CN116994038A

CN116994038A - Adjustment method and device of blood vessel classification model, electronic equipment and storage medium

Info

Publication number: CN116994038A
Application number: CN202310833431.5A
Authority: CN
Inventors: 刘恩佑; 郝增号; 王明阳; 张欢; 王少康; 陈宽
Original assignee: Infervision Medical Technology Co Ltd
Current assignee: Infervision Medical Technology Co Ltd
Priority date: 2023-07-07
Filing date: 2023-07-07
Publication date: 2023-11-03

Abstract

The embodiment of the invention discloses a method and a device for adjusting a blood vessel classification model, electronic equipment and a storage medium. The method comprises the following steps: acquiring a blood vessel point cloud image, a blood vessel classification model and a trained feature extraction model, wherein the blood vessel point cloud image comprises at least one blood vessel point; inputting the vessel point cloud image into the feature extraction model, and obtaining point features corresponding to the at least one vessel point respectively according to the output result of the feature extraction model; for each vascular point in the at least one vascular point, determining a direction vector of the vascular point according to a point characteristic corresponding to the vascular point; and determining an attention sequence based on the at least one blood vessel point and the direction vector corresponding to the at least one blood vessel point respectively, and adjusting the blood vessel classification model according to the attention sequence. The technical scheme of the embodiment of the invention can improve the classification accuracy of the blood vessel classification model.

Description

Adjustment method and device of blood vessel classification model, electronic equipment and storage medium

Technical Field

The embodiment of the invention relates to the technical field of image processing, in particular to a method and a device for adjusting a blood vessel classification model, electronic equipment and a storage medium.

Background

In the medical field, it is important to accurately classify blood vessels.

At present, a model mode is generally adopted for blood vessel classification, but the classification accuracy of the existing blood vessel classification model is low, and the problem is solved.

Disclosure of Invention

The embodiment of the invention provides a method and a device for adjusting a blood vessel classification model, electronic equipment and a storage medium, so as to improve the classification accuracy of the blood vessel classification model.

According to an aspect of the present invention, there is provided a method for adjusting a blood vessel classification model, which may include:

acquiring a blood vessel point cloud image, a blood vessel classification model and a trained feature extraction model, wherein the blood vessel point cloud image comprises at least one blood vessel point;

inputting the vessel point cloud image into a feature extraction model, and obtaining at least one point feature corresponding to the vessel point according to the output result of the feature extraction model;

for each vascular point in at least one vascular point, determining a direction vector of the vascular point according to the point characteristics corresponding to the vascular point;

And determining an attention sequence based on at least one blood vessel point and a direction vector corresponding to the at least one blood vessel point respectively, and adjusting the blood vessel classification model according to the attention sequence.

According to another aspect of the present invention, there is provided an adjusting apparatus for a blood vessel classification model, which may include:

the feature extraction model acquisition module is used for acquiring a blood vessel point cloud image, a blood vessel classification model and a trained feature extraction model, wherein the blood vessel point cloud image comprises at least one blood vessel point;

the point feature obtaining module is used for inputting the vessel point cloud image into the feature extraction model, and obtaining at least one point feature corresponding to the vessel point respectively according to the output result of the feature extraction model;

the direction vector determining module is used for determining a direction vector of each vascular point in the at least one vascular point according to the point characteristics corresponding to the vascular point;

and the blood vessel classification model adjusting module is used for determining an attention sequence based on at least one blood vessel point and the direction vector corresponding to the at least one blood vessel point respectively, and adjusting the blood vessel classification model according to the attention sequence.

According to another aspect of the present invention, there is provided an electronic device, which may include:

At least one processor; and

a memory communicatively coupled to the at least one processor; wherein,,

the memory stores a computer program executable by the at least one processor, the computer program being executable by the at least one processor to cause the at least one processor to implement the method for adapting a blood vessel classification model provided by any embodiment of the present invention when executed.

