Coronary artery center line segmentation naming method and device based on angiography image segmentation

Abstract

The invention provides a coronary artery center line segmentation naming method and device based on angiography image segmentation, and relates to the technical field of medical image processing, and the method comprises the steps: obtaining an angiography image of a coronary artery blood vessel; inputting the angiography image into a pre-trained joint segmentation segmentation model, and outputting a blood vessel segmentation result and a segmentation naming result; based on the blood vessel segmentation result and the segmentation naming result, generating an ordered center line of the coronary blood vessel; wherein the ordered center line is a center line of which each segment in the skeleton diagram of the coronary blood vessel has a clear starting point, a clear ending point and a clear direction; and based on the ordered centerline and the segmented naming result, generating a coronary artery centerline segmented naming result of the coronary artery blood vessel according with an anatomical rule. According to the method provided by the invention, double tasks are synchronously completed through the joint segmentation segmentation model, error accumulation is avoided, and the overall analysis efficiency and precision are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of medical image processing, and particularly to a method and device for segmenting and naming coronary artery centerlines based on contrast image segmentation. Background Art

[0002] In clinical practice, doctors usually need to anatomically segment the main branches of the coronary artery (such as the left anterior descending artery, circumflex artery, right coronary artery, etc.) and their secondary branches according to angiographic images, and accurately name them according to standardized naming specifications. This process is not only directly related to the accurate positioning of the lesion location, but also an important basis for formulating interventional treatment plans.

[0003] In the prior art, coronary artery vessel analysis usually adopts a step-by-step processing method: first, the vessel contour is extracted through an image segmentation algorithm, and then skeleton extraction and branch naming are performed based on the segmentation result. This method has the following typical problems: the segmentation model and the naming model are independently trained, resulting in the gradual transmission of errors and limited overall accuracy; the loss weights of the segmentation and segmentation tasks are fixed during the training process and cannot adapt to the requirements of different training stages; there is a lack of anatomical rule constraints after skeleton extraction, and noise branches and topological errors affect the generation of the centerline; the integration of segmentation naming and the centerline mostly relies on manual correction, with low efficiency and easy introduction of subjective biases.

[0004] How to synchronously complete the dual tasks of segmentation and segmentation, avoid error accumulation, and improve the overall analysis efficiency and accuracy is a technical problem that needs to be solved currently. Summary of the Invention

[0005] The present invention provides a method and device for segmenting and naming coronary artery centerlines based on contrast image segmentation to solve the defects existing in the prior art.

[0006] The present invention provides a method for segmenting and naming coronary artery centerlines based on contrast image segmentation, including the following steps: Obtain an angiographic image of a coronary artery vessel; wherein, the angiographic image includes: image data of at least one target shooting position; Input the angiographic image into a pre-trained joint segmentation and segmentation model, and output a vessel segmentation result and a segmentation naming result; wherein, the joint segmentation and segmentation model is: obtained by training a multi-task deep learning model based on a coronary artery vessel angiographic image dataset; Generate an ordered centerline of the coronary artery vessel based on the vessel segmentation result and the segmentation naming result; wherein, the ordered centerline is: a centerline in the skeleton diagram of the coronary artery vessel where each segment has a clear starting point, ending point, and direction; Generate the sectional naming result of the coronary artery centerline of the coronary artery that conforms to anatomical rules based on the ordered centerline and the sectional naming result.

[0007] According to a method for sectional naming of coronary artery centerlines based on angiographic image segmentation provided by the present invention, the combined segmentation and sectional naming model includes: an image encoding layer, a segmented image decoding layer, and a sectional image decoding layer. The image encoding layer, the segmented image decoding layer, and the sectional image decoding layer constitute a coronary artery segmentation task and a coronary artery sectional naming task; The step of inputting the angiographic image into a pre-trained combined segmentation and sectional naming model to output a vascular segmentation result and a sectional naming result includes: Input the angiographic image into the combined segmentation and sectional naming model, and respectively execute the coronary artery segmentation task and the coronary artery sectional naming task to output the vascular segmentation result and the sectional naming result; Among them, the coronary artery segmentation task: is used to extract the filled contour of the coronary artery, and assign each pixel in the angiographic image to a single category. The coronary artery sectional naming task: is used to anatomically segment the main branches and their secondary branches of the coronary artery on the basis of coronary artery segmentation, and assign the pixels of different segments to different categories.

[0008] According to a method for sectional naming of coronary artery centerlines based on angiographic image segmentation provided by the present invention, the training process of the combined segmentation and sectional naming model includes: Train a deep learning model based on a pre-constructed multi-task learning framework so that the deep learning model can learn multiple tasks simultaneously; among them, each task in the multiple tasks has a corresponding loss function, and the weighted sum of the loss functions of each task in the multiple tasks constitutes a total loss function; the total loss function adjusts the relevance between the segmentation loss and the sectional loss through a consistency index, and the consistency index is calculated based on the number of overlapping region pixels of the segmentation error map and the sectional error map; Dynamically adjust the weights of the deep learning model by minimizing the total loss function to obtain the combined segmentation and sectional naming model.

[0009] According to a method for sectional naming of coronary artery centerlines based on angiographic image segmentation provided by the present invention, the step of dynamically adjusting the weights of the deep learning model by minimizing the total loss function includes: In the early stage of training, the segmentation loss weight is higher than the sectional loss weight; In the middle stage of training, the sectional loss weight is higher than the segmentation loss weight; In the late stage of training, the segmentation loss weight is balanced with the sectional loss weight.

