A method, system, electronic device, and storage medium of determining microcatheter shape

Abstract

The application discloses a method for determining microcatheter shape, comprising: determining a plurality of different microcatheter shapes; and evaluating each microcatheter shape based on a pre-trained neural network model to determine a suitable microcatheter shape. The above technical solution can improve the accuracy of microcatheter shaping and reduce the time of microcatheter shaping. The application also discloses a system for determining microcatheter shape, an electronic device and a storage medium.

Description

Technical Field

[0001] This invention relates to the field of medical device technology, and in particular to a method, system, electronic device and storage medium for determining the morphology of microcatheters. Background Technology

[0002] Cerebral aneurysms constitute a significant portion of interventional procedures. A cerebral aneurysm is a tumor-like protrusion resulting from abnormal changes in blood vessels, and it is the leading cause of subarachnoid hemorrhage, which in turn is a major cause of death among cerebrovascular disease patients. Subarachnoid hemorrhage can lead to various pathological changes such as cerebral vascular stenosis, cerebral edema, cerebral infarction, and hydrocephalus, with high rates of disability and mortality. Early diagnosis and effective assessment of aneurysms are crucial for their treatment.

[0003] Treatment of cerebral aneurysms primarily relies on surgical methods, such as coil embolization. This involves selectively delivering a microcatheter into the aneurysm. Coils are then introduced into the aneurysm through the microcatheter and used to densely embolize it. A crucial aspect of this procedure is the successful shaping of the microcatheter. Proper shaping of the microcatheter tip significantly improves the accuracy of placement, stability, and maneuverability during the interventional procedure.

[0004] During interventional aneurysm treatment, variability in aneurysm identification and measurement leads to significant discrepancies in the measurement of the same indicator for the same aneurysm by different doctors, resulting in poor accuracy and repeatability, thus affecting treatment outcomes. All measurements are performed manually by doctors, increasing their workload and reducing their efficiency. Furthermore, all vascular interventional procedures are chosen based on the doctor's subjective experience, without evaluation of the surgical results.

[0005] The pre-shaping of the microcatheter is crucial for the success of interventional surgery for cerebral aneurysms and for ensuring the microcatheter successfully enters the aneurysm. Currently, even among experienced surgeons in top hospitals, only about 70% of those who have successfully shaped the microcatheter into the aneurysm after a single attempt. If the microcatheter fails to enter the aneurysm after a single shaping attempt, it can cause interventional obstacles during the surgery, preventing the rapid filling of the aneurysm with embolic material. In such cases, the microcatheter must be removed from the cerebral artery, replaced, and reshaped, with repeated attempts to insert it into the aneurysm. This inevitably prolongs the surgical time and introduces numerous problems associated with repeated microcatheter shaping.

[0006] Therefore, determining the appropriate microcatheter morphology to shorten operation time, reduce surgical complications, and improve patient prognosis is a challenge that needs to be overcome at this stage. Summary of the Invention

[0007] The purpose of this invention is to solve the technical problem of low accuracy in microcatheter shaping. This invention provides a method for determining microcatheter morphology. Based on a pre-trained neural network model, it can evaluate the morphology of multiple microcatheters and determine the appropriate morphology, improving the accuracy of matching microcatheter morphology with the carrier artery and aneurysm, reducing the time required for microcatheter shaping, and providing effective data support for surgery.

[0008] To address the aforementioned technical problems, embodiments of the present invention disclose a method for determining the morphology of a microcatheter, comprising:

[0009] Several different microcatheter morphologies are identified; each microcatheter morphology is evaluated based on a pre-trained neural network model to determine the appropriate microcatheter morphology.

[0010] By adopting the above technical solution, the accuracy of microcatheter shaping can be improved and the shaping time of microcatheter can be reduced.

[0011] According to another specific embodiment of the present invention, an embodiment of the present invention discloses a method for determining the morphology of a microcatheter, the step of determining several different microcatheter morphologies includes: determining an initial morphology of the microcatheter; and determining several different microcatheter morphologies based on the initial morphology of the microcatheter.

[0012] According to another specific embodiment of the present invention, an embodiment of the present invention discloses a method for determining the morphology of a microcatheter. The steps of determining the initial morphology of the microcatheter include: acquiring an image including an aneurysm and a parent artery; determining the location of the aneurysm based on the image; segmenting the aneurysm and the parent artery based on the location of the aneurysm to determine a mask image of the aneurysm and the parent artery; determining the centroid P of the aneurysm based on the mask image of the aneurysm, and extracting the centerline of the parent artery based on the mask image of the parent artery, and determining the starting point O of the centerline; drawing a perpendicular line from the centroid P to the centerline, wherein the intersection of the perpendicular line and the centerline is B; and establishing the initial morphology OBP of the microcatheter based on the perpendicular line segment BP and the centerline.

