Method and device for determining fractional flow reserve, electronic equipment and storage medium

Abstract

This invention discloses a method, apparatus, electronic device, and storage medium for determining fractional flow reserve (FLR). The method includes: acquiring multiple initial vascular images; determining a vascular midline point, and the corresponding vascular cross-sectional area and flow coefficient, based on the multiple initial vascular images; and inputting the vascular midline point, the corresponding vascular cross-sectional area, and the flow coefficient into a pre-trained fractional flow reserve prediction model to obtain the FLR. This technical solution allows the fractional flow reserve prediction model to predict the FLR based on the vascular midline point and the corresponding vascular cross-sectional area and flow coefficient, effectively improving the accuracy of the FLR.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method, apparatus, electronic device, and storage medium for determining fractional blood flow reserve. Background Technology

[0002] Fractional flow reserve (FFR) is the ratio of the maximum blood flow that a coronary artery can obtain in the presence of stenosis to the maximum blood flow that a normal artery can obtain. It can be used to determine the severity of the disease.

[0003] Deep learning constructs neural networks and combines low-level features with more abstract high-level features or attribute features to detect the distributed representation features of data, thereby completing classification or regression tasks. Currently, research on predicting hemodynamics using deep learning remains very limited. The limitations of these studies include: 1) most studies focus on two-dimensional flow fields with limited application scope; 2) the sample resolution in the datasets is too low to represent complex flow field distributions and geometries.

[0004] In the process of realizing this invention, the inventors discovered that at least the following technical problems exist in the prior art: the above-mentioned prior art solutions have the problem of low accuracy of fractional blood flow reserve. Summary of the Invention

[0005] This invention provides a method, apparatus, electronic device, and storage medium for determining fractional flow reserve, in order to improve the accuracy of fractional flow reserve.

[0006] According to one aspect of the present invention, a method for determining fractional blood flow reserve is provided, comprising:

[0007] Acquire multiple initial vascular images;

[0008] The vessel midline point, the vessel cross-sectional area, and the flow coefficient corresponding to the vessel midline point are determined based on the multiple initial vessel images.

[0009] The blood vessel midline point, along with the corresponding blood vessel cross-sectional area and flow coefficient, are input into a pre-trained score prediction model to obtain the blood flow reserve score.

[0010] According to another aspect of the present invention, a fractional blood flow reserve determination device is provided, comprising:

[0011] The image acquisition module is used to acquire multiple initial blood vessel images;

[0012] The blood vessel midline point determination module is used to determine the blood vessel midline point, as well as the blood vessel cross-sectional area and flow coefficient corresponding to the blood vessel midline point, based on the multiple initial blood vessel images.

[0013] The blood flow reserve fraction determination module is used to input the blood vessel midline point, the corresponding blood vessel cross-sectional area and flow coefficient into a pre-trained fraction prediction model to obtain the blood flow reserve fraction.

[0014] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0015] At least one processor;

[0016] and a memory communicatively connected to the at least one processor;

[0017] The memory stores a computer program that can be executed by the at least one processor, which is then executed by the at least one processor to enable the at least one processor to perform the fractional blood flow reserve determination method according to any embodiment of the present invention.

[0018] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions for causing a processor to execute and implement the fractional blood flow reserve determination method according to any embodiment of the present invention.

[0019] The technical solution of this invention acquires multiple initial vascular images, then determines the vascular midline point, the corresponding vascular cross-sectional area, and the flow coefficient based on the multiple initial vascular images. The determined vascular midline point, along with the corresponding vascular cross-sectional area and flow coefficient, are then input into a pre-trained fractional prediction model to obtain the blood flow reserve fraction. Compared with existing technologies, the fractional prediction model of this embodiment can effectively learn the flow field distribution and geometric structure of the blood vessels, thereby improving the accuracy of the predicted blood flow reserve fraction.

