Coronary scoring method, apparatus, electronic device, and storage medium
By identifying target vascular regions based on CTA data and generating DSA coronary artery scores using deep learning network models, the problems of high risk in traditional CAG and lack of functional evaluation in CTA are solved. This achieves non-invasive coronary Syntax scoring, meeting the clinical need for rapid and objective coronary artery risk assessment.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- THE FIRST AFFILIATED HOSPITAL OF SOOCHOW UNIV
- Filing Date
- 2026-02-05
- Publication Date
- 2026-06-05
AI Technical Summary
Traditional coronary angiography (CAG) carries certain risks and is expensive, limiting its widespread use in coronary heart disease screening. Meanwhile, coronary CTA lacks functional evaluation, resulting in limited diagnostic accuracy when relying solely on CTA.
By acquiring CTA data and combining it with pre-screening conditions to determine the target vascular region, the system outputs coronary artery segment images, names, and geometric parameters, which are then input into a pre-trained deep learning network model to generate a patient coronary artery score array that serves as the gold standard for DSA coronary artery scoring.
This paper presents a non-invasive, automated coronary Syntax scoring method that combines the accuracy of DSA with the advantages of CTA non-invasive examination, meeting the clinical need for rapid and objective CAD risk assessment.
Smart Images
Figure CN122156086A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of medical image processing and intelligent diagnostic technology, and in particular to a coronary artery scoring method, device, electronic device, and storage medium. Background Technology
[0002] In the field of coronary artery disease (CAD) diagnosis and treatment, coronary angiography (CAG) is widely used to assess the degree of coronary artery stenosis.
[0003] However, traditional CAG is an invasive method (i.e., DSA, which is an invasive examination requiring puncture of the coronary arteries to inject a special contrast agent into the coronary ostium before testing), carrying certain risks and high costs, limiting its widespread use in clinical screening. Coronary CTA, as a non-invasive imaging technique, has gradually gained importance in the preliminary diagnosis of CAD in recent years. Although CTA can provide anatomical information about the coronary arteries, the accuracy of diagnosis relying solely on CTA remains limited due to the lack of functional evaluation. Summary of the Invention
[0004] The purpose of this invention is to provide a coronary artery scoring method, device, electronic device, and storage medium, which can at least provide a non-invasive, automated coronary Syntax scoring method that combines the accuracy of DSA with the advantages of non-invasive CTA examination, in order to meet the clinical need for rapid and objective CAD risk assessment.
[0005] To address the aforementioned technical problems, in a first aspect, the present invention provides a coronary artery scoring method, characterized in that it includes at least:
[0006] Acquire computed tomography (CTA) angiography data of any patient, and determine the target vascular region based on the CTA data and pre-screening conditions, and then output at least the coronary artery segment images, coronary artery segment names and / or coronary artery segment weights, and coronary artery segment geometric parameters within the target vascular region.
[0007] At least the coronary artery segment images, the coronary artery segment names and / or the coronary artery segment weights, and the coronary artery segment geometric parameters are input into a pre-trained deep learning network model to output a patient coronary artery score array that uses the patient's digital subtraction angiography (DSA) coronary artery score as the gold standard.
[0008] Optionally, the patient coronary score array includes at least one of the following: the patient coronary optimized score (using the patient's DSA coronary score as the gold standard), the score change value, or the score change percentage.
[0009] Optionally, the pre-training process of the deep learning network model includes at least the following steps:
[0010] Acquire pre-trained CTA data of multiple patients and their corresponding pre-trained DSA coronary artery scores for model pre-training, and determine the corresponding key vascular region based on each of the pre-trained CTA data and the pre-screening conditions, and then output at least the pre-trained coronary artery segment image, pre-trained coronary artery segment name and / or pre-trained coronary artery segment weight, and pre-trained coronary artery segment geometric parameters within each of the key vascular regions.
[0011] At least the pre-trained coronary artery segment images, the names of the pre-trained coronary artery segments and / or the weights of the pre-trained coronary artery segments, and the geometric parameters of the pre-trained coronary artery segments are input into the deep learning network model to generate coronary artery scoring results;
[0012] At least based on the coronary artery score, the pre-trained DSA coronary artery score, and the loss function of the deep learning network model, the loss change data is calculated, and it is determined whether the loss value of the loss change data within a set interval converges to 0;
[0013] When the loss value of the loss change data converges to 0 within the set interval, it is determined that the deep learning network model pre-training is complete.
