In-stent ffr calculation method and device, electronic equipment and storage medium

Abstract

This application relates to the field of medical image processing and intelligent diagnostic technology, and discloses a method, device, electronic device, and storage medium for calculating in-stent FFR. The method includes: acquiring a stent PCCT image of any patient and locating the stent position in the image; performing stent boundary recognition based on the stent position in the stent PCCT image to obtain at least a boundary fitting curve; performing image region segmentation based at least on the boundary fitting curve to extract at least a vascular image; determining the vascular centerline based at least on the vascular image, extracting average Hu features along the vascular centerline, and then integrating all average Hu features to obtain an average Hu sequence; and inputting the average Hu sequence and the vascular image into a pre-trained deep learning network model to output the corresponding vascular FFR. This application can at least alleviate the problem of difficulty in determining in-stent vascular FFR in existing methods, and helps to reduce the interference of stents on the vascular FFR determination results.

Description

Technical Field

[0001] This invention relates to the field of medical image processing and intelligent diagnostic technology, and in particular to an in-stent FFR calculation method, device, electronic device and storage medium. Background Technology

[0002] Currently, during the diagnostic process, conventional CT imaging is insufficient to determine the FFR (fractional flow reserve) of vessels within stents. The advent of Photon-Counting CT (PCCT) has made this possible, but because it assesses the condition within a stent, the stent itself can still negatively impact the determination of the vessel's FFR. Summary of the Invention

[0003] The purpose of this invention is to provide a method, apparatus, electronic device and storage medium for calculating in-stent FFR, which can at least alleviate the problem of difficulty in determining in-stent vessel FFR and reduce the degree of interference of stent on vessel FFR determination results.

[0004] To address the aforementioned technical problems, in a first aspect, the present invention provides a method for calculating in-stent FFR, comprising at least:

[0005] Acquire PCCT images of stents in any patient and locate the stent position in the images;

[0006] Stent boundary identification is performed based on the stent location in the stent PCCT image to obtain at least a boundary fitting curve;

[0007] Image region segmentation is performed at least based on the boundary fitting curve to extract vascular images of at least the non-stent region;

[0008] Based on at least the vascular image, the vascular centerline is determined, and the average Hu feature is extracted along the vascular centerline. Then, all the average Hu features are integrated to obtain the average Hu sequence.

[0009] The average Hu sequence and the blood vessel image are input into a pre-trained deep learning network model to output the corresponding blood vessel FFR.

[0010] Optionally, the step of determining the vessel centerline based at least on the vessel image, extracting average Hu features along the vessel centerline, and then integrating all the average Hu features to obtain an average Hu sequence specifically includes at least:

[0011] The centerline of the blood vessel shall be determined at least based on the blood vessel image;

[0012] At least one CT value of a blood vessel region is extracted point by point along the center line of the blood vessel towards the normal direction of each point, and the average value of all the CT values ​​of the blood vessel regions is used as the average Hu feature of the corresponding point.

[0013] The average Hu sequence is obtained by integrating all the average Hu features.

[0014] Optionally, the step of inputting at least the average Hu sequence and the blood vessel image into a pre-trained deep learning network model to output the corresponding blood vessel FFR specifically includes at least:

[0015] The average Hu sequence and the blood vessel image are input into a pre-trained deep learning network model so that the deep learning network model convolves the blood vessel image to obtain a one-dimensional feature sequence.

[0016] Optionally, the step of inputting at least the average Hu sequence and the blood vessel image into the pre-trained deep learning network model to output the corresponding blood vessel FFR specifically includes at least:

[0017] After performing a concat operation on the one-dimensional feature sequence and the average Hu sequence through the deep learning network model, convolution is then performed to output the corresponding blood vessel FFR.

[0018] Based on the same concept, in a second aspect, the present invention also provides an in-stent FFR calculation device for performing the in-stent FFR calculation method described in any one of the first aspects;

[0019] The FFR calculation device within the support structure includes at least:

[0020] The stent positioning module is used to acquire stent PCCT images of any patient and locate the stent position in the images.