[0020] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a flowchart of a method for determining fractional blood flow reserve according to Embodiment 1 of the present invention;

[0023] Figure 2 This is a flowchart of a method for determining fractional blood flow reserve according to Embodiment 2 of the present invention;

[0024] Figure 3 This is a flowchart of a method for determining a flow coefficient according to Embodiment 2 of the present invention;

[0025] Figure 4 This is a flowchart of a method for determining fractional blood flow reserve according to Embodiment 3 of the present invention;

[0026] Figure 5 This is a flowchart of the prediction process of a score prediction model provided in Embodiment 3 of the present invention;

[0027] Figure 6 This is a flowchart of a method for determining fractional blood flow reserve according to Embodiment 4 of the present invention;

[0028] Figure 7 This is a schematic diagram of a fractional blood flow reserve determination device according to Embodiment 5 of the present invention;

[0029] Figure 8 This is a schematic diagram of the structure of an electronic device that implements the blood flow reserve determination method of the present invention. Detailed Implementation

[0030] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0031] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0032] Example 1

[0033] Figure 1 This is a flowchart of a method for determining fractional flow reserve (FVR) according to Embodiment 1 of the present invention. This embodiment is applicable to the automatic determination of FVR. The method can be executed by a FVR determination device, which can be implemented in hardware and / or software and can be configured in a computer terminal. Figure 1 As shown, the method includes:

[0034] S110. Acquire multiple initial blood vessel images.

[0035] In this embodiment, the initial vascular image refers to the image to be segmented and midline extracted. For example, the initial vascular image can be medical imaging data, such as images in Digital Imaging and Communications in Medicine (DICOM) format, computed tomography (CT) images, magnetic resonance imaging (MRI) images, etc. Optionally, the initial vascular image can be an image containing coronary arteries.

[0036] Specifically, multiple consecutive initial vascular images can be obtained from the preset storage location of the electronic device, or multiple consecutive initial vascular images can be obtained from other devices connected to the electronic device or the cloud, without limitation.

[0037] S120. Determine the midline point of the blood vessel based on the multiple initial blood vessel images, as well as the cross-sectional area and flow coefficient of the blood vessel corresponding to the midline point.

[0038] In this embodiment, the vessel midline point is a point on the vessel midline, and the vessel midline may include multiple vessel midline points. The vessel cross-sectional area refers to the area of ​​the vessel cross-section corresponding to the vessel midline point, and the vessel cross-section is perpendicular to the vessel midline. The flow coefficient is an evaluation index of the flow rate at the vessel midline point.

[0039] Specifically, multiple initial vascular images can be used as input data to a pre-trained vascular segmentation model. The model can predict and output vascular segmentation results based on the initial images. Further, the vascular midline is extracted from the segmentation results, and the corresponding vascular cross-sectional area and flow coefficient are obtained. The vascular segmentation result can be a three-dimensional vascular segmentation image; in other words, the three-dimensional distribution of blood vessels can be viewed through the segmentation result. For example, the vascular segmentation result can be a three-dimensional coronary artery tree model, etc.

[0040] S130. Input the midline point of the blood vessel, the cross-sectional area of ​​the blood vessel corresponding to the midline point, and the flow coefficient into the pre-trained score prediction model to obtain the blood flow reserve score.

[0041] In this embodiment, the score prediction model is a pre-trained network prediction model that can be used to predict blood flow reserve score.

[0042] Specifically, multiple vessel midline points, along with the corresponding vessel cross-sectional area and flow coefficient, can be used as input data for the model. These data are then fed into a pre-trained score prediction model. The score prediction model can then predict the blood flow reserve score based on these multiple vessel midline points, the corresponding vessel cross-sectional area, and the flow coefficient, and output the result.

[0043] The technical solution of this invention acquires multiple initial vascular images, then determines the vascular midline point, the corresponding vascular cross-sectional area, and the flow coefficient based on the multiple initial vascular images. The determined vascular midline point, along with the corresponding vascular cross-sectional area and flow coefficient, are then input into a pre-trained fractional prediction model to obtain the blood flow reserve fraction. Compared with existing technologies, the fractional prediction model of this embodiment can effectively learn the flow field distribution and geometric structure of the blood vessels, thereby improving the accuracy of the predicted blood flow reserve fraction.