[0014] Optionally, after calculating loss change data based at least on the coronary artery score, the pre-trained DSA coronary artery score, and the loss function of the deep learning network model, and determining whether the loss value of the loss change data converges to 0 within a set interval, the pre-training process of the deep learning network model further includes at least the following steps:
[0015] When the loss value of the loss change data within the set interval does not converge to 0, it is determined whether the training number of the deep learning network model has met the standard.
[0016] When the training times of the deep learning network model reach the target, the weights of the deep learning network model are updated at least according to the preset gradient descent method, and the training sample features and the coronary artery scoring results are saved as models during the training process.
[0017] Based on the same concept, in a second aspect, the present invention also provides a coronary artery scoring device for performing the coronary artery scoring method described in any one of the first aspects;
[0018] The coronary artery scoring device includes at least:
[0019] The parameter output module is used to acquire CTA data of any patient, and to determine the target vascular region based on the CTA data and pre-screening conditions, and then output at least the coronary artery segment image, coronary artery segment name and / or coronary artery segment weight, and coronary artery segment geometric parameters within the target vascular region.
[0020] The scoring output module is used to input at least the coronary artery segment image, the coronary artery segment name and / or the coronary artery segment weight, and the coronary artery segment geometric parameters into a pre-trained deep learning network model to output a patient coronary artery score array that uses the patient's DSA coronary artery score as the gold standard.
[0021] Optionally, the patient coronary score array includes at least one of the following: the patient coronary optimized score (using the patient's DSA coronary score as the gold standard), the score change value, or the score change percentage.
[0022] Optionally, the pre-training process of the deep learning network model includes at least the following steps:
[0023] Acquire pre-trained CTA data of multiple patients and their corresponding pre-trained DSA coronary artery scores for model pre-training, and determine the corresponding key vascular region based on each of the pre-trained CTA data and the pre-screening conditions, and then output at least the pre-trained coronary artery segment image, pre-trained coronary artery segment name and / or pre-trained coronary artery segment weight, and pre-trained coronary artery segment geometric parameters within each of the key vascular regions.
[0024] At least the pre-trained coronary artery segment images, the names of the pre-trained coronary artery segments and / or the weights of the pre-trained coronary artery segments, and the geometric parameters of the pre-trained coronary artery segments are input into the deep learning network model to generate coronary artery scoring results;
[0025] At least based on the coronary artery score, the pre-trained DSA coronary artery score, and the loss function of the deep learning network model, the loss change data is calculated, and it is determined whether the loss value of the loss change data within a set interval converges to 0;
[0026] When the loss value of the loss change data converges to 0 within the set interval, it is determined that the deep learning network model pre-training is complete.
[0027] Optionally, after calculating loss change data based at least on the coronary artery score, the pre-trained DSA coronary artery score, and the loss function of the deep learning network model, and determining whether the loss value of the loss change data converges to 0 within a set interval, the pre-training process of the deep learning network model further includes at least the following steps:
[0028] When the loss value of the loss change data within the set interval does not converge to 0, it is determined whether the training number of the deep learning network model has met the standard.
[0029] When the training times of the deep learning network model reach the target, the weights of the deep learning network model are updated at least according to the preset gradient descent method, and the training sample features and the coronary artery scoring results are saved as models during the training process.
[0030] Based on the same concept, in a third aspect, the present invention also provides an electronic device including a memory and a processor, the memory storing a computer program executable on the processor, the processor executing the program to implement the steps in the coronary artery scoring method of any one of the first aspects.
[0031] Based on the same concept, in a fourth aspect, the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps in the coronary artery scoring method of any one of the first aspects.
[0032] The technical solution provided by this invention first acquires CTA data from any patient, and then determines the target vascular region based on the CTA data and pre-screening conditions. It then outputs at least the coronary artery segment images, segment names and / or segment weights, and geometric parameters of the coronary artery segments within the target vascular region. Further, it inputs the coronary artery segment images, segment names and / or segment weights, and geometric parameters into a pre-trained deep learning network model to output a patient coronary artery score array that uses the patient's DSA coronary artery score as the gold standard. Therefore, this invention provides at least one non-invasive, automated coronary artery Syntax scoring method that combines the accuracy of DSA with the advantages of non-invasive CTA examination, meeting the clinical need for rapid and objective CAD risk assessment. Attached Figure Description
[0033] Figure 1 This is a flowchart of a coronary artery scoring method provided in an embodiment of the present invention;
[0034] Figure 2 This is a schematic diagram of a right-side advantage type provided by an embodiment of the present invention;
[0035] Figure 3 This is a schematic diagram of a left-side dominant type provided by an embodiment of the present invention;
[0036] Figure 4 This is a schematic diagram of the structure of a coronary artery scoring device provided in an embodiment of the present invention;
[0037] Figure 5This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation
[0038] 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. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0039] The terminology used in the embodiments of this application is for the purpose of describing particular embodiments only and is not intended to limit the application. The singular forms “a,” “said,” and “the” used in the embodiments of this application and the appended claims are also intended to include the plural forms, and “multiple” generally includes at least two unless the context clearly indicates otherwise.