[0013] According to another specific embodiment of the present invention, an embodiment of the present invention discloses a method for determining the morphology of a microcatheter, comprising the steps of determining several different microcatheter morphologies based on the initial morphology of the microcatheter, including: extracting a discrete point set of the initial morphology of the microcatheter; randomly perturbing at least one discrete point in the discrete point set to determine an updated discrete point set; fitting the updated discrete point set to determine the updated microcatheter morphology; and repeating the random perturbation and fitting several times to determine several different microcatheter morphologies.

[0014] According to another specific embodiment of the present invention, an embodiment of the present invention discloses a method for determining the morphology of a microcatheter, wherein discrete points are located on a mask image of the carrier artery or aneurysm.

[0015] According to another specific embodiment of the present invention, an embodiment of the present invention discloses a method for determining the morphology of a microcatheter. The method involves evaluating each microcatheter morphology based on a pre-trained neural network model to determine a suitable microcatheter morphology. The method includes: acquiring training sample data, which includes multiple microcatheter sample data, wherein the multiple microcatheter sample data are divided into suitable and unsuitable categories; determining the predicted labels of the training sample data based on the training sample data and mask images including aneurysms and aneurysm-bearing arteries; calculating the function value of a loss function based on the predicted labels and the labels of the training sample data; determining the gradient of the initial neural network model based on the function value of the loss function, performing backpropagation and continuously updating the weight parameters of the initial neural network model until the function value of the loss function converges to a stable state, thereby determining the pre-trained neural network model; and evaluating each microcatheter morphology based on the pre-trained neural network model to determine a suitable microcatheter morphology.

[0016] According to another specific embodiment of the present invention, an embodiment of the present invention discloses a method for determining the morphology of a microcatheter, comprising the steps of determining the predicted label of the training sample data based on training sample data and a mask image including an aneurysm and aneurysm-bearing artery, and an initial neural network model, including: extracting features of the mask image including the aneurysm and aneurysm-bearing artery and the training sample data based on the initial neural network model; processing the extracted features by the fully connected layer of the initial neural network model and outputting a feature vector; and determining the predicted label of the training sample data based on the feature vector by the softmax layer of the initial neural network model.

[0017] According to another specific embodiment of the present invention, an embodiment of the present invention discloses a method for determining the morphology of a microcatheter, wherein the method for calculating the function value of the loss function includes a cross-entropy calculation method or a mean-variance calculation method.

[0018] The present invention also discloses a system for determining the morphology of microcatheters, comprising: a generator that determines several different microcatheter morphologies; and a discriminator that evaluates each microcatheter morphology based on a pre-trained neural network model to determine a suitable microcatheter morphology.

[0019] By adopting the above technical solution, based on a pre-trained neural network model, the morphology of multiple microcatheters can be evaluated, and the appropriate microcatheter morphology can be determined. This improves the accuracy of matching the microcatheter morphology with the carrier artery and aneurysm, reduces the time for microcatheter morphology shaping, and provides effective data support for surgery.

[0020] The present invention also discloses an electronic device, including a memory, a processor, and a computer program stored in the memory that can run on the processor. When the processor executes the computer program, it implements the method for determining the microcatheter morphology in any of the above embodiments.

[0021] The present invention also discloses a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the method for determining the morphology of microcatheters in any of the above embodiments. Attached Figure Description

[0022] Figure 1 This invention illustrates a flowchart of the method for determining the morphology of microcatheters. Figure 1 ;

[0023] Figure 2 This invention illustrates a flowchart of the method for determining the morphology of microcatheters. Figure 2 ;

[0024] Figure 3 This invention illustrates a flowchart of the method for determining the morphology of microcatheters. Figure 3 ;

[0025] Figure 4 This invention illustrates a flowchart of the method for determining the morphology of microcatheters. Figure 4 ;

[0026] Figure 5 This invention illustrates a flowchart of the method for determining the morphology of microcatheters. Figure 5 ;

[0027] Figure 6 This invention illustrates a flowchart of the method for determining the morphology of microcatheters. Figure 6 ;

[0028] Figure 7 This invention illustrates a structural block diagram of the system for determining the morphology of microcatheters.