According to another aspect of the present invention, there is provided a computer readable storage medium having stored thereon computer instructions for causing a processor to implement the method for adapting a blood vessel classification model provided by any embodiment of the present invention when executed.

According to the technical scheme, a blood vessel point cloud image, a blood vessel classification model and a trained feature extraction model are acquired, wherein the blood vessel point cloud image comprises at least one blood vessel point; inputting the vessel point cloud image into a feature extraction model, and obtaining at least one point feature corresponding to the vessel point according to the output result of the feature extraction model; for each vascular point in at least one vascular point, determining a direction vector of the vascular point according to the point characteristics corresponding to the vascular point; and determining an attention sequence based on at least one blood vessel point and a direction vector corresponding to the at least one blood vessel point respectively, and adjusting the blood vessel classification model according to the attention sequence. According to the technical scheme, the blood vessel classification model is adjusted according to the attention sequence of the direction of the blood vessel, so that the attention of the direction of the blood vessel can be introduced into the blood vessel classification model, and the classification accuracy of the blood vessel classification model can be improved.

It should be understood that the description in this section is not intended to identify key or critical features of the embodiments of the invention, nor is it intended to be used to limit the scope of the invention. Other features of the present invention will become apparent from the description that follows.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly described below, and it is apparent that the drawings in the following description are only some embodiments of the present invention, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.

FIG. 1 is a flow chart of a method for adjusting a blood vessel classification model according to an embodiment of the present invention;

FIG. 2 is a schematic diagram of an attention sequence in a method for adjusting a blood vessel classification model according to an embodiment of the present invention;

FIG. 3 is a flow chart of another method for adapting a blood vessel classification model according to an embodiment of the present invention;

FIG. 4 is a flow chart of a method for tuning a blood vessel classification model according to an embodiment of the present invention;

FIG. 5 is a flow chart of a method for tuning a blood vessel classification model according to an embodiment of the present invention;

FIG. 6 is a flowchart of an alternative example of a method for adapting a blood vessel classification model according to an embodiment of the present invention;

FIG. 7 is a block diagram of an adjusting apparatus for a blood vessel classification model according to an embodiment of the present invention;

fig. 8 is a schematic structural diagram of an electronic device for implementing a method for adjusting a blood vessel classification model according to an embodiment of the present invention.

Detailed Description

In order that those skilled in the art will better understand the present invention, a technical solution in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in which it is apparent that the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present invention without making any inventive effort, shall fall within the scope of the present invention.

It should be noted that the terms "first," "second," and the like in the description and the claims of the present invention and the above figures are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments of the invention described herein may be implemented in sequences other than those illustrated or otherwise described herein. The cases of "target", "original", etc. are similar and will not be described in detail herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.

Fig. 1 is a flowchart of a method for adjusting a blood vessel classification model according to an embodiment of the present invention. The embodiment is applicable to the case of adjusting the blood vessel classification model. The method can be executed by the adjusting device of the blood vessel classification model, which can be realized by software and/or hardware, and can be integrated on electronic equipment, and the electronic equipment can be various user terminals or servers.

Referring to fig. 1, the method of the embodiment of the present invention specifically includes the following steps:

s101, acquiring a blood vessel point cloud image, a blood vessel classification model and a trained feature extraction model, wherein the blood vessel point cloud image comprises at least one blood vessel point.

Wherein the vessel point is a point in a vessel point cloud in the vessel point cloud image.

In the embodiment of the invention, the blood vessel point cloud image is a point cloud image of a blood vessel obtained by collecting the blood vessel. The vessel point cloud image may be, for example, a vessel point cloud image including only a vessel point on a vessel obtained by acquiring a point cloud image including a vessel through a Digital Radiography (DR), an electronic computed tomography (Computed Tomography, CT), a CT angiography (CTA), and/or a CT vessel imaging technique, and the source of the vessel point cloud image is not particularly limited in the embodiment of the present invention. The number of vessel point cloud images may be one or more. The blood vessel point cloud image may be a head blood vessel point cloud image, a leg blood vessel point cloud image, an elbow blood vessel point cloud image, or the like, and in the embodiment of the present invention, the type of the blood vessel point cloud image and the position where the blood vessel point cloud image is acquired are not specifically limited.