[0044] Example 2

[0045] Figure 2 This is a flowchart of a method for determining fractional blood flow reserve (FVRLR) according to Embodiment 2 of the present invention. The method of this embodiment can be combined with various optional schemes in the FVRLR determination methods provided in the above embodiments. The FVRLR determination method provided in this embodiment has been further optimized. Optionally, determining the vessel midline point, the vessel cross-sectional area and flow coefficient corresponding to the vessel midline point based on the multiple initial vessel images includes: inputting the multiple initial vessel images into a pre-trained vessel segmentation model to obtain vessel segmentation results; extracting the midline from the vessel segmentation results to obtain the vessel midline, wherein the vessel midline includes multiple vessel midline points; and determining the vessel cross-sectional area and flow coefficient corresponding to each vessel midline point.

[0046] like Figure 2 As shown, the method includes:

[0047] S210. Acquire multiple initial blood vessel images.

[0048] S220. Input the multiple initial blood vessel images into the pre-trained blood vessel segmentation model to obtain the blood vessel segmentation result.

[0049] Specifically, multiple initial blood vessel images can be used as input data for the model, which is then fed into a pre-trained blood vessel segmentation model. The blood vessel segmentation model can predict the blood vessel segmentation results based on the initial blood vessel images and output them.

[0050] The training steps of the blood vessel segmentation model include: acquiring an initial blood vessel sample image and a corresponding blood vessel segmentation annotation image; training an initial neural network model based on the initial blood vessel sample image and the corresponding blood vessel segmentation annotation image to obtain the blood vessel segmentation model.

[0051] For example, a blood vessel segmentation model can be pre-trained using a large number of initial blood vessel sample images. In the trained neural network model, features are extracted from the initial blood vessel sample images beforehand, and the model parameters in the neural network model are trained based on the extracted feature information. By continuously adjusting the model parameters, the distance deviation between the model's output and the blood vessel segmentation annotation image gradually decreases and tends to stabilize.

[0052] S230. The midline of the blood vessel segmentation result is extracted to obtain the blood vessel midline, wherein the blood vessel midline includes multiple blood vessel midline points.

[0053] The vessel midline refers to the central line image of a blood vessel, which can include multiple vessel midline points. Specifically, a series of morphological operations can be performed on the vessel segmentation results, and a smoothing algorithm can be used to smooth the image to obtain the vessel midline image.

[0054] S240. Determine the cross-sectional area and flow coefficient of the blood vessel corresponding to each midline point of the blood vessel.

[0055] In this embodiment, the cross-sectional area of ​​the blood vessel corresponding to the midline point of the blood vessel can be determined, the coronary inlet flow rate can be obtained, the coronary outlet flow rate can be determined based on the coronary inlet flow rate, and the flow coefficient corresponding to the midline point of the blood vessel can be determined based on the coronary outlet flow rate.

[0056] Among them, the coronary inlet flow can be the left coronary inlet flow or the right coronary inlet flow, which can include the blood flow distribution flow of the left coronary artery and the blood flow distribution flow of the right coronary artery. This coronary inlet flow can be obtained from the average value of the statistical population.

[0057] For example, Figure 3This is a flowchart illustrating a method for determining the flow coefficient provided in this embodiment. Specifically, the cross-sectional area of ​​the vessel corresponding to the vessel midline point can be obtained from the vessel segmentation results. Furthermore, the process of calculating the flow coefficient corresponding to the vessel midline point includes: allocating the inlet flow rates of the left and right coronary arteries, and then determining the coronary artery outlet flow rate based on Murray's law and the inlet flow rates of the left and right coronary arteries; furthermore, tracing back from the coronary artery terminal to the coronary artery inlet, calculating the flow rate of the unknown vessel using the known coronary artery outlet flow rate, thereby obtaining the total flow rate at the inlet, and then normalizing using the total flow rate at the inlet to obtain the flow coefficient corresponding to the vessel midline point in the coronary tree.

[0058] S250. Input the midline point of the blood vessel, the cross-sectional area of ​​the blood vessel corresponding to the midline point, and the flow coefficient into the pre-trained score prediction model to obtain the blood flow reserve score.

[0059] The technical solution of this invention involves inputting multiple initial vascular images into a pre-trained vascular segmentation model to obtain vascular segmentation results. Then, the midline of the vascular segmentation results is extracted to obtain the vascular midline. Subsequently, the cross-sectional area and flow coefficient of each vascular midline point are determined, thereby realizing the acquisition of vascular midline data. This provides accurate input data for model prediction and improves the accuracy of fractional flow reserve.