[0029] Figure 8 This diagram shows the structural block diagram of the electronic device in this invention;

[0030] Figure 9 A schematic diagram of the initial morphology (OBP) of the microcatheter in this invention is shown;

[0031] Figure 10 This invention includes a mask image of the tumor-bearing artery and the aneurysm. Detailed Implementation

[0032] The following specific embodiments illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. Although the description of the present invention is presented in conjunction with preferred embodiments, this does not mean that the features of the invention are limited to these embodiments. On the contrary, the purpose of describing the invention in conjunction with embodiments is to cover other options or modifications that may be derived based on the claims of the present invention. To provide a deep understanding of the invention, many specific details will be included in the following description. The invention may also be implemented without using these details. Furthermore, to avoid confusion or obscuring the focus of the invention, some specific details will be omitted in the description. It should be noted that, unless otherwise specified, the embodiments and features in the embodiments of the present invention can be combined with each other.

[0033] It should be noted that in this specification, similar reference numerals and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0034] In the description of this embodiment, it should be noted that the terms "upper", "lower", "inner", "bottom", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, or the orientation or positional relationship that the product of the invention is usually placed in during use. They are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limiting the present invention.

[0035] In the description of this embodiment, it should also be noted that, unless otherwise explicitly specified and limited, the terms "set up," "connected," and "linked" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this embodiment based on the specific circumstances.

[0036] To make the objectives, technical solutions, and advantages of the present invention clearer, the embodiments of the present invention will be described in further detail below with reference to the accompanying drawings.

[0037] Cerebral aneurysms constitute a significant portion of interventional procedures. A cerebral aneurysm is a tumor-like protrusion resulting from abnormal changes in blood vessels, and it is the leading cause of subarachnoid hemorrhage, which in turn is a major cause of death in cerebrovascular disease patients. Subarachnoid hemorrhage can lead to various pathological changes such as cerebral vascular stenosis, cerebral edema, cerebral infarction, and hydrocephalus, resulting in high rates of disability and mortality. Therefore, early diagnosis and effective assessment of aneurysms are crucial for their treatment.

[0038] Treatment of cerebral aneurysms primarily relies on surgical methods, such as coil embolization. To introduce the coil, a suitable microcatheter must first be selectively delivered into the aneurysm. The coil is then introduced through the microcatheter to densely embolize the aneurysm. Therefore, successful microcatheter shaping is a crucial aspect of this procedure. Microcatheter shaping, or the microcatheter morphology mentioned in this article, refers specifically to the shaping or morphology of the microcatheter tip.

[0039] The method, system, electronic device, and storage medium for determining the morphology of microcatheters provided by this invention are all for non-therapeutic purposes.

[0040] The method for determining the morphology of microcatheters provided by this invention directly targets preoperative images.

[0041] To facilitate further understanding, the technical terms used in this article are explained below:

[0042] (1) Loss Function

[0043] In training a deep neural network, to ensure the output closely approximates the desired predicted value, we compare the network's prediction with the target value. Based on the difference, we update the weight vector of each layer (typically pre-configured before the initial update). For example, if the prediction is too high, the weight vector is adjusted to predict a lower value. This adjustment continues until the deep neural network predicts the target value accurately or very closely. Therefore, we need to predefine "how to compare the difference between the predicted and target values," which is the loss function. These are crucial equations used to measure the difference between the predicted and target values. For example, a higher loss function output indicates a greater difference, making deep neural network training a process of minimizing this loss.

[0044] (2) Neural Network Model

[0045] The neural network model refers to any neural network model in the related technology. In any of the following embodiments, the neural network model refers to a neural network model used for machine learning, as long as it can serve the purpose of training and prediction using data samples. They will not be listed one by one here.

[0046] (3) Fully Connected Layers (FC)

[0047] A fully connected layer is a layer in which every node is connected to all nodes in the previous layer, and it is used to synthesize features extracted from other layers.

[0048] (4) softmax layer

[0049] The softmax layer can be understood as a module that normalizes the prediction results in the fully connected layer. The output value of softmax is a value in the range [0,1].

[0050] (5) Mask

[0051] A mask is a binary image used to represent a region of interest in an image (e.g., an image region representing a specific organ and / or tissue in a medical image). The parts of the image that belong to the region of interest and those that do not can have different pixel values ​​(e.g., 0, 1).

[0052] refer to Figure 1 This invention proposes a method for determining the morphology of microcatheters, comprising:

[0053] Step S1: Determine several different microcatheter morphologies;

[0054] Step S2: Based on the pre-trained neural network model, evaluate each microcatheter morphology to determine the appropriate microcatheter morphology.

[0055] The method for determining microcatheter morphology according to the present invention, based on a pre-trained neural network model, can evaluate the morphology of multiple microcatheters and determine the appropriate microcatheter morphology, thereby improving the accuracy of matching microcatheter morphology with the carrier artery and aneurysm, reducing the time for microcatheter morphology shaping, and providing effective data support for surgery.