In an embodiment of the present invention, the blood vessel classification model is a model that can classify blood vessels. The vessel classification model may be used to classify vessels characterized based on at least one vessel point, and may also be used to classify each vessel point on a vessel. The blood vessel classification model can be a blood vessel classification model which is already trained, and can also be a blood vessel classification model which is not already trained; in the case where the blood vessel classification model is an untrained blood vessel classification model, after the blood vessel classification model is adjusted according to the attention sequence, the adjusted blood vessel classification model may also be trained. The classification of the blood vessel classification model may be, for example, arterial or venous classification, for example, cardiac blood vessel, head blood vessel, leg blood vessel, main blood vessel, branch blood vessel, capillary blood vessel, etc., and in the embodiment of the present invention, the classification of the blood vessel classification model is not specifically limited.

In the embodiment of the invention, the feature extraction model can be a model which is trained and can extract the features of the vessel point cloud image; the feature extraction model may, for example, perform feature extraction for each of at least one vessel point in the vessel point cloud image. In the embodiment of the present invention, the kind of the feature extraction model is not particularly limited.

Optionally, the feature extraction model is implemented based on a Unet network, which at least includes a sparse convolution layer. Specifically, a convolutional layer in the Unet network may be replaced by a sparse convolutional layer, and the Unet network in which the convolutional layer is replaced by the sparse convolutional layer is used as a backbone network of the feature extraction model. In the embodiment of the invention, as each convolution kernel of the traditional convolution layer is calculated with each pixel point of the input image, the calculated amount of the traditional convolution layer is larger; in the sparse convolution layer, the input image is divided into a plurality of sub-images, each sub-image only comprises a small part of pixel points in the input image, each convolution kernel only calculates with a part of pixel points in the sub-image, and other pixel points are ignored, so that the Unet network at least comprises the sparse convolution layer, the calculated amount can be greatly reduced, the extraction efficiency and accuracy of the feature extraction model are improved, the Unet network can have higher resolution, the Unet network is more suitable for a blood vessel classification scene, the generalization capability and robustness of the feature extraction model can be improved, and the problems of over fitting and under fitting are avoided; and the result of the output of the Unet network is organized in a point-wise manner, the Unet network is also particularly suitable for determining the attention sequence.

It should be noted that, the technical solution of the embodiment of the present invention may be applied to not only adjusting a blood vessel classification model, but also acquiring a pipeline point cloud image, a pipeline classification model and a trained feature extraction model to adjust the pipeline classification model, where the above-mentioned pipeline may be a tubular object such as a sewer pipeline, and in the embodiment of the present invention, a specific type of the pipeline is not specifically limited. Wherein the pipeline point cloud image may be an image comprising at least one pipeline point.

S102, inputting the vessel point cloud image into a feature extraction model, and obtaining point features corresponding to at least one vessel point respectively according to an output result of the feature extraction model.

In the embodiment of the invention, the cloud image of the blood vessel points can be input into the feature extraction model, the feature extraction model can perform feature extraction on each blood vessel point in at least one blood vessel point to obtain the output result of the feature extraction model, and the output result can represent the features respectively corresponding to the blood vessel points, so that the point features respectively corresponding to the at least one blood vessel point can be obtained according to the output result of the feature extraction model. The point features are features corresponding to the blood vessel points.

S103, determining a direction vector of each vascular point in the at least one vascular point according to the point characteristics corresponding to the vascular point.