[0010] A method for segment naming of coronary artery centerlines based on angiographic image segmentation provided by the present invention. Based on the blood vessel segmentation result and the segment naming result, generating an ordered centerline of the coronary artery blood vessels includes: Extracting the initial skeleton map of the coronary artery blood vessels from the blood vessel segmentation result, and filtering out the noise branches in the initial skeleton map according to a preset branch pixel length threshold to obtain the skeleton map of the coronary artery blood vessels; Determining target segments in the skeleton map of the coronary artery blood vessels based on bifurcation points and end points according to the anatomical rules of the 16-segment method of the coronary artery tree; Determining the starting point of the coronary artery blood vessels according to the segment naming result; Generating an ordered centerline of the coronary artery blood vessels based on the skeleton map of the coronary artery blood vessels, the target segments, and the starting point of the coronary artery blood vessels.

[0011] A method for segment naming of coronary artery centerlines based on angiographic image segmentation provided by the present invention. Based on the ordered centerline and the segment naming result, generating a segment naming result of the coronary artery centerlines of the coronary artery blood vessels includes: Sequentially assigning segment category values to the coordinate points on the ordered centerline according to the ordered centerline and the segment naming result to determine the category of each coordinate point; Based on the determined category of each coordinate point, respectively determining each segment of the ordered centerline through a dynamic category correction strategy, and outputting a segment naming result of the coronary artery centerlines of the coronary artery blood vessels that conforms to anatomical rules.

[0012] A method for segment naming of coronary artery centerlines based on angiographic image segmentation provided by the present invention. Based on the determined category of each coordinate point, respectively determining each segment of the ordered centerline through a dynamic category correction strategy, and outputting a segment naming result of the coronary artery centerlines of the coronary artery blood vessels that conforms to anatomical rules includes: In the case where there are multiple segment naming results in a segment of the ordered centerline, if the pixel ratio of any category in the ordered centerline is lower than a preset threshold and the distribution is discontinuous, then determine that the category is a noise point and filter it out; In the case where there are multiple segment naming results in a segment of the ordered centerline, if the pixel ratio of any category in the ordered centerline is higher than a preset threshold and the distribution is continuous, then retain the category and perform mapping matching with the segment naming result, and output a segment naming result of the coronary artery centerlines of the coronary artery blood vessels that conforms to anatomical rules.

[0013] The present invention also provides a device for segment naming of coronary artery centerlines based on angiographic image segmentation, including the following modules: An acquisition module, configured to acquire angiographic images of coronary blood vessels; wherein, the angiographic images include: image data of at least one target shooting position; A segmentation and naming module, configured to input the angiographic images into a pre-trained joint segmentation and segmentation model, and output a blood vessel segmentation result and a segmentation naming result; wherein, the joint segmentation and segmentation model is: obtained by training a multi-task deep learning model based on a coronary angiographic image dataset; A centerline extraction module, configured to generate an ordered centerline of the coronary blood vessels based on the blood vessel segmentation result and the segmentation naming result; wherein, the ordered centerline is: a centerline in the skeleton diagram of the coronary blood vessels where each segment has a clear starting point, ending point, and direction; A centerline and segmentation naming fusion module, configured to generate a coronary centerline segmentation naming result of the coronary blood vessels that conforms to anatomical rules based on the ordered centerline and the segmentation naming result.

[0014] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, it implements the coronary centerline segmentation and naming method based on angiographic image segmentation as described in any one of the above.

[0015] The present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the coronary centerline segmentation and naming method based on angiographic image segmentation as described in any one of the above.

[0016] The present invention also provides a computer program product, including a computer program. When the computer program is executed by a processor, it implements the coronary centerline segmentation and naming method based on angiographic image segmentation as described in any one of the above.

[0017] A method and device for segment naming of coronary artery centerlines based on angiographic image segmentation provided by the present invention obtain angiographic images of coronary arteries; wherein, the angiographic images include: image data of at least one target shooting position; input the angiographic images into a pre-trained joint segmentation and segmentation model to output a blood vessel segmentation result and a segmentation naming result; wherein, the joint segmentation and segmentation model is: obtained by training a multi-task deep learning model based on a coronary artery angiographic image dataset; based on the blood vessel segmentation result and the segmentation naming result, generate an ordered centerline of the coronary artery; wherein, the ordered centerline is: a centerline in the skeleton diagram of the coronary artery where each segment has a clear starting point, ending point, and direction; based on the ordered centerline and the segmentation naming result, generate a segmentation naming result of the coronary artery centerline of the coronary artery that conforms to anatomical rules. It can be seen that the present invention synchronously completes dual tasks through a joint segmentation and segmentation model, avoiding error accumulation and improving the overall analysis efficiency and accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following will briefly introduce the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0019] Figure 1 It is a schematic diagram of an application scenario of the method for segment naming of coronary artery centerlines based on angiographic image segmentation provided by the present invention.

[0020] Figure 2 It is a schematic flowchart of the method for segment naming of coronary artery centerlines based on angiographic image segmentation provided by the present invention.

[0021] Figure 3 It is a schematic structural diagram of the device for segment naming of coronary artery centerlines based on angiographic image segmentation provided by the present invention.