[0056] In some embodiments, reference Figure 1 and combined Figure 2 Step S1, determining several different microcatheter morphologies, may further include the following steps:

[0057] Step S11: Determine the initial morphology of the microcatheter;

[0058] Step S12: Based on the initial morphology of the microcatheter, determine several different microcatheter morphologies.

[0059] In some embodiments, reference Figure 2 and combined Figure 3 , Figure 10 Step S11, determining the initial morphology of the microcatheter, including:

[0060] Step S111: Acquire images including the aneurysm and the parent artery;

[0061] Specifically, three-dimensional digital subtraction angiography (3D-DSA) imaging technology can be used to obtain images of aneurysms and their parent arteries. The basic principle of DSA is X-ray imaging after injecting contrast agent into the examined area to acquire images. 3D-DSA is a new angiography technique that transmits two-dimensional images acquired by rotating angiography to a computer workstation, where various processing software performs three-dimensional reconstruction of the vascular structure. It combines X-ray tube rotation technology, digital subtraction angiography technology, and three-dimensional reconstruction technology, clearly displaying the anatomical structure of the aneurysm and the adjacent relationships of the parent artery, effectively avoiding image overlap or obscuring.

[0062] Step S112: Determine the location of the aneurysm based on the images;

[0063] Identify the target location region of the aneurysm from 3D-DSA images.

[0064] Step S113: Based on the location of the aneurysm, segment the images of the aneurysm and the parent artery to determine the mask images of the aneurysm and the parent artery.

[0065] In other words, based on the target location region of the aneurysm, the images of the aneurysm and its parent artery are segmented. The parent artery is ultimately labeled "1", the aneurysm "2", and the background "0", resulting in a segmented mask image. Regarding the segmentation method, it can be done manually based on the operator's experience, such as using open-source medical image processing software like itk-snap or 3DSLICER to separately determine the mask images of the aneurysm and its parent artery. Alternatively, it can be automated using software purchased or developed internally or through other channels. If the automated segmentation results are inaccurate, manual intervention can be used to correct them, achieving semi-automatic segmentation. The aforementioned segmentation algorithms include traditional segmentation algorithms, such as region-based watershed and region growing algorithms, and graph theory-based maximum flow and minimum cut algorithms, to segment aneurysms and their parent arteries, and determine mask images for each. Deep learning algorithms, such as UNET, VNet, and NNUNet, can also be used for aneurysm and parent artery segmentation, or traditional pattern recognition combined with conditional random fields can be used to achieve aneurysm and parent artery segmentation, and determine mask images for each. In this application, no specific limitation is made to the segmentation algorithm, as long as it can achieve image segmentation.

[0066] Step S114: Based on the mask image of the aneurysm, determine the centroid P of the aneurysm, and based on the mask image of the carrier artery, extract the centerline of the carrier artery and determine the starting point O of the centerline.

[0067] The centroid of the aneurysm represents the geometric center of the aneurysm, and the centroid can be determined by calculating the centroid of the surface pixels of the mask image of the segmented aneurysm.

[0068] In practice, the confirmation of the starting point O is usually done by manual marking.

[0069] The centerline of the tumor-bearing artery is extracted based on the mask image of the artery. This centerline extraction can be based on traditional algorithms such as distance transform, minimum path, and morphological processing, or it can be implemented using a Venn diagram, or alternatively, deep learning-based methods. It should be noted that the methods for centerline extraction are not limited to those mentioned above.

[0070] Step S115: Draw a perpendicular line from the centroid P to the center line, where the intersection of the perpendicular line and the center line is B;

[0071] Step S116: Based on the vertical segment BP and the centerline, establish the initial morphology OBP of the microcatheter.

[0072] Based on the centerline and perpendicular line generated in the above steps, the initial microcatheter morphology can be determined. The initial microcatheter morphology OBP can be referenced. Figure 9 .

[0073] In some embodiments, reference Figure 2 and combined Figure 4 Step S12: Based on the initial morphology of the microcatheter, determine several different microcatheter morphologies, including:

[0074] Step S121: Obtain the discrete point set of the initial morphology of the microcatheter;

[0075] The discrete point set of the initial morphology of the microcatheter can be understood as the pixel set of the initial morphology OBP of the microcatheter represented in the image.

[0076] Step S122: Randomly perturb at least one discrete point in the discrete point set to determine the updated discrete point set;

[0077] Step S123: Fit the updated discrete point set to determine the updated microcatheter morphology;

[0078] Step S124: Repeat the random perturbation and fitting several times to determine several different microcatheter morphologies.