[0021] A boundary fitting module is used to perform stent boundary identification based on the stent location in the stent PCCT image, so as to obtain at least a boundary fitting curve;

[0022] A region segmentation module is used to perform image region segmentation at least based on the boundary fitting curve, so as to extract at least the vascular image of the non-stent region;

[0023] The feature integration module is used to determine the blood vessel centerline based on at least the blood vessel image, extract the average Hu feature along the blood vessel centerline, and then integrate all the average Hu features to obtain the average Hu sequence.

[0024] The result output module is used to input at least the average Hu sequence and the blood vessel image into the pre-trained deep learning network model to output the corresponding blood vessel FFR.

[0025] Optionally, the feature integration module is specifically used for at least:

[0026] The centerline of the blood vessel shall be determined at least based on the blood vessel image;

[0027] At least one CT value of a blood vessel region is extracted point by point along the center line of the blood vessel towards the normal direction of each point, and the average value of all the CT values ​​of the blood vessel regions is used as the average Hu feature of the corresponding point.

[0028] The average Hu sequence is obtained by integrating all the average Hu features.

[0029] Optionally, the result output module is specifically used for at least:

[0030] The average Hu sequence and the blood vessel image are input into a pre-trained deep learning network model so that the deep learning network model convolves the blood vessel image to obtain a one-dimensional feature sequence.

[0031] Optionally, the result output module is specifically used for at least:

[0032] After performing a concat operation on the one-dimensional feature sequence and the average Hu sequence through the deep learning network model, convolution is then performed to output the corresponding blood vessel FFR.

[0033] 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 in-support FFR calculation method of any one of the first aspects.

[0034] 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, wherein the computer program, when executed by a processor, implements the steps in the in-stent FFR calculation method of any one of the first aspects.

[0035] The technical solution provided by this invention first acquires a PCCT image of a stent from any patient and locates the stent position within the image. Further, stent boundary recognition is performed in the PCCT image based on the stent position to obtain at least a boundary fitting curve. Further, image region segmentation is performed based on the boundary fitting curve to extract at least the vascular image of the non-stent region. Further, the vascular centerline is determined based on the vascular image, and average Hu features are extracted along the vascular centerline, then all average Hu features are integrated to obtain an average Hu sequence. Finally, the average Hu sequence and the vascular image are input into a pre-trained deep learning network model to output the corresponding vascular FFR. Therefore, this invention provides at least one PCCT-based in-stent vascular FFR calculation process, which can alleviate the difficulty of determining in-stent vascular FFR in existing methods and reduce the interference of stents on vascular FFR determination results. Attached Figure Description

[0036] Figure 1 This is a flowchart of an in-stent FFR calculation method provided by an embodiment of the present invention;

[0037] Figure 2 This is a schematic diagram of the structure of an in-frame FFR calculation device provided in an embodiment of the present invention;

[0038] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0039] 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.

[0040] 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.

[0041] 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.

[0042] 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.

[0043] 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).”

[0044] 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.

[0045] 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.

[0046] Figure 1 This is a flowchart of an in-stent FFR calculation method provided by an embodiment of the present invention. This embodiment is at least applicable to clinical in-stent vessel FFR calculation and disease diagnosis scenarios. The in-stent FFR calculation method can be, but is not limited to, executed by the in-stent FFR calculation device in this embodiment of the invention. This execution device can be implemented in software and / or hardware. Figure 1 As shown, the method for calculating FFR within the stent includes at least the following steps:

[0047] S1. Obtain PCCT images of stents from any patient and locate the stent position in the images.

[0048] Step S1 can be achieved using any existing stent detection network, which can employ neural network models or multi-classification network models, etc.

[0049] In one specific implementation, the stent positioning process can be as follows:

[0050] (1) Obtain the pre-identified region of the stent in the PCCT image of the stent.