In the embodiment of the invention, for each vascular point in at least one vascular point, the point characteristics corresponding to the vascular point can be processed, for example, the point characteristics can be processed by a filtering algorithm, the direction vector of the vascular point is determined, the direction vector of the vascular point is the direction of the vascular at the vascular point, which can be used for representing the direction of the vascular tangent at the vascular point, and the direction of the vascular at the vascular point can be determined through the direction vector.

S104, determining an attention sequence based on at least one blood vessel point and a direction vector corresponding to the at least one blood vessel point respectively, and adjusting the blood vessel classification model according to the attention sequence.

In the embodiment of the present invention, an Attention sequence may be determined based on at least one vascular point and a direction vector corresponding to the at least one vascular point, referring to fig. 2, where the Attention sequence is an Attention (Attention) sequence that may represent an association relationship and a direction relationship between the at least one vascular point, and the blood vessel classification model is adjusted according to the Attention sequence.

It should be noted that, according to the attention sequence, the manner of adjusting the blood vessel classification model may be to add a module related to the attention sequence to the blood vessel classification model, or may be to adjust parameters in the blood vessel classification model according to the attention sequence, or the like.

An optional technical scheme, according to the corresponding point characteristic of the vascular point, determines the direction vector of the vascular point, including: processing point features corresponding to the blood vessel points based on a target filtering algorithm to obtain at least one feature value and feature vectors corresponding to the at least one feature value respectively; and taking the feature vector corresponding to the feature value with the smallest value in the at least one feature value as the direction vector of the vascular point.

In the embodiment of the present invention, the specific type of the target filtering algorithm is not specifically limited.

The target filtering algorithm may be, for example, a Hessian matrix filtering algorithm among franki filtering algorithms. Specifically, gaussian filtering processing can be performed on point features corresponding to the vascular points under multiple scales; according to the result of Gaussian filtering processing, calculating a second derivative construction Hessian matrix of each vascular point, wherein a first derivative in the second derivative construction Hessian matrix represents the change of the gray level of the vascular point, and the second derivative represents the change of the gray level of the vascular point, namely the gradient of the gray level of the vascular point; constructing a Hessian matrix according to the second derivative, and calculating to obtain three characteristic values; for each of the three eigenvalues, solving eigenvectors of a second derivative construction Hessian matrix according to the eigenvalues, thereby obtaining eigenvectors respectively corresponding to the three eigenvalues, wherein the eigenvectors can represent directions corresponding to the eigenvalues; and taking the feature vector corresponding to the feature value with the smallest numerical value in the three feature values as the direction vector of the vascular point. It should be noted that, since the gradient of the gray scale in two perpendicular directions of the blood vessel is large, that is, the gradient of the gray scale in two perpendicular directions on the plane intersecting the blood vessel is large, the gradient of the gray scale in the direction along the blood vessel is small, and the feature value may represent the gradient of the gray scale of the blood vessel in the corresponding direction, two feature values are large, one feature value is about 0, and the feature vector corresponding to the feature value with the smallest value of the three feature values may be used as the direction vector of the blood vessel point. According to the technical scheme, the direction vector calculated by the Franage filtering algorithm can be utilized to reduce the display memory occupied by the attention sequence, so that the calculation load is reduced, the calculation efficiency is improved, the Franage filtering algorithm can naturally reflect the blood vessel flow direction, the prior condition in the direction is added into the blood vessel classification to assist the training of the blood vessel classification model, the accuracy and the stability of the blood vessel classification model classification are improved, and the accuracy and the robustness of the blood vessel classification can be effectively improved based on the mode of combining the direction vector determination and the blood vessel classification of the Franage filtering algorithm.

In the embodiment of the invention, the characteristic of the characteristic value of the gradient which can represent the gray scale of the vascular point along each direction can be obtained according to the point characteristic corresponding to the vascular point after being processed by the target filtering algorithm, so that the direction vector of the vascular point is represented, the determination of the prescription vector of the vascular point is realized, and the follow-up attention sequence is facilitated.