[0022] Figure 4 It is a schematic structural diagram of an electronic device provided by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0023] In order to make the objectives, technical solutions, and advantages of the present invention clearer, the following will clearly and completely describe the technical solutions in the present invention in conjunction with the drawings in the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the protection scope of the present invention.

[0024] The following will combine with Figures 1 - 4 to describe a method and device for segment naming of coronary artery centerlines based on angiographic image segmentation according to the present invention.

[0025] It should be noted that it takes a long time for manual completion of a single complete coronary artery segmentation naming, and the reading consistency among different physicians is poor. At the same time, complex vascular variations and lesion characteristics are prone to misjudgment. The segmentation model and the naming model are independently trained, resulting in the gradual transmission of errors and limited overall accuracy. The model performs unstably in complex scenarios (such as bifurcations and stenoses); after skeleton extraction, there is a lack of anatomical rule constraints, and noise branches and topological errors affect the generation of the centerline; the integration of segmentation naming and the centerline lacks a dynamic correction mechanism, and algorithm noise is likely to affect the interpretability of the final result. Based on this, the present invention provides a method for segment naming of coronary artery centerlines based on angiographic image segmentation to solve at least one of the above problems.

[0026] Figure 1 is a schematic diagram of the application scenario of the method for segment naming of coronary artery centerlines based on angiographic image segmentation provided by the present invention. As Figure 1 shown, the system consists of an angiography device 11 and a processing electronic device 12. During the operation, first, a catheter is inserted from the femoral artery of the patient's thigh or other peripheral arteries and guided to the ascending aorta, and then inserted into the left or right coronary artery ostium. After injecting the contrast agent, the angiography device 11 takes the developed angiographic images and transmits the image data to the electronic device 12. The electronic device 12 performs segmentation and processing on the angiographic images to generate a clear cardiovascular segmentation map, providing a basis for subsequent analysis.

[0027] The specific implementation form of the electronic device 12 is diverse. It can be an inspection terminal equipped with a medical X-ray inspection device, a doctor's office computer, a hospital information management system (HIS) management server of the hospital, or a cloud server. The specific selection can be flexibly adjusted according to the actual application scenario.

[0028] In the clinical scenario, when a patient has vascular-related discomfort symptoms, static vascular examinations (such as CT and MR) and dynamic examinations (such as DSA) are usually required. After the examinations are completed, the system scores the lesion area based on the image data, providing scientific and reliable data references for doctors' diagnosis and treatment.

[0029] Figure 2 is a schematic flowchart of the method for segment naming of coronary artery centerlines based on angiographic image segmentation provided by the present invention. As Figure 2 shown, the method includes the following: Step 100: Obtain angiographic images of the coronary artery vessels; wherein, the angiographic images include: image data of at least one target shooting position.

[0030] Specifically, the angiography image contains at least the image data of one available shooting position, and the available shooting positions include, but are not limited to, the left coronary left shoulder position, the left coronary foot position, the right coronary left anterior oblique position, and the right coronary cranial position. These positions can ensure the integrity of some vascular structures. It should be noted that the image frame filled with contrast agent is selected as the effective image. In terms of image size selection, the target image size can be set to a multiple of 2 with the same length and width, such as 512×512 pixels. The specific size can be adjusted according to the actual application scenario and is not limited to this.

[0031] Step 200: Input the angiography image into the pre-trained joint segmentation and segmentation naming model, and output the vascular segmentation result and the segmentation naming result; wherein, the joint segmentation and segmentation naming model is obtained by training a multi-task deep learning model based on a coronary angiography image dataset.

[0032] It should be noted that the joint segmentation and segmentation naming model includes an image encoding layer, a segmentation image decoding layer, and a segmentation image decoding layer. The image encoding layer, the segmentation image decoding layer, and the segmentation image decoding layer constitute the coronary artery segmentation task and the coronary artery segmentation naming task.

[0033] Step 200 inputs the angiography image into the pre-trained joint segmentation and segmentation naming model and outputs the vascular segmentation result and the segmentation naming result, including: Input the angiography image into the joint segmentation and segmentation naming model, respectively execute the coronary artery segmentation task and the coronary artery segmentation naming task, and output the vascular segmentation result and the segmentation naming result; Among them, the coronary artery segmentation task is used to extract the filled contour of the coronary artery and assign each pixel in the angiography image to a single category. The coronary artery segmentation naming task is used to anatomically segment the main branches (such as the left anterior descending branch, the circumflex branch, the right coronary artery, etc.) and their secondary branches of the coronary artery on the basis of coronary artery segmentation, and assign the pixels of different segments to different categories.

[0034] Furthermore, the training process of the joint segmentation and segmentation naming model provided in this embodiment is described in detail, including: Step 210: Train a deep learning model based on a pre-constructed multi-task learning framework so that the deep learning model can learn multiple tasks simultaneously; wherein, each task in the multiple tasks has a corresponding loss function, and the weighted sum of the loss functions of each task in the multiple tasks constitutes the total loss function; the total loss function adjusts the correlation between the segmentation loss and the segmentation loss through a consistency index, and the consistency index is calculated based on the number of overlapping region pixels of the segmentation error map and the segmentation error map.

[0035] Step 220: Dynamically adjust the weights of the deep learning model by minimizing the total loss function to obtain the joint segmentation and segmentation model.

[0036] Specifically, the image encoding layer is implemented by a feature extraction network, including but not limited to models based on convolutional neural networks (CNNs) and models based on Transformers. After the coronary artery image is encoded by the feature extraction network, a multi-scale feature matrix of the image is obtained for use by the decoding layers of different tasks.