[0040] It should be understood that the term "and / or" used in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.
[0041] It should be understood that although the terms first, second, third, etc., may be used in the embodiments of this application, these descriptions should not be limited to these terms. These terms are only used to distinguish the descriptions. For example, first may also be referred to as second without departing from the scope of the embodiments of this application, and similarly, second may also be referred to as first.
[0042] Depending on the context, the words “if” or “suppose” as used here can be interpreted as “when” or “in response to determination” or “in response to detection.” Similarly, depending on the context, the phrases “if determination” or “if detection (of the stated condition or event)” can be interpreted as “when determination” or “in response to determination” or “when detection (of the stated condition or event)” or “in response to detection (of the stated condition or event).”
[0043] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that an article or device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such an article or device. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the article or device that includes said element.
[0044] It should be noted that any symbols and / or numbers present in the specification that are not marked in the accompanying drawings are not reference numerals.
[0045] Figure 1 This is a flowchart of a coronary artery scoring method provided by an embodiment of the present invention. This embodiment is at least applicable to clinical coronary artery disease risk assessment scenarios. The coronary artery scoring method can be, but is not limited to, executed by the coronary artery scoring device in this embodiment of the present invention, and the executing entity can be implemented in software and / or hardware. Figure 1 As shown, this coronary artery scoring method includes at least the following steps:
[0046] S1. Obtain computed tomography (CTA) angiography data of any patient, and determine the target vascular region based on the CTA data and pre-screening conditions, and then output at least the coronary artery segment images, coronary artery segment names and / or coronary artery segment weights, and coronary artery segment geometric parameters within the target vascular region.
[0047] The pre-screening criteria can be adaptively configured based on the actual clinical diagnosis of coronary artery disease (CAD). Different coronary artery lesions correspond to different pre-screening criteria, such as a segmental stenosis rate greater than 50%. It can be understood that the target vessel area is the vessel area that is already diseased or has a potential risk of coronary artery disease.
[0048] It is known that coronary artery segment naming or weighting is related to the coronary artery dominance type, which can be divided into left-sided dominance and right-sided dominance. Specifically, Figure 2 This is a schematic diagram of a right-side advantage type provided in an embodiment of the present invention. See also: Figure 2 The posterior descending artery originates from the right coronary artery (segment 4), meaning the posterior interventricular groove is entirely supplied by the right coronary artery or by a combination of the right coronary artery and the circumflex artery. Furthermore, Figure 3 This is a schematic diagram of a left-side dominant type provided by an embodiment of the present invention, such as... Figure 3 As shown, if the posterior descending artery originates from the left coronary artery (segment 15), meaning the posterior interventricular groove is entirely supplied by the circumflex artery, there is no balanced coronary artery selection in the Syntax score.