[0079] Specifically, a discrete point set is extracted from the initial shape (OBP) of the microcatheter. This discrete point set refers to the set of three-dimensional coordinates of points on the OBP, which can be used to describe the OBP's orientation, edge distribution, width, and other information. For example, the discrete point set can refer to a set of pixels along the microcatheter shape. At least one discrete point in this set is randomly perturbed. This random perturbation can be understood as making small random changes to the spatial coordinates of the discrete point. For example, a pixel {x, y, z} is updated to {x+α, y+β, z+γ} after random perturbation, where α, β, and γ represent the movement values ​​along the x, y, and z directions, respectively. After forming a new discrete point set, it is fitted to obtain the updated microcatheter shape. This random perturbation and fitting of the new discrete point set after random perturbation is repeated several times to obtain several different microcatheter shapes.

[0080] Before the random perturbation begins, the number of microcatheter morphologies to be generated is set, or the number of iterations for the microcatheter morphology is set, until the set number of iterations is reached or no new microcatheter morphologies are generated, which completes the determination of several different microcatheter morphologies.

[0081] It is important to note that the random perturbation of discrete points needs to be constrained to prevent the generated new discrete points from exceeding the boundaries of the parent artery or aneurysm, in which case the generated microcatheter morphology would be meaningless. Therefore, the new discrete points generated by randomly perturbing the discrete points must still be located on the mask image of the parent artery or the mask image of the aneurysm.

[0082] The above describes how to determine several different microcatheter morphologies based on the initial images of the aneurysm and the parent artery, i.e., step S1. The following describes how to determine the appropriate microcatheter morphology, i.e., step S2.

[0083] refer to Figure 1 and combined Figure 5 Step S2, which evaluates each of the above microcatheter morphologies based on a pre-trained neural network model to determine a suitable microcatheter morphology, includes:

[0084] Step S21: Obtain training sample data, which includes multiple microcatheter sample data, and the multiple microcatheter sample data are divided into two categories: suitable and unsuitable.

[0085] The training sample data generally consists of multiple microcatheter samples suitable for the aforementioned aneurysms and carrier arteries, selected by experts and evaluated by automatic or semi-automatic methods, as well as multiple microcatheter samples that are not suitable for the aforementioned aneurysms and carrier arteries.

[0086] Each suitable and unsuitable microcatheter sample is labeled to determine the label of the training sample data. For example, suitable microcatheter samples are labeled as "1" and unsuitable microcatheter samples are labeled as "0", thus obtaining the label of the training sample data. The advantage of setting multiple unsuitable microcatheter samples in addition to multiple suitable microcatheter samples is that it can utilize more comprehensive data for model training, and the increased amount of unsuitable microcatheter samples participating in model training further ensures the model accuracy.

[0087] For example, the ratio of suitable to unsuitable microcatheter samples is 1:1. This sample selection strategy ensures a balance between suitable and unsuitable microcatheter samples during training.

[0088] Step S22: Based on the training sample data, including the mask images of aneurysms and the aneurysm-bearing arteries, the initial neural network model determines the predicted labels of the training sample data;

[0089] Here, the mask image of the aneurysm and its parent artery refers to the mask image obtained in step S113. Based on the training sample data, including the mask image of the aneurysm and its parent artery, the initial neural network model determines the predicted labels for the training sample data.

[0090] refer to Figure 5 and combined Figure 6 The step of determining the predicted labels of the training data samples based on training sample data, including mask images of aneurysms and aneurysm-bearing arteries, further includes:

[0091] Step S221: Based on the initial neural network model, extract features from the mask image of the aneurysm and the carrier artery, as well as the training sample data;

[0092] Step S222: The fully connected layer of the initial neural network model processes the extracted features and outputs a feature vector;

[0093] Step S223: The softmax layer of the initial neural network model determines the predicted labels of the training sample data based on the feature vectors.

[0094] Based on the initial neural network model, features of the mask image including the aneurysm and the carrier artery, as well as features of the training sample data, are extracted. These features can be latent or explicit. Explicit features represent easily observable features, such as the curvature and thickness of the microcatheter, the thickness of the blood vessel, and the appearance of the aneurysm. Latent features represent hidden features, such as the relationship between the curvature of the microcatheter and the location of the aneurysm, and the influence of the thickness of the blood vessel on the curvature of the microcatheter.

[0095] The fully connected layer of the initial neural network model processes the extracted features and outputs a feature vector. The softmax layer of the initial neural network model determines the predicted label of the training sample data based on the feature vector.