[0037] Specifically, both the image segmentation part and the segmentation naming part are provided with forward-propagable loss functions, and the loss functions include but are not limited to mean squared error loss, cross-entropy loss, and logarithmic loss. In the present invention, a multi-task learning framework is set up in the model. The model simultaneously learns multiple tasks, and each task is provided with an independent loss function. These loss functions are combined into a total loss function. The model updates the weights by minimizing the total loss function, thereby achieving the collaborative optimization of the segmentation and segmentation tasks.

[0038] In some embodiments, the total loss function can be calculated by the following formula: (1) Where, is the total loss of the model, is the segmentation loss, is the segmentation naming loss, and are preset parameters respectively.

[0039] In some embodiments, the total loss function can also be calculated by the following formula: (2) The consistency index k proposed in this embodiment is used to quantify the correlation between the segmentation task and the segmentation task. Its calculation method includes the following steps: Error map generation: The model generates pixel-level difference maps with the true values through the segmentation task and the segmentation task respectively, namely the segmentation error map and the segmentation error map.

[0040] Definition of consistent pixels: When there is a spatially overlapping area between the segmentation error map and the segmentation error map, if a certain pixel point shows significant differences in both error maps at the same time, then this pixel point is defined as a consistent pixel. These consistent pixels characterize the segmentation error-induced segmentation error regions.

[0041] Index calculation: The mathematical expression of the consistency index k is the ratio of the total number of consistent pixels to the total number of significantly different pixels in the segmentation error map. The specific formula is as follows: (3) When the segmentation error-induced segmentation fault is more significant, the system generates more consistent pixels, causing the k value to approach 1. This metric can effectively reflect the impact of the segmentation task on the segmentation task, providing a quantitative basis for model optimization.

[0042] In the embodiments of the present invention, in addition to the foregoing loss function fusion method, another method for optimizing the image segmentation result is provided. According to the predicted probability of each pixel, binaryzation is performed by setting a threshold to distinguish the foreground (vascular region) from the background. The specific implementation method is as follows: The segmentation loss and the segmentation loss of the image respectively calculate their own loss functions and independently update the network parameters; the vascular segmentation probability is determined according to the segmentation loss, and the vascular segmentation image is updated; when calculating the segmentation loss, the intersection of the segmentation result and the predicted segmentation result is taken, and the segmentation and segmentation losses are only calculated for the pixels within the predicted segmentation region; for the region of the difference between the predicted union and the intersection, only the segmentation loss is calculated to ensure the accuracy of the segmentation result.

[0043] In the embodiments of the present invention, the loss weight is dynamically adjusted according to the training stage. In the early stage of training, the proportion of the segmentation loss is increased, and the proportion of the segmentation loss is decreased to give priority to training the overall vascular contour; in the middle stage of training, the proportion of the segmentation loss is increased, and the proportion of the segmentation loss is decreased. On the basis of stabilizing the vascular contour, the segmentation naming categories within the contour are optimized; in the later stage of training, the segmentation loss and the segmentation loss are balanced, and the overall model performance is optimized to ensure the collaborative convergence of the segmentation and segmentation tasks.

[0044] When evaluating the performance of the semantic segmentation and instance segmentation multi-task model on the validation set, the following steps are adopted: Input the validation images not participating in the training into the trained model to obtain the prediction results; According to the intersection over union (IoU) and Dice coefficient metrics, quantitatively evaluate the performance of the model on the segmentation and segmentation tasks to ensure the accuracy and robustness of the output results. The calculation methods are as follows: (4) (5) Among them, is the predicted pixel region, is the true pixel region, represents the intersection area of the predicted pixel region and the true pixel region.

[0045] The embodiment of the present invention constructs a joint segmentation model, which synchronously completes the coronary segmentation and segmentation naming tasks through a multi-task learning framework. The model uses a shared encoding layer to extract multi-scale features, and designs independent decoding layers (segmentation decoding layer and segmentation decoding layer) to handle different tasks respectively. The segmentation task generates vascular contours through the U-Net structure, while the segmentation task names the branches based on anatomical classification. The feature fusion is enhanced by skip connections and attention mechanisms to ensure the consistency of segmentation and segmentation results at the pixel level. This joint model avoids the problem of error accumulation in traditional step-by-step processing and significantly improves the efficiency and accuracy of coronary artery analysis. A dynamic loss weight allocation strategy is introduced in model training. In the early stage of training, segmentation loss is dominant, and accurate vascular contours are established first; in the middle of training, segmentation loss is turned to refine the branch naming classification; in the late stage of training, the two are balanced for global optimization. The correlation between tasks is quantified by consistency indicators, and the dual-task errors are integrated in the total loss function. This mechanism ensures that the model can automatically identify segmentation deviations caused by segmentation errors, optimize parameters in a targeted manner, and improve model robustness.

[0046] Step 300: Generate an ordered centerline of the coronary vessel based on the vessel segmentation result and the segment naming result; wherein the ordered centerline is a centerline in which each segment in the skeleton diagram of the coronary vessel has a clear starting point, an end point and a direction.

[0047] Step 300 generates an ordered centerline of the coronary vessels based on the vessel segmentation result and the segment naming result, including: Step 310: extract the blood vessel segmentation result to obtain an initial skeleton image of the coronary blood vessels, filter the noise branches in the initial skeleton image according to a preset branch pixel length threshold, and obtain the skeleton image of the coronary blood vessels.