[0060] Example 3

[0061] Figure 4 This is a flowchart of a method for determining blood flow reserve fraction according to Embodiment 3 of the present invention. The method of this embodiment can be combined with various optional schemes in the blood flow reserve fraction determination methods provided in the above embodiments. The blood flow reserve fraction determination method provided in this embodiment has been further optimized. Optionally, the step of inputting the blood vessel midline point, and the blood vessel cross-sectional area and flow coefficient corresponding to the blood vessel midline point into a pre-trained fraction prediction model to obtain the blood flow reserve fraction includes: performing sparse mapping on the blood vessel midline point, and the blood vessel cross-sectional area and flow coefficient corresponding to the blood vessel midline point to obtain blood vessel sparse mapping features; inputting the blood vessel sparse mapping features into the sparse segmentation network to obtain blood vessel sparse segmentation features; performing inverse sparse mapping on the blood vessel sparse segmentation features to obtain blood vessel original image spatial features; and inputting the blood vessel original image spatial features into a recurrent neural network to obtain the blood flow reserve fraction.

[0062] like Figure 4 As shown, the method includes:

[0063] S310. Acquire multiple initial blood vessel images.

[0064] S320. Determine the midline point of the blood vessel based on the multiple initial blood vessel images, as well as the cross-sectional area and flow coefficient of the blood vessel corresponding to the midline point.

[0065] S330. Perform sparsification mapping on the midline point of the blood vessel, as well as the cross-sectional area and flow coefficient of the blood vessel corresponding to the midline point, to obtain the sparse mapping feature of the blood vessel.

[0066] In this embodiment, sparsity mapping can be achieved through a hash mapping table to obtain vascular sparsity mapping features.

[0067] For example, the coordinates of the midline point of the blood vessel, the cross-sectional area of ​​the blood vessel corresponding to the midline point, and the flow coefficient corresponding to the midline point of the blood vessel are sparsely mapped using a pre-established hash mapping table.

[0068] S340. Input the sparse mapping features of blood vessels into the sparse segmentation network to obtain sparse segmentation features of blood vessels.

[0069] Among them, sparse segmentation network refers to a neural network model used to process the sparse mapping features of blood vessels. Sparse segmentation network can be used to learn the flow field distribution and geometry of blood vessels.

[0070] For example, the sparse segmentation network can be a sparse convolutional network. Specifically, the sparse mapping features of blood vessels are input into the sparse convolutional network to obtain sparse segmentation features of blood vessels. It should be noted that sparse convolutional networks can maintain high resolution of coronary artery images, thereby characterizing the flow field distribution and geometry of coronary artery models at high resolution. Furthermore, due to the sparsity of sparse convolutional networks, the computational cost of the model can be reduced.

[0071] S350. Perform inverse sparsification mapping on the sparse segmentation features of the blood vessels to obtain the spatial features of the original blood vessel image.

[0072] Specifically, inverse sparsification mapping can be achieved through a hash mapping table, which maps the sparse segmentation features of blood vessels back to the original image space, thereby obtaining the original image space features of blood vessels.

[0073] S360. Input the spatial features of the original vascular image into a recurrent neural network to obtain the blood flow reserve fraction.

[0074] In this embodiment, a recurrent neural network can be used to capture the dependencies between midline points of various blood vessels.

[0075] For example, a recurrent neural network can be a Long Short-Term Memory (LSTM) network. Specifically, the spatial features of the original vascular map can be input into the LSTM network to obtain the blood flow reserve fraction.

[0076] It should be noted that Long Short-Term Memory (LSTM) networks can control the transmission state through gating, thereby remembering information that needs to be remembered for a long time and forgetting unimportant information. This allows them to be used to process data with long, variable sequences. In the fractional flow reserve (FVR) prediction scenario of this embodiment, the FVRs between each midline point are correlated. Furthermore, during recurrent neural network (RNN) training, the loss weights for stenotic vessels are increased, making the RNN pay more attention to changes in FVR at the stenosis.