[0051] Specifically, the stent pre-identification area can be marked as area A, that is, area A in the stent PCCT image is the stent pre-identification area.

[0052] (2) Perform feature extraction on the pre-identified area of ​​the stent to generate a feature image of a preset size.

[0053] Specifically, the pre-identified region of the support can be imaged to generate a local region image; further, feature extraction can be performed on the local region image to generate a feature image of a preset size.

[0054] It is understood that the preset size can refer to a fixed length or a fixed dimension. Specifically, region image extraction can be performed on the aforementioned region A to generate a first local region image, a second local region image, and a third local region image; further, a neural network is used to extract features from the first local region image to generate a first feature image of fixed length; further, a neural network is used to extract features from the second local region image to generate a second feature image of fixed length; further, a neural network is used to extract features from the third local region image to generate a third feature image of fixed length.

[0055] It is understandable that the first feature image, the second feature image, and the third feature image can have the same length. The reason for this setting is that, from the perspective of matrix representation, only feature images with the same size or length can be combined.

[0056] (3) Combine the features of the feature images to generate a pre-identified scaffold image.

[0057] Specifically, feature combination can be performed on the generated fixed-length first feature image, second feature image, and third feature image to generate a pre-identified scaffold image.

[0058] (4) Perform stent detection on the pre-identified stent image to locate the stent position in the stent PCCT image.

[0059] Specifically, stent detection can be performed on pre-identified stent images. If a pre-identified stent image is detected as a stent, the detection result of the stent in the stent PCCT image can be obtained, and the stent position in the stent PCCT image can be located.

[0060] S2. Perform stent boundary identification based on stent location in stent PCCT images to obtain at least a boundary fitting curve.

[0061] Since vascular stents are mostly tubular mesh structures, the stent boundaries in stent PCCT images may appear as "discontinuous highlights." Therefore, based on the stent location determined in step S1, pixels above a preset CT threshold can be selected from the stent PCCT image, and the outer and inner contours of the stent can be fitted accordingly. In another specific implementation, the boundary fitting curve refers at least to the aforementioned fitted inner contour of the stent.

[0062] S3. Perform image region segmentation based at least on the boundary fitting curve to extract vascular images of at least the non-stent region.

[0063] One method for extracting vascular images in non-stent regions can be an image region segmentation method based on deep learning algorithms, which may include the following steps:

[0064] Data acquisition and cleaning: Select the original PCCT images of the stent, acquire the selected original PCCT images of the stent and desensitize them, clean the data according to the preset exclusion criteria, remove case images that affect identification and annotation, eliminate potential image interference in the images, and integrate them into a research dataset for subsequent annotation, training and validation.

[0065] Data annotation: Desensitized image data is manually annotated using strategies such as semi-automatic iterative annotation and hierarchical annotation to reduce the annotation workload of doctors and improve the efficiency of building the image region segmentation database. Contour annotation follows a coarse-to-fine principle. First, the vascular region features are manually coarsely annotated to obtain the coarsely annotated vascular region features. Based on the coarsely annotated vascular region features, a preliminary image region segmentation model is built. Subsequently, the vascular region contours are automatically generated based on this model, manually annotated and modified, and the model is repeatedly trained and tested using subsequent batches of finely annotated data to iteratively optimize the model accuracy. This is the semi-automatic iterative annotation method.

[0066] Data partitioning: The labeled stent PCCT image atlas is divided into training and test sets according to a certain ratio. For example, the training set can be divided into 10 equal-sized folds, and after 10 cross-validations, the optimal model hyperparameters are selected through grid search. The model hyperparameters are then fine-tuned using 10x cross-validation on the training set. In this process, the training data is divided into roughly equal-sized subsets; training uses a subset, while validation is performed on the remaining subset. The optimized model then enters the testing phase, with the test set independent of the training set.