[0048] Step 320 : According to the anatomical rules of the coronary artery tree 16 segmentation method, a target segment is determined based on bifurcation points and end points in the skeleton image of the coronary vessels.

[0049] Step 330: Determine the starting point of the coronary vessel according to the segment naming result.

[0050] Step 340: Generate an ordered centerline of the coronary vessel based on the skeleton map of the coronary vessel, the target segment and the starting point of the coronary vessel.

[0051] Specifically, first, a morphological thinning algorithm or a distance transformation method is used to extract an initial skeleton graph of the coronary artery from the blood vessel segmentation result. The initial skeleton graph mainly includes the main branch skeletons of the blood vessels and is used for subsequent segmentation processing. Next, the noise branches in the initial skeleton graph are filtered to obtain a skeleton graph, specifically including removing the skeleton parts with too short branch lengths, setting a branch pixel length threshold, and considering the branches shorter than the threshold as invalid segments. At the same time, based on the definition of pixel-based segmentation, the skeleton graph is divided into the following segmentation types: one segment from the starting point to the bifurcation point, one segment from the starting point to the branch end, and one segment from the bifurcation point to the branch end. Then, according to the segmentation naming result and combined with the 16-segment method of the coronary artery tree, the starting position of the coronary artery is determined. Specifically, the starting segment of the left coronary artery is segment 5, and the starting segment of the right coronary artery is segment 1. Based on the segmentation categories of the left and right coronary arteries, the spatial position of the starting point is mapped to the skeleton graph. Finally, based on the skeleton graph and the starting point position, an ordered centerline is generated. The ordered centerline is a centerline in which each segment in the skeleton graph has a clear starting point, ending point, and direction, ensuring that the output result conforms to the anatomical structure characteristics of the coronary artery tree and providing structured data support for clinical diagnosis.

[0052] In the embodiment of the present invention, based on the segmentation result, a morphological thinning algorithm is used to extract the blood vessel skeleton graph, and noise filtering is performed in combination with anatomical rules. A branch length threshold is defined, and branches that are too short are regarded as noise and invalid branches are removed. The skeleton segments are divided based on the bifurcation points and end points to ensure that the topological structure conforms to the characteristics of the coronary artery tree. The starting point is determined through the anatomical segmentation naming result, an ordered centerline is generated, and each segment is given a clear starting point, ending point, and direction, providing structured data support for clinical practice.

[0053] Step 400: Generate a segmented naming result of the coronary centerline of the coronary artery that conforms to anatomical rules based on the ordered centerline and the segmented naming result.

[0054] Step 400 generates a segmented naming result of the coronary centerline of the coronary artery based on the ordered centerline and the segmented naming result, including: Step 410: Assign segment category values to the coordinate points on the ordered centerline in sequence according to the ordered centerline and the segmented naming result to determine the category of each coordinate point.

[0055] Step 420: Based on the category of each determined coordinate point, perform a determination on each ordered centerline segment through a dynamic category correction strategy, and output a segmented naming result of the coronary centerline of the coronary artery that conforms to anatomical rules.

[0056] Step 420 specifically includes: Step 421: In the case where there are multiple segmented naming results in an ordered centerline, if the pixel ratio of any category in the ordered centerline is lower than a preset threshold and the distribution is discontinuous, then determine that the category is noise and filter it out.

[0057] Step 422: In the case where there are multiple segmented naming results in an ordered centerline, if the pixel ratio of any category in the ordered centerline is higher than a preset threshold and the distribution is continuous, then retain the category and perform mapping matching with the segmented naming result, and output the segmented naming result of the coronary artery centerline of the coronary artery vessel that conforms to the anatomical rules.

[0058] Specifically, first, according to the ordered centerline and the segmented naming result that conforms to the anatomy, the segmentation category assignment is sequentially performed on the coordinate points on the ordered centerline to ensure that each coordinate point has a clear segmentation label. For the case where there are multiple segmented naming results in a centerline segment, if the proportion of a single category in the overall segmentation is low and discontinuous, then this category is excluded as noise, and the entire segmentation is unified into the segmented naming result that conforms to the anatomy. For the case where there are multiple segmented naming results in a centerline segment, if the proportion of a single category in the overall segmentation is high and continuous, then retain this category, regard it as the difference between the pixel-level segmentation and the anatomical segmentation, and take the anatomical segmentation naming as the standard to ensure that the output result conforms to the clinical diagnosis standard.

[0059] The embodiment of the present invention proposes a dynamic category correction strategy for the integration of segmented naming and the centerline. The abnormal classification pixels with low proportion and discrete distribution are directly excluded; for the continuous and dominant segmentation differences, the anatomical naming is given priority. This mechanism effectively balances the conflict between pixel-level prediction noise and anatomical rules, ensuring that the output result has both algorithm accuracy and clinical interpretability.