[0077] Figure 5 This is a flowchart illustrating the prediction process of a score prediction model provided in this embodiment. Specifically, after extracting the coronary artery midline, the coordinates of the vessel midline points, as well as the corresponding vessel cross-sectional area and flow coefficient, are sparsely mapped to obtain vessel sparse mapping features. These features are then input into a sparse segmentation network to obtain vessel sparse segmentation features. Inverse sparse mapping is performed on the vessel sparse segmentation features to obtain the original vessel image spatial features. Finally, these features are input into an LSTM network to obtain the blood flow reserve score.

[0078] Based on the above embodiments, optionally, the training process of the score prediction model includes: acquiring blood vessel midline sample data and the blood flow reserve target score corresponding to the blood vessel midline sample data, wherein the blood vessel midline sample data includes blood vessel midline sample points, and the blood vessel cross-sectional area and flow coefficient corresponding to the blood vessel midline sample points; training an initial neural network model based on the blood vessel midline sample data and the blood flow reserve target score corresponding to the blood vessel midline sample data to obtain the score prediction model.

[0079] For example, during the training of the score prediction model, features are extracted from the blood vessel midline sample data in advance, and the model parameters in the neural network model are trained based on the extracted feature information. By continuously adjusting the model parameters, the distance deviation between the model's output and the target blood flow reserve score gradually decreases and tends to stabilize. The loss function of the model can be the mean square error (MSE) loss function. The MSE loss function can be used to calculate the loss between the predicted blood flow reserve score and the target blood flow reserve score at each midline point, and then backpropagate to update the network parameters.

[0080] In some embodiments, after obtaining the fractional flow reserve, the coronary artery can be rendered based on the fractional flow reserve and the distance from each midline point to the coronary artery surface to obtain coronary hemodynamic results.

[0081] The technical solution of this invention involves acquiring multiple initial vascular images, determining the vascular midline points, corresponding vascular cross-sectional areas, and flow coefficients based on these images, performing sparse mapping on the vascular midline points and their corresponding cross-sectional areas and flow coefficients to obtain vascular sparse mapping features, inputting these features into a sparse segmentation network to obtain vascular sparse segmentation features, and then performing inverse sparse mapping on these features to obtain the original vascular image spatial features. These features are then input into a recurrent neural network to obtain the blood flow reserve fraction. Compared with existing technologies, the fraction prediction model of this embodiment can effectively learn the flow field distribution and geometric structure of blood vessels, capturing the order dependencies between midline points, thereby improving the accuracy of the predicted blood flow reserve fraction.

[0082] Example 4

[0083] Figure 6 This is a flowchart of a method for determining blood flow reserve fraction according to Embodiment 4 of the present invention. The method of this embodiment can be combined with various optional schemes in the blood flow reserve fraction determination methods provided in the above embodiments. The blood flow reserve fraction determination method provided in this embodiment has been further optimized. Optionally, the fraction prediction model includes a sparse point cloud segmentation model and a recurrent neural network; correspondingly, the step of inputting the vessel midline point, and the vessel cross-sectional area and flow coefficient corresponding to the vessel midline point into the pre-trained fraction prediction model to obtain the blood flow reserve fraction includes: performing sparse mapping on the vessel midline point, and the vessel cross-sectional area and flow coefficient corresponding to the vessel midline point to obtain vessel sparse mapping features; inputting the vessel sparse mapping features into the sparse point cloud segmentation model to obtain vessel sparse segmentation features; performing inverse sparse mapping on the vessel sparse segmentation features to obtain vessel original image spatial features; and inputting the vessel original image spatial features into the recurrent neural network to obtain the blood flow reserve fraction.

[0084] like Figure 6 As shown, the method includes:

[0085] S410: Acquire multiple initial blood vessel images.

[0086] S420. Determine the midline point of the blood vessel based on the multiple initial blood vessel images, as well as the cross-sectional area and flow coefficient of the blood vessel corresponding to the midline point.

[0087] S430. Perform sparsification mapping on the midline point of the blood vessel, as well as the cross-sectional area and flow coefficient of the blood vessel corresponding to the midline point, to obtain the sparse mapping feature of the blood vessel.