[0067] Blood vessel region segmentation: Based on fully convolutional neural networks, models including but not limited to BASNet, UNet, and Mask R-CNN can be used. The training loss function is adjusted, replacing the original model's fusion loss with Focal Loss. This simplifies the loss function while improving edge detection capabilities and training stability. A small amount of data is used as a validation set during training to verify the model's performance. By observing changes in the loss metrics of the training and validation sets, training is terminated promptly to prevent overfitting.

[0068] Test data: Input the test set, load the image region segmentation model, output the vascular image of the non-stent region, and save the output results.

[0069] S4. At least the vascular centerline is determined based on the vascular image, and the average Hu feature is extracted along the vascular centerline. Then, all average Hu features are integrated to obtain the average Hu sequence.

[0070] Among them, the average Hu feature can be matched one-to-one with each point on the center line of the blood vessel.

[0071] In yet another specific implementation, step S4 may optionally include at least the following:

[0072] (1) Determine the centerline of the blood vessels at least based on the vascular images;

[0073] (2) Extract at least one CT value of a blood vessel region along the center line of the blood vessel towards the normal direction of each point, and take the average value of all CT values ​​of the blood vessel regions as the average Hu feature of the corresponding point.

[0074] (3) Integrate all average Hu features to obtain the average Hu sequence.

[0075] S5. Input the average Hu sequence and blood vessel image into the pre-trained deep learning network model to output the corresponding blood vessel FFR.

[0076] There are several options for choosing a deep learning network model, such as a convolutional neural network model.

[0077] In yet another specific implementation, step S5 may optionally include at least the following:

[0078] (1) At least the average Hu sequence and blood vessel image are input into the pre-trained deep learning network model so that the deep learning network model can convolve the blood vessel image to obtain a one-dimensional feature sequence;

[0079] (2) Continue to perform Concat operation on the one-dimensional feature sequence and the average Hu sequence through the deep learning network model, and then perform convolution to output the corresponding blood vessel FFR.

[0080] In yet another specific implementation, the pre-training process of the deep learning network model may optionally include at least the following steps:

[0081] (1) Obtain pre-trained vascular images of multiple patients and their corresponding pre-trained vascular FFRs for model pre-training, and determine the vascular centerline based on each pre-trained vascular image, and extract the pre-trained average Hu features along the vascular centerline, and then integrate all the pre-trained average Hu features to obtain the pre-trained average Hu sequence.

[0082] (2) At least the pre-trained vascular images and the pre-trained average Hu sequence are input into the deep learning network model to generate FFR results;

[0083] (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 FFR results, the pre-trained vascular FFR 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.

[0084] (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.

[0085] The technical solution provided in this embodiment first acquires a PCCT image of a stent from any patient and locates the stent position within the image. Further, it performs stent boundary recognition based on the stent position in the PCCT image to obtain at least a boundary fitting curve. Further, it performs image region segmentation based on the boundary fitting curve to extract at least the vascular image of the non-stent region. Further, it determines the vascular centerline based on the vascular image and extracts the average Hu feature along the vascular centerline, then integrates all average Hu features to obtain an average Hu sequence. Finally, it inputs the average Hu sequence and the vascular image into a pre-trained deep learning network model to output the corresponding vascular FFR. Therefore, this embodiment provides at least one PCCT-based in-stent vascular FFR calculation process, which can alleviate the difficulty of determining in-stent vascular FFR in existing methods and help reduce the interference of stents on the vascular FFR determination results.

[0086] Figure 2 This is a schematic diagram of an in-stent FFR calculation device provided in an embodiment of the present invention. This embodiment is at least applicable to clinical scenarios involving FFR calculation and disease diagnosis of blood vessels within stents. The in-stent FFR calculation device can be implemented using software and / or hardware. Figure 2As shown, the in-stent FFR calculation device is used to perform the in-stent FFR calculation method of any of the foregoing embodiments or implementations.

[0087] The in-frame FFR calculation unit includes at least:

[0088] The stent positioning module 110 is used to acquire stent PCCT images of any patient and locate the stent position in the images.