[0096] Step S23: Calculate the function value of the loss function based on the predicted label and the label of the training sample data;

[0097] Step S24: Based on the function value of the loss function, determine the gradient of the initial neural network model, perform backpropagation and continuously update the weight parameters of the initial neural network model until the function value of the loss function converges to a stable state, and determine the neural network model based on the pre-trained model.

[0098] The loss function's value can be used to measure the difference between the predicted label of the initial neural network model and the label of the training sample data. The weight parameters of each layer of the neural network are updated based on the difference between the two until the predicted label of the initial neural network model and the label of the training sample data are very close, and finally the pre-trained neural network model is determined.

[0099] The methods for calculating the value of the loss function include the cross-entropy calculation method or the mean-variance calculation method.

[0100] The backpropagation algorithm is used to adjust the weight parameters in the initial neural network model during training, thereby reducing the error loss of the initial neural network model and ultimately determining the pre-trained neural network model. Specifically, forward propagation of the input signal to the output generates error loss. By backpropagating this error loss information, the weight parameters in the initial neural network model are updated, causing the error loss to converge to a steady state. This steady state means that the error loss no longer decreases, or remains relatively convergent within a certain range. The backpropagation algorithm is an error-loss-driven backpropagation process aimed at obtaining the optimal parameters of the initial neural network model, such as the weight matrix.

[0101] Step S25: Evaluate each microcatheter morphology based on a pre-trained neural network model to determine the appropriate microcatheter morphology.

[0102] For example, a mask image of the segmented aneurysm-bearing artery and aneurysm is input into a pre-trained neural network model. For example, the background is labeled "0", the aneurysm-bearing artery is labeled "1", and the aneurysm is labeled "2". Furthermore, one of the several different microcatheter morphologies generated in step S12 is input into the pre-trained neural network model.

[0103] The microcatheter morphology is evaluated based on a pre-trained neural network model to determine a suitable microcatheter morphology.

[0104] The predicted value output by softmax is the microcatheter morphology score, which ranges from 0 to 1. 1 represents the optimal microcatheter morphology. The closer the score is to 1, the more suitable the microcatheter morphology is for the aneurysm-bearing artery and the aneurysm mentioned above. In other words, the greater the probability that the microcatheter morphology matches the aneurysm.

[0105] In practice, a threshold S can be preset. When the score of the microcatheter morphology based on the pre-trained neural network model is between 0 and S (excluding S), the final output result is an unsuitable microcatheter morphology. If the score is between S and 1, the final output result is a suitable microcatheter morphology.

[0106] refer to Figures 1-10 The following describes an exemplary implementation of a method for determining the morphology of microcatheters.

[0107] This exemplary implementation method first determines several different microcatheter morphologies, and then evaluates each of the above microcatheter morphologies based on a pre-trained neural network model to determine a suitable microcatheter morphology.

[0108] Images of aneurysms and the parent artery are obtained using 3D-DSA imaging technology;

[0109] Identify the target location region of the aneurysm from 3D-DSA images;

[0110] The images of the aneurysm and the parent artery are segmented, with the parent artery labeled as "1", the aneurysm labeled as "2", and the background labeled as "0", to obtain the segmented mask images of the aneurysm and the parent artery.

[0111] The centroid P of the aneurysm is determined based on the centroids of the surface pixels in the mask image of the aneurysm.

[0112] Based on the mask image of the tumor-bearing artery, the centerline of the tumor-bearing artery is extracted, and the starting point O is marked by manual marking.

[0113] Establish a perpendicular line from the center of mass P to the center line, where the intersection of the perpendicular line and the center line is B;

[0114] The initial morphology of the microcatheter, OBP, is determined based on the vertical segment BP and the centerline.

[0115] This determines the initial morphology (OBP) of the microcatheter, and then several different microcatheter morphologies are generated based on the initial OBP.

[0116] The step of determining several different microcatheter morphologies based on the initial morphology of the microcatheter includes:

[0117] Based on the initial shape of the microcatheter (OBP), obtain the discrete point set of the initial shape of the microcatheter.

[0118] Let N be the number of microcatheter morphologies that need to be generated in the end.

[0119] By randomly and slightly modifying the spatial coordinates of at least one discrete point in the initial discrete point set of the microcatheter morphology, at least one of the discrete points becomes a new discrete point, thus determining the updated discrete point set. The updated discrete point set is then fitted to determine the updated microcatheter morphology.

[0120] Repeat the above operation several times until the number of microcatheters with different shapes has reached the set number N of microcatheter shapes, or until the shape of the newly generated microcatheters no longer changes. This completes the determination of several different microcatheter shapes.

[0121] At this point, N different microcatheter morphologies were determined.

[0122] Next, the neural network model is trained to obtain a pre-trained neural network model that can evaluate the above N different microcatheter morphologies to determine the suitable microcatheter morphology.

