the corresponding method is large The blood vessels can be segmented by the segmentation method provided by the physician, and the position ratio is calculated after segmentation, and the calcification or stenosis position is marked to the corresponding position in the standard pattern. The blood vessel branch of the strip can be directly calculated by calculating the position ratio. The calcification or stenosis position is indicated to the corresponding position in the standard pattern. In some embodiments, the novel system may further comprise a data output module for reporting the output of the stenosis degree and the calcification position. For example, the information also includes blood and science information, which are also included in the report.

Referring to the first figure, the novel system comprises six components, namely a server 1, an image processing module 2, a neural network learning module 3, a vascular calcification or stenosis degree recognition and labeling module 4, and a graph generating module. 5. The server 1 has a database for storing the vascular image files of different angles of the patient and the degree of calcification or stenosis. The image processing module 2 performs image processing on the blood vessel image file, and extracts a picture file that is only a blood vessel. The vascular calcification or stenosis features were trained using a neural network learning module 3 model. A section of the calcified or stenotic vessel is taken for a period of time, and the calcification or stenosis degree is determined by using blood vessels of different angles, and then the trained model is used to identify the vascular calcification or stenosis and the indicator module 4 automatically determines where the calcification or stenosis is located. Which segment of the blood vessel is automatically marked with calcification or stenosis and calcification or stenosis; or the calcification or stenosis position and calcification or stenosis can be manually indicated by the naked eye and the result can be entered into the module. Next, a cardiovascular standard pattern is established, and the resulting calcification or stenosis position is mapped to the corresponding position of the standard pattern by the chart generation module 5 and a detection result pattern indicating the position and extent of calcification or stenosis is automatically generated. .

1‧‧‧Server

2‧‧‧Image Processing Module

3‧‧‧ class neural network learning module

4‧‧‧Vascular calcification or stenosis identification and labeling module

5‧‧‧Chart generation module

The first image is a schematic representation of the linkage of the various components of an automated report generation system for cardiac vascular calcification sites.

Claims (8)

An automatic report generation system for the location and extent of cardiac vascular calcification or stenosis, comprising: a server having a database for storing vascular images of different angles and degrees of calcification or stenosis; an image processing module The vascular image file is used for image processing, and the image file of only the blood vessel is intercepted; a type of neural network learning module which uses the neural network training model for vascular calcification or stenosis; a blood vessel calcification or stenosis degree identification and labeling module It will cut a section of the calcified or stenotic blood vessels, and use different angles of blood vessels to judge the degree of calcification or stenosis, and then use the trained model to automatically mark the calcification or stenosis position and the degree of calcification or stenosis, or artificially Mark calcification or stenosis and calcification or stenosis and enter the results into the module; and a pattern generation module to establish a standard pattern of cardiovascular vessels and map the resulting calcification or stenosis to the corresponding position of the standard pattern A pattern of detection results indicating the location and extent of calcification or stenosis is automatically generated. The system of claim 1, wherein the different angles of the vascular image and the location and extent of calcification or stenosis are collected from a plurality of patients to establish big data. The system of claim 1, wherein the degree of calcification or stenosis is expressed as a degree of vascular stenosis classified as 0, 25%, 50%, and 75% of the degree of stenosis. The system of claim 1, wherein the image of the calcification or stenosis position and the degree of calcification or stenosis corresponds to the standard pattern, wherein the large blood vessel can be segmented by a segmentation method provided by a physician, and the segmentation is calculated. After the position ratio, the calcification or stenosis position is marked to the corresponding position in the standard pattern. The system of claim 1, wherein the image of the calcification or stenosis position and the degree of calcification or stenosis corresponds to the standard pattern, and the vascular branch line of the strip can directly mark the calcification or stenosis position by calculating the position ratio. The corresponding position in the standard pattern. The system of claim 1, wherein the neural network learning module is a convolutional neural network or other type of supervised neural network. The system of claim 1, further comprising a data output module for outputting a report such as vascular calcification or stenosis degree and calcification position. The system of claim 7, wherein the information further includes blood and science information, which are also output in the report.