[0089] Boundary fitting module 120 is used to perform stent boundary identification based on stent location in stent PCCT images to obtain at least a boundary fitting curve;

[0090] Region segmentation module 130 is used to perform image region segmentation at least based on boundary fitting curves to extract vascular images of at least non-stent regions;

[0091] The feature integration module 140 is used to determine the blood vessel centerline based on at least the blood vessel image, extract the average Hu feature along the blood vessel centerline, and then integrate all the average Hu features to obtain the average Hu sequence.

[0092] The output module 150 is used to input at least the average Hu sequence and the blood vessel image into the pre-trained deep learning network model to output the corresponding blood vessel FFR.

[0093] Optionally, the feature integration module 140 is specifically used for at least:

[0094] Determine the centerline of the blood vessels at least based on vascular images;

[0095] At least one CT value of a blood vessel region is extracted point by point along the center line of the blood vessel towards the normal direction of each point, and the average value of the CT values ​​of all blood vessel regions is used as the average Hu feature of the corresponding point.

[0096] The average Hu sequence is obtained by integrating all average Hu features.

[0097] Optionally, the result output module 150 is specifically used for at least:

[0098] At least the average Hu sequence and blood vessel image are input into the pre-trained deep learning network model so that the deep learning network model can convolve the blood vessel image to obtain a one-dimensional feature sequence.

[0099] Optionally, the result output module 150 is specifically used for at least:

[0100] After performing a concat operation on the one-dimensional feature sequence and the average Hu sequence through a deep learning network model, convolution is then performed to output the corresponding blood vessel FFR.

[0101] The technical solution provided in this embodiment first acquires PCCT images of stents from any patient using a stent positioning module and locates the stent position within the images. Further, a boundary fitting module performs stent boundary recognition in the PCCT images based on the stent position to obtain at least a boundary fitting curve. Further, a region segmentation module performs image region segmentation based at least on the boundary fitting curve to extract at least the vascular image of the non-stent region. Further, a feature integration module determines the vascular centerline based at least on the vascular image and extracts average Hu features along the vascular centerline, then integrates all average Hu features to obtain an average Hu sequence. Finally, a result output module inputs at least the average Hu sequence and the vascular image into a pre-trained deep learning network model to output the corresponding vascular FFR. Therefore, this embodiment provides at least one PCCT-based in-stent vascular FFR calculation process, which can alleviate the difficulty of determining in-stent vascular FFR in existing methods and help reduce the interference of stents on vascular FFR determination results.

[0102] This embodiment provides an electronic device. Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. See also: Figure 3 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 above-mentioned bracket-in-the-frame FFR calculation methods 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 in-stent FFR calculation method in any optional implementation of the above embodiments, so as to at least achieve the following functions: acquire stent PCCT images of any patient and locate the stent position in the images; perform stent boundary recognition based on the stent position in the stent PCCT images to at least obtain a boundary fitting curve; perform image region segmentation based on the boundary fitting curve to at least extract the vascular images of the non-stent regions; determine the vascular centerline based on the vascular images and extract the average Hu features along the vascular centerline, and then integrate all the average Hu features to obtain the average Hu sequence; input the average Hu sequence and the vascular images into a pre-trained deep learning network model to output the corresponding vascular FFR.

[0103] This embodiment provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the in-stent FFR calculation method provided in all embodiments of this application: acquiring a stent PCCT image of any patient and locating the stent position in the image; performing stent boundary recognition based on the stent position in the stent PCCT image to obtain at least a boundary fitting curve; performing image region segmentation based at least on the boundary fitting curve to extract at least the vascular image of the non-stent region; determining the vascular centerline based at least on the vascular image and extracting the average Hu feature along the vascular centerline, and then integrating all average Hu features to obtain the average Hu sequence; and inputting the average Hu sequence and the vascular image into a pre-trained deep learning network model to output the corresponding vascular FFR.

[0104] 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.

[0105] 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.

[0106] 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.

[0107] 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).