[0123] Based on training sample data and mask images including aneurysms and carrier arteries, an initial neural network model determines predicted labels for the training sample data. The training sample data typically consists of multiple microcatheter samples suitable for the aforementioned aneurysms and carrier arteries, as well as multiple microcatheter samples unsuitable for the aneurysms and carrier arteries, selected by expert screening, automated methods, or semi-automatic methods. Each suitable and unsuitable microcatheter sample is labeled to determine the label of the training sample data. For example, suitable microcatheter samples are labeled as "1", and unsuitable microcatheter samples are labeled as "0", thus determining the label of the training sample data.

[0124] The loss function is calculated by combining the predicted labels and the labels of the training sample data.

[0125] Based on the value of the loss function, the gradient of the initial neural network model is determined. Backpropagation is performed and the weight parameters of the initial neural network model are continuously updated until the value of the loss function no longer decreases or remains relatively convergent within a certain range, thus determining the neural network model based on the pre-trained model.

[0126] At this point, the neural network model has been trained.

[0127] Several different microcatheter morphologies generated in advance are input into the pre-trained neural network model. The pre-trained neural network model evaluates the microcatheter morphologies by scoring them. The scoring range is 0-1, where 1 represents the optimal microcatheter morphology. The closer the score is to 1, the more suitable the microcatheter morphology is for the aneurysm-bearing artery and the aneurysm, and the greater the probability that the microcatheter morphology matches the aneurysm.

[0128] In practice, a threshold S can be preset. When the score of the microcatheter morphology based on the pre-trained neural network model is between 0 and S (excluding S), the final output result is an unsuitable microcatheter morphology. If the score is between S and 1, the final output result is a suitable microcatheter morphology. In this way, several different microcatheter morphologies are classified into two categories: suitable microcatheter morphologies and unsuitable microcatheter morphologies.

[0129] This exemplary implementation uses a pre-trained neural network model to score or classify several different microcatheter morphologies generated in advance. This allows doctors or professionals to shape microcatheters in a short time by referring to suitable or optimal microcatheter morphologies, improving the accuracy of microcatheter shaping, reducing the probability of doctors or professionals having to reshape them if the first shaping is unsuccessful, and reducing the time required for microcatheter shaping.

[0130] refer to Figure 7The present invention also provides a system 10 for determining the morphology of a microcatheter, comprising:

[0131] Generator 11, generator 11 determines several different microcatheter morphologies;

[0132] Discriminator 12, based on a pre-trained neural network model, evaluates each microcatheter morphology to determine the appropriate microcatheter morphology.

[0133] By adopting the above technical solution, based on a pre-trained neural network model, the morphology of multiple microcatheters can be evaluated, and the appropriate microcatheter morphology can be determined. This improves the accuracy of matching the microcatheter morphology with the carrier artery and aneurysm, reduces the time for microcatheter morphology shaping, and provides effective data support for surgery.

[0134] refer to Figure 8 The present invention also provides an electronic device 20, including a memory 21, a processor 22, and a computer program stored on the memory 21 that can run on the processor 22. When the processor 22 executes the computer program, it implements the method for determining the microcatheter morphology in any of the foregoing embodiments. The memory may include, for example, system memory, a fixed non-volatile storage medium, etc. The system memory may store, for example, an operating system, application programs, a boot loader, and other programs.

[0135] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method for determining the morphology of a microcatheter in any of the foregoing embodiments. Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable non-transitory storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0136] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0137] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0138] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0139] It should be noted that in some alternative implementations, the functions marked in the boxes may occur in a different order than those shown in the accompanying drawings. For example, two consecutive boxes may actually be executed substantially in parallel, or sometimes in reverse order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and combinations of boxes in the block diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified functions or actions, or using a combination of dedicated hardware and computer instructions.

[0140] Furthermore, although the present invention has been illustrated and described with reference to certain preferred embodiments, those skilled in the art should understand that the above description is a further detailed explanation of the invention in conjunction with specific embodiments, and should not be construed as limiting the specific implementation of the invention to these descriptions. Those skilled in the art can make various changes in form and detail, including several simple deductions or substitutions, without departing from the spirit and scope of the invention.