[0049] In one specific implementation, the coronary segment weight corresponding to the proximal RCA (i.e., half the distance from the right coronary ostium to the acute rim of the heart) in the right-dominant type can be 1; the coronary segment weight corresponding to the mid-RCA (i.e., from the end of Seg1 to the acute rim of the heart) in the right-dominant type can be 1; the coronary segment weight corresponding to the distal RCA (i.e., from the acute rim of the heart to the ostium of the posterior descending artery) in the right-dominant type can be 1; the coronary segment weight corresponding to the posterior descending artery (i.e., originating from the right coronary artery and running along the posterior interventricular groove) in the right-dominant type can be 1; the posterior branch PL (posterior branch of the right coronary artery: from Seg4 to the posterior crux, running along the left) in the right-dominant type can be 1. The weight of the coronary artery segment corresponding to the atrioventricular groove or left ventricular diaphragmatic surface can be 0.5; in the right-dominant type, the weight of the coronary artery segment corresponding to the posterior branch PL (the first branch of Seg16) from the RCA can be 0.5; in the right-dominant type, the weight of the coronary artery segment corresponding to the posterior branch PL (the second branch of Seg16) from the RCA can be 0.5; in the right-dominant type, the weight of the coronary artery segment corresponding to the posterior branch PL (the third branch of Seg16) from the RCA can be 0.5; in the right-dominant type, the weight of the coronary artery segment corresponding to the left main coronary artery (from the left coronary ostium to the bifurcation of the anterior descending and circumflex arteries) can be 5; in the right-dominant type, the weight of the proximal segment of the LAD (i.e., from the main... The coronary segment weight corresponding to the bifurcation point of the main interventricular septal branch (S1) can be 3.5; in the right-dominant type, the coronary segment weight corresponding to the mid-LAD segment (i.e., from the distal end of Seg6 to the point where the LAD forms an angle in the RAO view; if this angle is not identifiable, then the segment is located at half the distance from the first large interventricular septal branch S1 to the apex) can be 2.5; in the right-dominant type, the coronary segment weight corresponding to the distal LAD segment (i.e., the distal portion of the LAD, from the distal end of Seg7 to the apex) can be 1; in the right-dominant type, the coronary segment weight corresponding to the first diagonal branch (i.e., the first diagonal branch originating from Seg6 or Seg7) can be 1; right In the lateral dominance type, the coronary segment weight corresponding to the first diagonal branch a (i.e., the small first diagonal branch originating from seg6 or seg7 before seg8) can be 1; in the right-dominant type, the coronary segment weight corresponding to the second diagonal branch (i.e., the second diagonal branch originating from seg8, or the transition between seg7 and seg8) can be 0.5; in the right-dominant type, the coronary segment weight corresponding to the second diagonal branch a (i.e., the small second diagonal branch originating from seg8, or the transition between seg7 and seg8) can be 0.5; in the right-dominant type, the coronary segment weight corresponding to the proximal segment of the circumflex branch (i.e., from the bifurcation of the main coronary artery to and including the opening of the first obtuse marginal branch OM1) can be 1.5; In the right-dominant type, the coronary segment weight corresponding to the anterior or intermediate branch (i.e., the intermediate branch RI at the bifurcation of the LM main trunk, belonging to the obtuse marginal branch region) can be 1; In the right-dominant type, the coronary segment weight corresponding to the obtuse marginal branch a (i.e., the first branch of the LCX, generally running to the obtuse marginal region of the heart) can be 1; In the right-dominant type, the coronary segment weight corresponding to the obtuse marginal branch b (i.e., the second branch of the LCX in the same direction as SEG12) can be 1; In the right-dominant type, the distal segment of the circumflex branch (i.e., from the origin of the last obtuse marginal branch of the LCX to the end of the LCX, generally running along the left posterior atrioventricular groove; possibly...) The coronary artery segment weight corresponding to a segment that is very small or missing can be 0.5; in the right-dominant type, the coronary artery segment weight corresponding to the left ventricular posterior branch (i.e., the left ventricular posterior branch originating from seg13 and running along the posterolateral wall of the left ventricle) can be 0.5; in the right-dominant type, the coronary artery segment weight corresponding to left ventricular posterior branch a (i.e., the first small left ventricular posterior branch originating after seg14 and running in the same direction as seg14) can be 0.5; in the right-dominant type, the coronary artery segment weight corresponding to left ventricular posterior branch b (i.e., the second small left ventricular posterior branch originating after seg14 and 14a and running in the same direction as seg14) can be 0.5.