[0088] S440. Input the sparse mapping features of blood vessels into the sparse point cloud segmentation model to obtain sparse segmentation features of blood vessels.

[0089] S450. Perform inverse sparsification mapping on the sparse segmentation features of the blood vessels to obtain the spatial features of the original blood vessel image.

[0090] S460. Input the spatial features of the original vascular image into a recurrent neural network to obtain the blood flow reserve fraction.

[0091] In this embodiment, the sparse point cloud segmentation model can be used to segment the sparse mapping features of blood vessels. It can be obtained by training based on the sparse mapping sample features of blood vessels and the sparse segmentation sample features of blood vessels corresponding to the sparse mapping sample features of blood vessels.

[0092] The technical solution of this invention involves acquiring multiple initial vascular images, determining the vascular midline points, corresponding vascular cross-sectional areas, and flow coefficients based on these images, performing sparse mapping on the vascular midline points and their corresponding cross-sectional areas and flow coefficients to obtain vascular sparse mapping features, inputting these features into a sparse point cloud segmentation model to obtain vascular sparse segmentation features, and then performing inverse sparse mapping on these features to obtain the original vascular image spatial features. These features are then input into a recurrent neural network to obtain the blood flow reserve fraction. Compared with existing technologies, the fraction prediction model of this embodiment can effectively learn the flow field distribution and geometric structure of the blood vessels, as well as capture the sequential dependencies between midline points, thereby improving the accuracy of the predicted blood flow reserve fraction.

[0093] Example 5

[0094] Figure 7 This is a schematic diagram of a fractional blood flow reserve determination device provided in Embodiment 5 of the present invention. Figure 7 As shown, the device includes:

[0095] Image acquisition module 510 is used to acquire multiple initial blood vessel images;

[0096] The vessel midline point determination module 520 is used to determine the vessel midline point, the vessel cross-sectional area and flow coefficient corresponding to the vessel midline point based on the multiple initial vessel images;

[0097] The blood flow reserve fraction determination module 530 is used to input the blood vessel midline point, as well as the blood vessel cross-sectional area and flow coefficient corresponding to the blood vessel midline point, into a pre-trained fraction prediction model to obtain the blood flow reserve fraction.

[0098] The technical solution of this invention acquires multiple initial vascular images, then determines the vascular midline point, the corresponding vascular cross-sectional area, and the flow coefficient based on the multiple initial vascular images. The determined vascular midline point, along with the corresponding vascular cross-sectional area and flow coefficient, are then input into a pre-trained fractional prediction model to obtain the blood flow reserve fraction. Compared with existing technologies, the fractional prediction model of this embodiment can effectively learn the flow field distribution and geometric structure of the blood vessels, thereby improving the accuracy of the predicted blood flow reserve fraction.

[0099] In some optional implementations, the vessel midline point determination module 520 includes:

[0100] The blood vessel segmentation result determination unit is used to input the multiple initial blood vessel images into a pre-trained blood vessel segmentation model to obtain blood vessel segmentation results;

[0101] A vessel midline extraction unit is used to extract the midline from the vessel segmentation results to obtain the vessel midline, wherein the vessel midline includes multiple vessel midline points;

[0102] The midline information determination unit is used to determine the cross-sectional area and flow coefficient of the blood vessel corresponding to each midline point of the blood vessel.

[0103] In some optional implementations, the centerline information determination unit is specifically used for:

[0104] Determine the cross-sectional area of ​​the blood vessel corresponding to the midline point of the blood vessel;

[0105] Obtain the coronary inlet flow rate and determine the coronary outlet flow rate based on the coronary inlet flow rate;

[0106] The flow coefficient corresponding to the midline point of the blood vessel is determined based on the coronary artery outlet flow rate.

[0107] In some alternative implementations, the score prediction model includes a sparse segmentation network and a recurrent neural network, wherein the sparse segmentation network is used to learn the flow field distribution and geometry of the blood vessels, and the recurrent neural network is used to capture the dependencies between the midline points of each of the blood vessels.

[0108] In some optional implementations, the fractional flow reserve determination module 530 is specifically used for:

[0109] The blood vessel midline point, as well as the corresponding blood vessel cross-sectional area and flow coefficient, are subjected to sparsification mapping to obtain the blood vessel sparsification mapping features.