[0060] The above is a description of the steps of the coronary artery centerline segmentation and naming method based on angiographic image segmentation provided by the present invention. From the description of the above steps, it can be seen that according to the coronary artery centerline segmentation and naming method based on angiographic image segmentation provided by the present invention, an angiographic image of the coronary artery is obtained; wherein, the angiographic image includes: image data of at least one target shooting position; the angiographic image is input into a pre-trained joint segmentation and segmentation model to output a blood vessel segmentation result and a segmentation naming result; wherein, the joint segmentation and segmentation model is: obtained by training a multi-task deep learning model based on a coronary artery angiographic image dataset; based on the blood vessel segmentation result and the segmentation naming result, an ordered centerline of the coronary artery is generated; wherein, the ordered centerline is: a centerline in which each segment in the skeleton diagram of the coronary artery has a clear starting point, ending point, and direction; based on the ordered centerline and the segmentation naming result, a coronary artery centerline segmentation naming result of the coronary artery that conforms to anatomical rules is generated. It can be seen that the present invention synchronously completes two tasks through the joint segmentation and segmentation model, avoiding error accumulation and improving the overall analysis efficiency and accuracy.

[0061] Next, the coronary artery centerline segmentation and naming device based on angiographic image segmentation provided by the present invention will be described. The coronary artery centerline segmentation and naming device described below can be correspondingly referred to the coronary artery centerline segmentation and naming method described above.

[0062] Figure 3 is a schematic structural diagram of the coronary artery centerline segmentation and naming device based on angiographic image segmentation provided by the present invention, as Figure 3 shown, the coronary artery centerline segmentation and naming device based on angiographic image segmentation provided by the present invention includes: An acquisition module 301, configured to acquire an angiographic image of a coronary artery; wherein, the angiographic image includes: image data of at least one target shooting position; A segmentation and naming module 302, configured to input the angiographic image into a pre-trained joint segmentation and segmentation model to output a blood vessel segmentation result and a segmentation naming result; wherein, the joint segmentation and segmentation model is: obtained by training a multi-task deep learning model based on a coronary artery angiographic image dataset; A centerline extraction module 303, configured to generate an ordered centerline of the coronary artery based on the blood vessel segmentation result and the segmentation naming result; wherein, the ordered centerline is: a centerline in which each segment in the skeleton diagram of the coronary artery has a clear starting point, ending point, and direction; A centerline and segmentation naming fusion module 304, configured to generate a coronary artery centerline segmentation naming result of the coronary artery that conforms to anatomical rules based on the ordered centerline and the segmentation naming result.

[0063] The coronary artery centerline segmentation and naming device based on contrast image segmentation provided by the present invention obtains the angiography image of the coronary artery; wherein, the angiography image includes: image data of at least one target shooting position; inputs the angiography image into a pre-trained joint segmentation and segmentation model to output a blood vessel segmentation result and a segmentation naming result; wherein, the joint segmentation and segmentation model is: obtained by training a multi-task deep learning model based on a coronary artery angiography image dataset; based on the blood vessel segmentation result and the segmentation naming result, generates an ordered centerline of the coronary artery; wherein, the ordered centerline is: a centerline in the skeleton diagram of the coronary artery where each segment has a clear starting point, ending point, and direction; based on the ordered centerline and the segmentation naming result, generates a coronary artery centerline segmentation and naming result of the coronary artery that conforms to anatomical rules. It can be seen that the present invention completes two tasks synchronously through the joint segmentation and segmentation model, avoids error accumulation, and improves the overall analysis efficiency and accuracy.

[0064] Based on the above embodiment, in this embodiment, the joint segmentation and segmentation model includes: an image encoding layer, a segmentation image decoding layer, and a segmentation image decoding layer, and the image encoding layer, the segmentation image decoding layer, and the segmentation image decoding layer constitute a coronary artery segmentation task and a coronary artery segmentation naming task; The segmentation naming module 302 is specifically used for: Inputs the angiography image into the joint segmentation and segmentation model, respectively executes the coronary artery segmentation task and the coronary artery segmentation naming task, and outputs the blood vessel segmentation result and the segmentation naming result; Wherein, the coronary artery segmentation task: is used to extract the filled contour of the coronary artery, and assigns each pixel in the angiography image to a single category, and the coronary artery segmentation naming task: is used to anatomically segment the main branches and their secondary branches of the coronary artery on the basis of coronary artery segmentation, and assigns the pixels of different segments to different categories.

[0065] Based on the above embodiment, in this embodiment, the device further includes a training module, which is specifically used for: Training a deep learning model based on a pre-constructed multi-task learning framework so that the deep learning model can learn multiple tasks simultaneously; wherein, each task in the multiple tasks has a corresponding loss function, and the weighted sum of the loss functions of each task in the multiple tasks constitutes a total loss function; the total loss function adjusts the correlation between the segmentation loss and the segmentation loss through a consistency index, and the consistency index is calculated based on the number of overlapping region pixels of the segmentation error map and the segmentation error map; Dynamically adjusts the weights of the deep learning model by minimizing the total loss function to obtain the joint segmentation and segmentation model.

[0066] Based on the above embodiments, in this embodiment, the device further includes a dynamic adjustment module, which is specifically used for: In the early stage of training, the segmentation loss weight is higher than the segmentation loss weight; In the middle stage of training, the segmentation loss weight is higher than the segmentation loss weight; In the late stage of training, the segmentation loss weight and the segmentation loss weight are balanced.

[0067] Based on the above embodiments, in this embodiment, the centerline extraction module 303 is specifically used for: Extract the initial skeleton diagram of the coronary artery from the blood vessel segmentation result, and filter the noise branches in the initial skeleton diagram according to a preset branch pixel length threshold to obtain the skeleton diagram of the coronary artery; Based on the anatomical rules of the 16-segment method of the coronary artery tree, determine the target segments in the skeleton diagram of the coronary artery based on the bifurcation points and end points; Determine the starting point of the coronary artery according to the segmentation naming result; Generate the ordered centerline of the coronary artery based on the skeleton diagram of the coronary artery, the target segments, and the starting point of the coronary artery.