[0050] In another specific implementation, the coronary segment weight corresponding to the proximal RCA (i.e., half the distance from the right coronary ostium to the acute rim of the heart) in the left-dominant type can be 0; the coronary segment weight corresponding to the mid-RCA (i.e., from the end of Seg1 to the acute rim of the heart) in the left-dominant type can be 0; the coronary segment weight corresponding to the distal RCA (i.e., from the acute rim of the heart to the ostium of the posterior descending artery) in the left-dominant type can be 0; the coronary segment weight corresponding to the left main LM (from the left coronary ostium to the bifurcation of the anterior descending and circumflex arteries) in the left-dominant type can be 6; the coronary segment weight corresponding to the proximal LAD (i.e., from the bifurcation of the main artery to and including the first great ventricular septal branch) in the left-dominant type can be 6. The weight of the coronary artery segment corresponding to the origin of S1 can be 3.5; in the left-dominant type, the weight of the coronary artery segment corresponding to the mid-LAD (i.e., from the distal end of Seg6 to the point where the LAD forms an angle in the RAO view; if this angle is not identifiable, then the segment is located at half the distance from the first large ventricular septal branch S1 to the apex) can be 2.5; in the left-dominant type, the weight of the coronary artery segment corresponding to the distal LAD (i.e., the distal part of the LAD, from the distal end of Seg7 to the apex) can be 1; in the left-dominant type, the weight of the coronary artery segment corresponding to the first diagonal branch (i.e., the first diagonal branch originating from Seg6 or Seg7) can be 1; in the left-dominant type, the weight of the first diagonal branch can be 1; The coronary segment weight corresponding to the diagonal branch a (i.e., the small first diagonal branch originating from seg6 or seg7 before seg8) can be 1; the coronary segment weight corresponding to the second diagonal branch (i.e., the second diagonal branch originating from seg8, or the transition between seg7 and seg8) in the left-dominant type can be 0.5; the coronary segment weight corresponding to the second diagonal branch a (i.e., the small second diagonal branch originating from seg8, or the transition between seg7 and seg8) in the left-dominant type can be 0.5; the coronary segment weight corresponding to the proximal segment of the circumflex branch (i.e., from the bifurcation of the main coronary artery to and including the opening of the first obtuse marginal branch OM1) in the left-dominant type can be 2. 5; In the left-dominant type, the coronary segment weight corresponding to the anterior or intermediate branch (i.e., the intermediate branch RI at the bifurcation of the LM main trunk, belonging to the obtuse marginal branch region) can be 1; In the left-dominant type, the coronary segment weight corresponding to the obtuse marginal branch a (i.e., the first branch of the LCX, generally running to the obtuse marginal region of the heart) can be 1; In the left-dominant type, the coronary segment weight corresponding to the obtuse marginal branch b (i.e., the second branch of the LCX in the same direction as seg12) can be 1; In the left-dominant type, the coronary segment weight corresponding to the distal segment of the circumflex branch (i.e., from the origin of the last obtuse marginal branch of the LCX to the end of the LCX, generally running along the left posterior atrioventricular groove; may be very small or absent) can be 1.5. In the left-dominant type, the coronary segment weight corresponding to the left ventricular posterior branch (i.e., the left ventricular posterior branch originating from seg13 and running along the posterolateral wall of the left ventricle) can be 1; in the left-dominant type, the coronary segment weight corresponding to left ventricular posterior branch a (i.e., the first small left ventricular posterior branch originating after seg14 and running in the same direction as seg14) can be 1; in the left-dominant type, the coronary segment weight corresponding to left ventricular posterior branch b (i.e., the second small left ventricular posterior branch originating after seg14 and 14a and running in the same direction as seg14) can be 1; in the left-dominant type, the coronary segment weight corresponding to the posterior descending artery (i.e., the posterior descending artery originating from the LCX, generally present in the left-dominant type; when seg15 is present, seg4 is generally absent) can be 1.
[0051] S2. Input at least the coronary segment images, coronary segment names and / or coronary segment weights, and coronary segment geometric parameters into the pre-trained deep learning network model to output a patient coronary score array that uses the patient's digital subtraction angiography (DSA) coronary score as the gold standard.
[0052] There are several options for choosing a deep learning network model, such as a convolutional neural network model.
[0053] In another specific implementation, the patient coronary score array may optionally include at least one of the following: a patient coronary optimization score (which should at least be close to or equal to the DSA coronary score) using the patient's DSA coronary score as the gold standard, a score change value (which may refer to the score difference between the patient's coronary optimization score and the DSA coronary score), or a score change percentage (which may refer to the ratio of the aforementioned score difference to the DSA coronary score, or the ratio of the aforementioned score difference to the patient's coronary optimization score, etc.).
[0054] In yet another specific implementation, the pre-training process of the deep learning network model may optionally include at least the following steps:
[0055] (1) Obtain pre-trained CTA data of multiple patients and their corresponding pre-trained DSA coronary artery scores for model pre-training, and determine the corresponding key vascular region based on each pre-trained CTA data and the pre-screening conditions, and then output at least the pre-trained coronary artery segment image, pre-trained coronary artery segment name and / or pre-trained coronary artery segment weight, and pre-trained coronary artery segment geometric parameters in each key vascular region.
[0056] (2) At least the pre-trained coronary artery segment images, pre-trained coronary artery segment names and / or pre-trained coronary artery segment weights, and pre-trained coronary artery segment geometric parameters are input into the deep learning network model to generate coronary artery scoring results;
[0057] (3) Calculate the loss change data (loss change data can refer to the loss value generated by each round of training of the model or the loss value change data obtained by statistical methods) based at least on the coronary artery score, the pre-trained DSA coronary artery score and the loss function of the deep learning network model (the loss function can be, for example, the cross-entropy loss function), and determine whether the loss value of the loss change data converges to 0 within the set interval.