Claims

1. A method for determining the morphology of a microcatheter, characterized in that, The microcatheter morphology is a pre-shaped form for introducing coils into the aneurysm, and the method includes: Determine several different microcatheter morphologies; The step of determining several different microcatheter morphologies includes: Determine the initial morphology of the microcatheter; Based on the initial morphology of the microcatheter, several different microcatheter morphologies are determined; The step of determining several different microcatheter morphologies based on the initial morphology of the microcatheter includes: Extract the discrete point set of the initial shape of the microcatheter, wherein the discrete point set of the initial shape refers to the set of three-dimensional coordinates of the points on the initial shape of the microcatheter. At least one discrete point in the discrete point set is randomly perturbed to determine an updated discrete point set, wherein the random perturbation is a random, minor change to the spatial coordinates of the discrete point. Fit the updated discrete point set to determine the updated microcatheter morphology; The random perturbation and fitting were repeated several times to determine several different microcatheter morphologies; Based on a pre-trained neural network model, each of the microcatheter morphologies is evaluated to determine a suitable microcatheter morphology. The evaluation method involves scoring the microcatheter morphology, and the score is used to characterize the degree of adaptation of the corresponding microcatheter morphology to the carrier artery and aneurysm.

2. The method for determining the morphology of a microcatheter as described in claim 1, characterized in that, The step of determining the initial morphology of the microcatheter includes: Acquire images including the aneurysm and the parent artery; Based on the images, the location of the aneurysm was determined; Based on the location of the aneurysm, the aneurysm and the carrier artery are segmented to determine a mask image of the aneurysm and the carrier artery; Based on the mask image of the aneurysm, the centroid P of the aneurysm is determined, and based on the mask image of the aneurysm-carrying artery, the center line of the aneurysm-carrying artery is extracted, and the starting point O of the center line is determined. Draw a perpendicular line from the centroid P to the center line, wherein the intersection point of the perpendicular line and the center line is B; Based on the vertical segment BP and the centerline, the initial morphology OBP of the microcatheter is established.

3. The method for determining the morphology of microcatheters as described in claim 2, characterized in that, The discrete points are located on the masked image of the tumor-bearing artery or the aneurysm.

4. The method for determining the morphology of a microcatheter as described in claim 2, characterized in that, The step of evaluating each microcatheter morphology based on the pre-trained neural network model to determine a suitable microcatheter morphology includes: Acquire training sample data, which includes multiple microcatheter sample data, wherein the multiple microcatheter sample data are divided into two categories: suitable and unsuitable. Based on the training sample data and the mask image including the aneurysm and the carrier artery, the initial neural network model determines the predicted label of the training sample data; The loss function value is calculated based on the predicted label and the label of the training sample data; Based on the value of the loss function, the gradient of the initial neural network model is determined, backpropagation is performed, and the weight parameters of the initial neural network model are continuously updated until the value of the loss function converges to a stable state, thus determining the pre-trained neural network model. Based on the pre-trained neural network model, each microcatheter morphology is evaluated to determine the appropriate microcatheter morphology.

5. The method for determining the morphology of a microcatheter as described in claim 4, characterized in that, The step of determining the predicted labels of the training sample data based on the training sample data, including mask images of aneurysms and aneurysm-bearing arteries, and the initial neural network model includes: Based on the initial neural network model, features of the mask image including the aneurysm and the carrier artery, and the training sample data are extracted; The fully connected layer of the initial neural network model processes the extracted features and outputs a feature vector; The softmax layer of the initial neural network model determines the predicted label of the training sample data based on the feature vector.

6. The method for determining the morphology of a microcatheter as described in claim 4, characterized in that, The loss function value is calculated using either the cross-entropy calculation method or the mean-variance calculation method.

7. A system for determining the morphology of microcatheters, characterized in that, The microcatheter is pre-shaped for coil introduction into the aneurysm, and the system includes: A generator determines several different microcatheter morphologies. Determining the several different microcatheter morphologies includes: determining an initial microcatheter morphology; and based on the initial microcatheter morphology, determining several different microcatheter morphologies. The step of determining several different microcatheter morphologies based on the initial microcatheter morphology includes: extracting a discrete point set of the initial microcatheter morphology, wherein the discrete point set of the initial morphology refers to the set of three-dimensional coordinates of points on the initial microcatheter morphology; randomly perturbing at least one discrete point in the discrete point set to determine an updated discrete point set, wherein the random perturbation is a random, minor modification to the spatial coordinates of the discrete points; fitting the updated discrete point set to determine an updated microcatheter morphology; and repeating the random perturbation and fitting process several times to determine several different microcatheter morphologies. The discriminator, based on a pre-trained neural network model, evaluates each of the microcatheter morphologies to determine a suitable microcatheter morphology. The evaluation method involves scoring the microcatheter morphology, and the score is used to characterize the degree of adaptation of the corresponding microcatheter morphology to the carrier artery and aneurysm.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method for determining the morphology of a microcatheter as described in any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method for determining the morphology of a microcatheter as described in any one of claims 1 to 6.

Images