[0068] Based on the above embodiments, in this embodiment, the centerline and segmentation naming fusion module 304 is specifically used for: According to the ordered centerline and the segmentation naming result, assign segmentation category values to the coordinate points on the ordered centerline in sequence to determine the category of each coordinate point; Based on the category of each determined coordinate point, judge each ordered centerline segment respectively through a dynamic category correction strategy, and output the coronary centerline segmentation naming result of the coronary artery that conforms to the anatomical rules.

[0069] Based on the above embodiments, in this embodiment, the centerline and segmentation naming fusion module 304 is specifically further used for: In the case where there are multiple segmentation naming results in an ordered centerline segment, if the pixel ratio of any category in the ordered centerline is lower than a preset threshold and the distribution is discontinuous, then determine that the category is noise and filter it out; In the case where there are multiple segmentation naming results in an ordered centerline segment, if the pixel ratio of any category in the ordered centerline is higher than a preset threshold and the distribution is continuous, then retain the category and perform mapping matching with the segmentation naming result, and output the coronary centerline segmentation naming result of the coronary artery that conforms to the anatomical rules.

[0070] Figure 4 Illustrates a schematic diagram of the physical structure of an electronic device, such asFigure 4 As shown, the electronic device can be a robot or other electronic device, and the electronic device can include: a processor 410, a communications interface 420, a memory 430, and a communication bus 440. Among them, the processor 410, the communications interface 420, and the memory 430 complete mutual communication through the communication bus 440. The processor 410 can call the logical instructions in the memory 430 to execute the method for segment naming of the coronary artery centerline based on angiographic image segmentation, including: Obtain an angiographic image of the coronary artery; wherein, the angiographic image includes: image data of at least one target shooting position; Input the angiographic image into a pre-trained joint segmentation and segmentation model to output a blood vessel segmentation result and a segmentation naming result; wherein, the joint segmentation and segmentation model is: obtained by training a multi-task deep learning model based on a coronary artery angiographic image dataset; Generate an ordered centerline of the coronary artery based on the blood vessel segmentation result and the segmentation naming result; wherein, the ordered centerline is: a centerline in the skeleton diagram of the coronary artery where each segment has a clear starting point, ending point, and direction; Generate a coronary artery centerline segmentation naming result of the coronary artery that conforms to anatomical rules based on the ordered centerline and the segmentation naming result.

[0071] In addition, when the logical instructions in the above-mentioned memory 430 are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs that can store program codes.

[0072] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the coronary artery centerline segmentation and naming method based on angiographic image segmentation provided by the above-mentioned various methods, including: Obtain an angiographic image of the coronary artery; wherein, the angiographic image includes: image data of at least one target shooting position; Input the angiographic image into a pre-trained joint segmentation and segmentation model to output a blood vessel segmentation result and a segmentation naming result; wherein, the joint segmentation and segmentation model is obtained by training a multi-task deep learning model based on a coronary artery angiographic image dataset; Generate an ordered centerline of the coronary artery based on the blood vessel segmentation result and the segmentation naming result; wherein, the ordered centerline is a centerline in the skeleton diagram of the coronary artery where each segment has a clear starting point, ending point, and direction; Generate a coronary artery centerline segmentation naming result of the coronary artery that conforms to anatomical rules based on the ordered centerline and the segmentation naming result.

[0073] On another aspect, the present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it is implemented to execute the coronary artery centerline segmentation and naming method based on angiographic image segmentation provided by the above-mentioned various methods, including: Obtain an angiographic image of the coronary artery; wherein, the angiographic image includes: image data of at least one target shooting position; Input the angiographic image into a pre-trained joint segmentation and segmentation model to output a blood vessel segmentation result and a segmentation naming result; wherein, the joint segmentation and segmentation model is obtained by training a multi-task deep learning model based on a coronary artery angiographic image dataset; Generate an ordered centerline of the coronary artery based on the blood vessel segmentation result and the segmentation naming result; wherein, the ordered centerline is a centerline in the skeleton diagram of the coronary artery where each segment has a clear starting point, ending point, and direction; Generate a coronary artery centerline segmentation naming result of the coronary artery that conforms to anatomical rules based on the ordered centerline and the segmentation naming result.

[0074] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. A person of ordinary skill in the art can understand and implement it without creative labor.

[0075] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the essence of the above technical solution, or the part that contributes to the prior art, can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0076] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features. And these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for segmenting and naming the coronary artery centerline based on angiography image segmentation, characterized in that: include: Acquire an angiographic image of a coronary vessel; wherein the angiographic image comprises: image data of at least one target photographing position; The angiography image is input into a pre-trained joint segmentation model, and a vessel segmentation result and a segmentation naming result are output; wherein the joint segmentation model is obtained by training a multi-task deep learning model based on a coronary angiography image dataset; Based on the blood vessel segmentation result and the segment naming result, an ordered center line of the coronary blood vessel is generated; wherein the ordered center line is a center line in which each segment in the skeleton image of the coronary blood vessel has a clear starting point, an end point and a direction; Based on the ordered center lines and the segment naming results, a coronary center line segment naming result of the coronary blood vessel that complies with anatomical rules is generated.