[0058] (4) When the loss value of the loss change data within the set interval converges to 0 (that is, when the loss change data is within the set interval, the magnitude and trend of all loss values converge to 0), the deep learning network model pre-training is determined to be complete.
[0059] In yet another specific implementation, optionally, after the aforementioned step (3), the pre-training process of the deep learning network model further includes at least the following steps:
[0060] (5) When the loss value of the loss change data within the set interval does not converge to 0 (i.e., the loss change data is within the set interval, but not all the magnitude and trend of the loss value converge to 0), determine whether the training times of the deep learning network model have met the standard.
[0061] (6) When the number of training iterations of the deep learning network model reaches the target, the weights of the deep learning network model should be updated at least according to the preset gradient descent method (e.g., mini-batch gradient descent method), and the training sample features and coronary artery scoring results during the training process should be saved. (Of course, other pre-trained coronary artery segment images, pre-trained coronary artery segment names and / or pre-trained coronary artery segment weights, and pre-trained coronary artery segment geometric parameters can still be input into the deep learning network model for training.)
[0062] In addition, when the number of training iterations of the deep learning network model is insufficient, the weights of the deep learning network model should be updated at least according to the set gradient descent method, and the pre-trained coronary artery segment images, pre-trained coronary artery segment names and / or pre-trained coronary artery segment weights, and pre-trained coronary artery segment geometric parameters should be re-input into the deep learning network model for training.
[0063] The technical solution provided in this embodiment first acquires CTA data from any patient. Based on the CTA data and pre-screening conditions, the target vessel region is determined. Then, at least the coronary artery segment images, segment names, and / or segment weights, as well as geometric parameters of the coronary artery segments within the target vessel region, are output. Further, the coronary artery segment images, segment names, and / or segment weights, as well as geometric parameters, are input into a pre-trained deep learning network model to output a patient coronary artery score array that uses the patient's DSA coronary artery score as the gold standard. Therefore, this embodiment provides at least one non-invasive, automated coronary artery Syntax scoring method that combines the accuracy of DSA with the advantages of non-invasive CTA examination, meeting the clinical need for rapid and objective CAD risk assessment.
[0064] Figure 4 This is a schematic diagram of a coronary artery scoring device provided in an embodiment of the present invention. This embodiment is at least applicable to clinical coronary artery disease risk assessment scenarios. The coronary artery scoring device can be implemented using software and / or hardware. Figure 4 As shown, the coronary artery scoring device is used to perform the coronary artery scoring method of any of the foregoing embodiments or implementations.
[0065] Coronary scoring devices should include at least:
[0066] The parameter output module 110 is used to acquire CTA data of any patient, and to determine the target vascular region based on the CTA data and pre-screening conditions, and then output at least the coronary artery segment image, coronary artery segment name and / or coronary artery segment weight, and coronary artery segment geometric parameters within the target vascular region.
[0067] The scoring output module 120 is used to input at least coronary segment images, coronary segment names and / or coronary segment weights, and coronary segment geometric parameters into a pre-trained deep learning network model to output an array of patient coronary scores that use the patient's DSA coronary score as the gold standard.
[0068] Optionally, the patient coronary score array includes at least one of the following: the patient coronary optimization score (using the patient's DSA coronary score as the gold standard), the score change value, or the score change percentage.
[0069] Optionally, the pre-training process of a deep learning network model includes at least the following steps:
[0070] Acquire pre-trained CTA data from multiple patients and their corresponding pre-trained DSA coronary artery scores for model pre-training. Based on each pre-trained CTA data, determine the corresponding key vascular region by combining pre-screening conditions, and then output at least the pre-trained coronary artery segment image, pre-trained coronary artery segment name and / or pre-trained coronary artery segment weight, and pre-trained coronary artery segment geometric parameters within each key vascular region.
[0071] At least the pre-trained coronary artery segment images, pre-trained coronary artery segment names and / or pre-trained coronary artery segment weights, and pre-trained coronary artery segment geometric parameters are input into the deep learning network model to generate coronary artery scoring results;
[0072] At least the loss change data should be calculated based on the coronary artery score, the pre-trained DSA coronary artery score, and the loss function of the deep learning network model, and it should be determined whether the loss value of the loss change data converges to 0 within a set interval.
[0073] When the loss value of the loss change data converges to 0 within a set interval, the deep learning network model is considered to have completed pre-training.