[0110] The sparse mapping features of blood vessels are input into the sparse segmentation network to obtain sparse segmentation features of blood vessels.

[0111] The sparse segmentation features of the blood vessels are inversely sparsified to obtain the spatial features of the original blood vessel image.

[0112] The spatial features of the original vascular image are input into a recurrent neural network to obtain the fractional blood flow reserve.

[0113] In some optional implementations, the training process of the score prediction model includes:

[0114] Obtain blood vessel midline sample data and the target blood flow reserve score corresponding to the blood vessel midline sample data, wherein the blood vessel midline sample data includes blood vessel midline sample points, as well as the blood vessel cross-sectional area and flow coefficient corresponding to the blood vessel midline sample points;

[0115] The initial neural network model is trained based on the blood vessel midline sample data and the corresponding blood flow reserve target score to obtain the score prediction model.

[0116] In some optional implementations, the score prediction model includes a sparse point cloud segmentation model and a recurrent neural network; the blood flow reserve score determination module 530 is further configured to:

[0117] The blood vessel midline point, as well as the corresponding blood vessel cross-sectional area and flow coefficient, are subjected to sparsification mapping to obtain the blood vessel sparsification mapping features.

[0118] The sparse mapping features of blood vessels are input into the sparse point cloud segmentation model to obtain sparse segmentation features of blood vessels.

[0119] The sparse segmentation features of the blood vessels are inversely sparsified to obtain the spatial features of the original blood vessel image.

[0120] The spatial features of the original vascular image are input into a recurrent neural network to obtain the fractional blood flow reserve.

[0121] The fractional flow reserve determination device provided in the embodiments of the present invention can execute the fractional flow reserve determination method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of executing the method.

[0122] Example 6

[0123] Figure 8A schematic diagram of an electronic device 10 that can be used to implement embodiments of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0124] like Figure 8 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded into the RAM 13 from storage unit 18. The RAM 13 can also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An I / O interface 15 is also connected to the bus 14.

[0125] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0126] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as a fractional blood flow reserve determination method, which includes:

[0127] Acquire multiple initial vascular images;

[0128] The vessel midline point, the vessel cross-sectional area, and the flow coefficient corresponding to the vessel midline point are determined based on the multiple initial vessel images.

[0129] The blood vessel midline point, along with the corresponding blood vessel cross-sectional area and flow coefficient, are input into a pre-trained score prediction model to obtain the blood flow reserve score.

[0130] In some embodiments, the fractional flow reserve determination method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or mounted on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the fractional flow reserve determination method described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform the fractional flow reserve determination method by any other suitable means (e.g., by means of firmware).

[0131] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0132] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0133] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0134] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0135] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or computing systems that include middleware components (e.g., application servers), or computing systems that include frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0136] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0137] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0138] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method of blood flow reserve fraction determination, characterized by, The method comprises the following steps: obtaining multiple initial blood vessel images; determining blood vessel centerline points, and blood vessel cross-sectional areas and flow coefficients corresponding to the blood vessel centerline points based on the multiple initial blood vessel images; inputting the blood vessel centerline points, and the blood vessel cross-sectional areas and flow coefficients corresponding to the blood vessel centerline points into a pre-trained score prediction model to obtain a blood flow reserve score; the score prediction model comprises a sparse segmentation network and a recurrent neural network; wherein the inputting the blood vessel centerline points, and the blood vessel cross-sectional areas and flow coefficients corresponding to the blood vessel centerline points into the pre-trained score prediction model to obtain the blood flow reserve score comprises: performing sparse mapping on the blood vessel centerline points, and the blood vessel cross-sectional areas and flow coefficients corresponding to the blood vessel centerline points to obtain blood vessel sparse mapping features; inputting the blood vessel sparse mapping features into the sparse segmentation network to obtain blood vessel sparse segmentation features; performing inverse sparse mapping on the blood vessel sparse segmentation features to obtain blood vessel original graph space features; inputting the blood vessel original graph space features into the recurrent neural network to obtain the blood flow reserve score.