[0108] 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 method for calculating FFR within a stent, characterized in that, At least including: Acquire PCCT images of stents in any patient and locate the stent position in the images; Stent boundary identification is performed based on the stent location in the stent PCCT image to obtain at least a boundary fitting curve; Image region segmentation is performed at least based on the boundary fitting curve to extract vascular images of at least the non-stent region; Based on at least the vascular image, the vascular centerline is determined, and the average Hu feature is extracted along the vascular centerline. Then, all the average Hu features are integrated to obtain the average Hu sequence. The average Hu sequence and the blood vessel image are input into a pre-trained deep learning network model to output the corresponding blood vessel FFR.

2. The method for calculating FFR within a stent according to claim 1, characterized in that, The step of determining the vascular centerline based at least on the vascular image, extracting average Hu features along the vascular centerline, and then integrating all the average Hu features to obtain an average Hu sequence specifically includes at least the following: The centerline of the blood vessel is determined based at least on the blood vessel image; At least one CT value of a blood vessel region is extracted point by point along the center line of the blood vessel towards the normal direction of each point, and the average value of all the CT values ​​of the blood vessel regions is used as the average Hu feature of the corresponding point. The average Hu sequence is obtained by integrating all the average Hu features.

3. The method for calculating FFR within a stent according to claim 1, characterized in that, The step of inputting at least the average Hu sequence and the blood vessel image into a pre-trained deep learning network model to output the corresponding blood vessel FFR specifically includes at least the following: The average Hu sequence and the blood vessel image are input into a pre-trained deep learning network model so that the deep learning network model convolves the blood vessel image to obtain a one-dimensional feature sequence.

4. The method for calculating FFR within a stent according to claim 3, characterized in that, The step of inputting the average Hu sequence and the blood vessel image into a pre-trained deep learning network model to output the corresponding blood vessel FFR specifically includes at least the following: After performing a concat operation on the one-dimensional feature sequence and the average Hu sequence through the deep learning network model, convolution is then performed to output the corresponding blood vessel FFR.

5. An in-frame FFR calculation device, characterized in that, Used to perform the in-stent FFR calculation method according to any one of claims 1-4; The FFR calculation device within the support structure includes at least: The stent positioning module is used to acquire stent PCCT images of any patient and locate the stent position in the images. A boundary fitting module is used to perform stent boundary identification based on the stent location in the stent PCCT image, so as to obtain at least a boundary fitting curve; A region segmentation module is used to perform image region segmentation at least based on the boundary fitting curve, so as to extract at least the vascular image of the non-stent region; The feature integration module is used to determine the blood vessel centerline based on at least the blood vessel image, extract the average Hu feature along the blood vessel centerline, and then integrate all the average Hu features to obtain the average Hu sequence. The result output module is used to input at least the average Hu sequence and the blood vessel image into the pre-trained deep learning network model to output the corresponding blood vessel FFR.

6. The in-support FFR calculation device according to claim 5, characterized in that, The feature integration module is specifically used for at least: The centerline of the blood vessel is determined based at least on the blood vessel image; At least one CT value of a blood vessel region is extracted point by point along the center line of the blood vessel towards the normal direction of each point, and the average value of all the CT values ​​of the blood vessel regions is used as the average Hu feature of the corresponding point. The average Hu sequence is obtained by integrating all the average Hu features.

7. The in-support FFR calculation device according to claim 5, characterized in that, The result output module is specifically used for at least: The average Hu sequence and the blood vessel image are input into a pre-trained deep learning network model so that the deep learning network model convolves the blood vessel image to obtain a one-dimensional feature sequence.

8. The in-support FFR calculation device according to claim 7, characterized in that, The result output module is specifically used for at least: After performing a concat operation on the one-dimensional feature sequence and the average Hu sequence through the deep learning network model, convolution is then performed to output the corresponding blood vessel FFR.

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 in-stent FFR calculation 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 in-stent FFR calculation method according to any one of claims 1 to 4.

Images