[0074] Optionally, after calculating the loss change data based at least on the coronary artery score, the pre-trained DSA coronary artery score, and the loss function of the deep learning network model, and determining whether the loss value of the loss change data converges to 0 within a set interval, the pre-training process of the deep learning network model may further include at least the following steps:
[0075] When the loss value of the loss change data does not converge to 0 within a set interval, determine whether the number of training times of the deep learning network model has met the standard.
[0076] When the training iterations of the deep learning network model reach the target, the weights of the deep learning network model should be updated at least according to the preset gradient descent method, and the training sample features and coronary artery scoring results during the training process should be saved.
[0077] The technical solution provided in this embodiment first acquires CTA data of any patient through a parameter output module. Based on the CTA data and pre-screening conditions, the target vessel region is determined. Then, at least the coronary artery segment images, segment names and / or segment weights, and geometric parameters of the coronary artery segments within the target vessel region are output. Further, the scoring output module inputs at least the coronary artery segment images, segment names and / or segment weights, and geometric parameters into a pre-trained deep learning network model to output a patient coronary artery score array that uses the patient's DSA coronary artery score as the gold standard. Therefore, this embodiment provides at least one non-invasive, automated coronary artery Syntax scoring method that combines the accuracy of DSA with the advantages of non-invasive CTA examination, meeting the clinical need for rapid and objective CAD risk assessment.
[0078] This embodiment provides an electronic device. Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. See also: Figure 5 The electronic device 1000 includes a processor 1001 and a memory 1002. The memory 1002 stores computer-readable instructions. When the computer-readable instructions are executed by the processor 1001, the steps in any of the coronary artery scoring methods described above are performed. Through the above technical solution, the processor 1001 and the memory 1002 are interconnected and communicate with each other through a communication bus and / or other forms of connection mechanism (not shown). The memory 1002 stores a computer program that can be executed by the processor. When the electronic device 1000 is running, the processor 1001 executes the computer program to execute the coronary artery scoring method in any optional implementation of the above embodiments, so as to at least achieve the following functions: acquire CTA data of any patient, determine the target vascular region based on the CTA data and combined with pre-screening conditions, and then output at least the coronary artery segment image, coronary artery segment naming and / or coronary artery segment weight, and coronary artery segment geometric parameters within the target vascular region; at least input the coronary artery segment image, coronary artery segment naming and / or coronary artery segment weight, and coronary artery segment geometric parameters into a pre-trained deep learning network model to output a patient coronary artery score array that uses the patient's DSA coronary artery score as the gold standard.
[0079] This embodiment provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the coronary artery scoring method provided in all embodiments of this application: acquiring CTA data of any patient, determining a target vascular region based on the CTA data and pre-screening conditions, and then outputting at least the coronary artery segment images, coronary artery segment names and / or coronary artery segment weights, and coronary artery segment geometric parameters within the target vascular region; and at least inputting the coronary artery segment images, coronary artery segment names and / or coronary artery segment weights, and coronary artery segment geometric parameters into a pre-trained deep learning network model to output a patient coronary artery score array that uses the patient's DSA coronary artery score as the gold standard.
[0080] Any combination of one or more computer-readable media may be used. A computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. A computer-readable storage medium can be, for example—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in connection with an instruction execution system, apparatus, or device.
[0081] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including—but not limited to—electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, capable of transmitting, propagating, or transmitting programs for use by or in connection with an instruction execution system, apparatus, or device.
[0082] The program code contained on a computer-readable medium may be transmitted using any suitable medium, including—but not limited to—wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.
[0083] Computer program code for performing the operations of this invention can be written in one or more programming languages or a combination thereof. Programming languages include object-oriented programming languages—such as Java, Smalltalk, and C++—as well as conventional procedural programming languages—such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).
[0084] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.
Claims
1. A coronary artery scoring method, characterized in that, At least including: Acquire computed tomography (CTA) angiography data of any patient, and determine the target vascular region based on the CTA data and pre-screening conditions, and then output at least the coronary artery segment images, coronary artery segment names and / or coronary artery segment weights, and coronary artery segment geometric parameters within the target vascular region. At least the coronary artery segment images, the coronary artery segment names and / or the coronary artery segment weights, and the coronary artery segment geometric parameters are input into a pre-trained deep learning network model to output a patient coronary artery score array that uses the patient's digital subtraction angiography (DSA) coronary artery score as the gold standard.
2. The coronary artery scoring method according to claim 1, characterized in that, The patient coronary artery score array includes at least one of the following: the patient coronary artery optimization score (using the patient's DSA coronary artery score as the gold standard), the score change value, or the score change percentage.