2. The method of claim 1, wherein, The determining the blood vessel centerline points, and the blood vessel cross-sectional areas and flow coefficients corresponding to the blood vessel centerline points based on the multiple initial blood vessel images comprises: inputting the multiple initial blood vessel images into a pre-trained blood vessel segmentation model to obtain blood vessel segmentation results; performing centerline extraction on the blood vessel segmentation results to obtain blood vessel centerlines, wherein the blood vessel centerlines comprise multiple blood vessel centerline points; determining blood vessel cross-sectional areas and flow coefficients corresponding to the blood vessel centerline points.

3. The method of claim 2, wherein, The determining the blood vessel cross-sectional areas and flow coefficients corresponding to the blood vessel centerline points comprises: determining blood vessel cross-sectional areas corresponding to the blood vessel centerline points; obtaining coronary artery inlet flow, and determining coronary artery outlet flow based on the coronary artery inlet flow; determining flow coefficients corresponding to the blood vessel centerline points based on the coronary artery outlet flow.

4. The method of claim 1, wherein, The sparse segmentation network is used for learning flow field distribution and geometric shape of blood vessels, and the recurrent neural network is used for capturing dependency relationship between the blood vessel centerline points.

5. The method of claim 1, wherein, The training process of the score prediction model comprises: obtaining blood vessel centerline sample data and blood flow reserve target scores corresponding to the blood vessel centerline sample data, wherein the blood vessel centerline sample data comprises blood vessel centerline sample points, and blood vessel cross-sectional areas and flow coefficients corresponding to the blood vessel centerline sample points; training an initial neural network model based on the blood vessel centerline sample data and the blood flow reserve target scores corresponding to the blood vessel centerline sample data to obtain the score prediction model.

6. The method of claim 1, wherein, The score prediction model comprises a sparse point cloud segmentation model and a recurrent neural network; correspondingly, the inputting the blood vessel centerline points, and the blood vessel cross-sectional areas and flow coefficients corresponding to the blood vessel centerline points into the pre-trained score prediction model to obtain the blood flow reserve score comprises: performing sparse mapping on the blood vessel centerline points, and the blood vessel cross-sectional areas and flow coefficients corresponding to the blood vessel centerline points to obtain blood vessel sparse mapping features; inputting the blood vessel sparse mapping features into the sparse point cloud segmentation model to obtain blood vessel sparse segmentation features; The blood vessel sparse segmentation feature is inversely sparse mapping to obtain a blood vessel original graph space feature; The blood vessel original graph space feature is input into a recurrent neural network to obtain a blood flow reserve score.

7. A blood flow reserve fraction determination apparatus characterized by comprising: The method comprises the following steps: An image acquisition module is configured to acquire a plurality of initial blood vessel images; A blood vessel centerline point determination module is configured to determine a blood vessel centerline point based on the plurality of initial blood vessel images, and a blood vessel cross-sectional area and a flow coefficient corresponding to the blood vessel centerline point; A blood flow reserve score determination module is configured to input the blood vessel centerline point, the blood vessel cross-sectional area and the flow coefficient corresponding to the blood vessel centerline point into a pre-trained score prediction model to obtain a blood flow reserve score; the score prediction model comprises a sparse segmentation network and a recurrent neural network; The blood flow reserve score determination module is specifically configured to perform sparse mapping on the blood vessel centerline point, the blood vessel cross-sectional area and the flow coefficient corresponding to the blood vessel centerline point to obtain blood vessel sparse mapping features; and input the blood vessel sparse mapping features into the sparse segmentation network to obtain blood vessel sparse segmentation features. The blood vessel sparse segmentation features are inversely sparse mapping to obtain a blood vessel original graph space feature; and the blood vessel original graph space feature is input into a recurrent neural network to obtain a blood flow reserve score.

8. An electronic device, comprising: The electronic device comprises: at least one processor; and a memory connected in communication with the at least one processor; The memory stores a computer program that can be executed by the at least one processor; and the computer program is executed by the at least one processor to enable the at least one processor to execute the blood flow reserve score determination method of any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer readable storage medium stores computer instructions for enabling the processor to execute the blood flow reserve score determination method of any one of claims 1-6 when executed.

Images