2. The method for segmenting and naming the coronary artery centerline based on angiography image segmentation according to claim 1, characterized in that: The joint segmentation model includes: an image encoding layer, a segmentation image decoding layer and a segmentation image decoding layer, wherein the image encoding layer, the segmentation image decoding layer and the segmentation image decoding layer constitute a coronary artery segmentation task and a coronary artery segmentation naming task; The step of inputting the angiography image into a pre-trained joint segmentation model and outputting a blood vessel segmentation result and a segmentation naming result includes: Inputting the angiography image into the joint segmentation model, performing the coronary segmentation task and the coronary segmentation naming task respectively, and outputting the blood vessel segmentation result and the segmentation naming result; Among them, the coronary artery segmentation task is used to extract the filled contour of the coronary artery and assign each pixel in the angiography image to a single category; the coronary artery segmentation naming task is used to anatomically segment the main branches of the coronary artery and their secondary branches based on the coronary artery segmentation, and assign pixels of different segments to different categories.

3. The method for segmenting and naming the coronary artery centerline based on angiography image segmentation according to claim 1, characterized in that: The training process of the joint segmentation model includes: A deep learning model is trained based on a pre-built multi-task learning framework so that the deep learning model learns multiple tasks simultaneously; wherein each of the multiple tasks has a corresponding loss function, and the weighted sum of the loss functions of each of the multiple tasks constitutes a total loss function; the total loss function adjusts the correlation between the segmentation loss and the segmentation loss through a consistency index, and the consistency index is calculated based on the number of pixels in the overlapping area of ​​the segmentation error map and the segmentation error map; The weights of the deep learning model are dynamically adjusted by minimizing the total loss function to obtain the joint segmentation model.

4. The method for segmenting and naming the coronary artery centerline based on angiography image segmentation according to claim 3, characterized in that: The dynamically adjusting the weight of the deep learning model by minimizing the total loss function comprises: In the early stage of training, the segmentation loss weight is higher than the segmentation loss weight; In the middle of training, the weight of segmentation loss is higher than that of segmentation loss; In the later stages of training, the segmentation loss weight is balanced with the fragmentation loss weight.

5. The method for segmenting and naming the coronary artery centerline based on angiography image segmentation according to claim 1, characterized in that: The step of generating the ordered centerline of the coronary vessels based on the vessel segmentation result and the segment naming result includes: Extracting the blood vessel segmentation result to obtain an initial skeleton image of the coronary blood vessel, filtering noise branches in the initial skeleton image according to a preset branch pixel length threshold, to obtain a skeleton image of the coronary blood vessel; According to the anatomical rules of the coronary artery tree 16 segmentation method, a target segment is determined based on bifurcation points and end points in the skeleton image of the coronary vessels; Determining the starting point of the coronary vessel according to the segmented naming result; An ordered centerline of the coronary vessel is generated based on the skeleton map of the coronary vessel, the target segment and the starting point of the coronary vessel.

6. The method for segmenting and naming the coronary artery centerline based on angiography image segmentation according to claim 1, characterized in that: The generating of the segmented naming result of the coronary centerline of the coronary vessel based on the ordered centerline and the segmented naming result comprises: According to the ordered center line and the segment naming result, the coordinate points on the ordered center line are assigned segment categories in turn to determine the category of each coordinate point; Based on the determined category of each coordinate point, each ordered centerline is judged separately through a dynamic category correction strategy, and the segmented naming result of the coronary centerline of the coronary blood vessels that conforms to anatomical rules is output.

7. The method for segmenting and naming the coronary artery centerline based on angiography image segmentation according to claim 6, characterized in that: Based on the determined category of each coordinate point, each ordered center line is judged separately through a dynamic category correction strategy, and the segmented naming result of the coronary center line of the coronary blood vessel that conforms to the anatomical rules is output, including: In the case where there are multiple segmented naming results in an ordered center line, if the pixel ratio of any category in the ordered center line is lower than a preset threshold and the distribution is discontinuous, the category is determined to be a noise point and filtered out; When there are multiple segmentation naming results in an ordered centerline, if the pixel ratio of any category in the ordered centerline is higher than a preset threshold and is continuously distributed, the category is retained and mapped and matched with the segmentation naming result, and the segmentation naming result of the coronary centerline of the coronary vessel that conforms to the anatomical rules is output.

8. A device for naming coronary artery centerline segments based on angiography image segmentation, characterized in that: include: An acquisition module, used for acquiring an angiographic image of a coronary vessel; wherein the angiographic image comprises: image data of at least one target shooting position; A segmentation and naming module is used to input the angiography image into a pre-trained joint segmentation and segmentation model, and output a vessel segmentation result and a segmentation and naming result; wherein the joint segmentation and segmentation model is obtained by training a multi-task deep learning model based on a coronary angiography image dataset; A centerline extraction module, for generating an ordered centerline of the coronary vessel based on the vessel segmentation result and the segment naming result; wherein the ordered centerline is a centerline in which each segment in the skeleton image of the coronary vessel has a clear starting point, an end point and a direction; The centerline and segment naming fusion module is used to generate a coronary centerline segment naming result of the coronary blood vessel that complies with anatomical rules based on the ordered centerline and the segment naming result.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the method for naming the coronary artery centerline segments based on angiography image segmentation as described in any one of claims 1 to 7 is implemented.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for naming the coronary centerline segments based on angiography image segmentation as described in any one of claims 1 to 7 is implemented.

Images