3. The coronary artery scoring method according to claim 1, characterized in that, The pre-training process of the deep learning network model includes at least the following steps: Acquire pre-trained CTA data of multiple patients and their corresponding pre-trained DSA coronary artery scores for model pre-training, and determine the corresponding key vascular region based on each of the pre-trained CTA data and the pre-screening conditions, and then output at least the pre-trained coronary artery segment image, pre-trained coronary artery segment name and / or pre-trained coronary artery segment weight, and pre-trained coronary artery segment geometric parameters within each of the key vascular regions. At least the pre-trained coronary artery segment images, the names of the pre-trained coronary artery segments and / or the weights of the pre-trained coronary artery segments, and the geometric parameters of the pre-trained coronary artery segments are input into the deep learning network model to generate coronary artery scoring results; At least based on the coronary artery score, the pre-trained DSA coronary artery score, and the loss function of the deep learning network model, the loss change data is calculated, and it is determined whether the loss value of the loss change data within a set interval converges to 0; When the loss value of the loss change data converges to 0 within the set interval, it is determined that the deep learning network model pre-training is complete.
4. The coronary artery scoring method according to claim 1, characterized in that, After calculating the loss change data based at least on the coronary artery score, the pre-trained DSA coronary artery score, and the loss function of the deep learning network model, and determining whether the loss value of the loss change data converges to 0 within a set interval, the pre-training process of the deep learning network model further includes at least the following steps: When the loss value of the loss change data within the set interval does not converge to 0, it is determined whether the training number of the deep learning network model has met the standard. When the training times of the deep learning network model reach the target, the weights of the deep learning network model are updated at least according to the preset gradient descent method, and the training sample features and the coronary artery scoring results are saved as models during the training process.
5. A coronary artery scoring device, characterized in that, Used to perform the coronary artery scoring method according to any one of claims 1-4; The coronary artery scoring device includes at least: The parameter output module is used to acquire CTA data of any patient, and to determine the target vascular region based on the CTA data and pre-screening conditions, and then output at least the coronary artery segment image, coronary artery segment name and / or coronary artery segment weight, and coronary artery segment geometric parameters within the target vascular region. The scoring output module is used to input at least the coronary artery segment image, the coronary artery segment name and / or the coronary artery segment weight, and the coronary artery segment geometric parameters into a pre-trained deep learning network model to output a patient coronary artery score array that uses the patient's DSA coronary artery score as the gold standard.
6. The coronary artery scoring device according to claim 5, characterized in that, The patient coronary artery score array includes at least one of the following: the patient coronary artery optimization score (using the patient's DSA coronary artery score as the gold standard), the score change value, or the score change percentage.
7. The coronary artery scoring device according to claim 5, characterized in that, The pre-training process of the deep learning network model includes at least the following steps: Acquire pre-trained CTA data of multiple patients and their corresponding pre-trained DSA coronary artery scores for model pre-training, and determine the corresponding key vascular region based on each of the pre-trained CTA data and the pre-screening conditions, and then output at least the pre-trained coronary artery segment image, pre-trained coronary artery segment name and / or pre-trained coronary artery segment weight, and pre-trained coronary artery segment geometric parameters within each of the key vascular regions. At least the pre-trained coronary artery segment images, the names of the pre-trained coronary artery segments and / or the weights of the pre-trained coronary artery segments, and the geometric parameters of the pre-trained coronary artery segments are input into the deep learning network model to generate coronary artery scoring results; At least based on the coronary artery score, the pre-trained DSA coronary artery score, and the loss function of the deep learning network model, the loss change data is calculated, and it is determined whether the loss value of the loss change data within a set interval converges to 0; When the loss value of the loss change data converges to 0 within the set interval, it is determined that the deep learning network model pre-training is complete.
8. The coronary artery scoring device according to claim 5, characterized in that, After calculating the loss change data based at least on the coronary artery score, the pre-trained DSA coronary artery score, and the loss function of the deep learning network model, and determining whether the loss value of the loss change data converges to 0 within a set interval, the pre-training process of the deep learning network model further includes at least the following steps: When the loss value of the loss change data within the set interval does not converge to 0, it is determined whether the training number of the deep learning network model has met the standard. When the training times of the deep learning network model reach the target, the weights of the deep learning network model are updated at least according to the preset gradient descent method, and the training sample features and the coronary artery scoring results are saved as models during the training process.
9. An electronic device comprising a memory and a processor, the memory storing a computer program executable on the processor, characterized in that, When the processor executes the program, it implements the steps in the coronary artery scoring method according to any one of claims 1 to 4.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps in the coronary artery scoring method according to any one of claims 1 to 4.