In another alternative technical scheme, the blood vessel classification model comprises a blood vessel classification module and a blood vessel staining module; after adjusting the vessel classification model according to the attention sequence, further comprising: updating the blood vessel classification model according to the obtained adjustment result; acquiring a target blood vessel image, and inputting the target blood vessel image into a blood vessel classification module; and inputting the output result of the blood vessel classification module into a blood vessel staining module to obtain a blood vessel staining result aiming at the target blood vessel image.

It should be noted that the blood vessel classification model may include a blood vessel classification module for classifying blood vessels and a blood vessel staining module for staining blood vessels, i.e. the blood vessel classification model may not only classify blood vessels but also stain blood vessels. In the case where only classification of blood vessels is required, the blood vessel classification model may be updated according to the obtained adjustment result after adjustment of the blood vessel classification model according to the attention sequence; and acquiring a target blood vessel image, inputting the target blood vessel image into a blood vessel classification module, and obtaining a blood vessel classification result aiming at the target blood vessel image according to the output result of the blood vessel classification module. If the blood vessel is required to be dyed, after the blood vessel classification model is adjusted according to the attention sequence, the blood vessel classification model can be updated according to the obtained adjustment result; and acquiring a target blood vessel image, inputting the target blood vessel image into a blood vessel classification module, and inputting the output result of the blood vessel classification module into a blood vessel staining module to obtain a blood vessel staining result aiming at the target blood vessel image.

According to the technical scheme provided by the embodiment of the invention, the attention to the blood vessel direction can be introduced into the blood vessel classification model capable of realizing blood vessel dyeing, so that the blood vessel dyeing can be realized without manual operation or by adopting a complex algorithm, and the dyeing efficiency and accuracy of the blood vessel classification model capable of realizing blood vessel dyeing can be improved.

Fig. 3 is a flowchart of another method for adjusting a blood vessel classification model according to an embodiment of the present invention. The present embodiment is optimized based on the above technical solutions. In this embodiment, optionally, determining the attention sequence based on at least one vascular point and a direction vector corresponding to the at least one vascular point respectively includes: for each vascular point in the at least one vascular point, determining a corresponding point corresponding to the vascular point from the at least one vascular point based on the at least one vascular point and the direction vector corresponding to the at least one vascular point respectively; and determining the attention sequence according to the corresponding points corresponding to the at least one blood vessel point respectively. Wherein, the explanation of the same or corresponding terms as the above embodiments is not repeated herein.

Referring to fig. 3, the method of this embodiment may specifically include the following steps:

S201, acquiring a blood vessel point cloud image, a blood vessel classification model and a trained feature extraction model, wherein the blood vessel point cloud image comprises at least one blood vessel point.

S202, inputting the vessel point cloud image into a feature extraction model, and obtaining point features corresponding to at least one vessel point respectively according to an output result of the feature extraction model.

S203, determining a direction vector of each vascular point in the at least one vascular point according to the point characteristics corresponding to the vascular point.

S204, determining corresponding points corresponding to the blood vessel points from the at least one blood vessel points based on the at least one blood vessel point and the direction vectors corresponding to the at least one blood vessel point respectively for each of the at least one blood vessel point.

In the embodiment of the invention, for each vascular point in at least one vascular point, a corresponding point which corresponds to the vascular point and can represent the front-back association with the vascular point can be determined from the at least one vascular point based on the at least one vascular point and the direction vector corresponding to the at least one vascular point.

S205, determining an attention sequence according to corresponding points corresponding to at least one blood vessel point respectively.

In the embodiment of the invention, the vascular point sequence corresponding to each vascular point in the at least one vascular point can be determined according to the corresponding point corresponding to the vascular point respectively, wherein the vascular point sequence is a sequence of vascular points which are corresponding to the vascular point and are characterized by the front-back relevance of the vascular point and the vascular point, and the attention sequence is determined according to the vascular point sequence corresponding to the vascular point respectively.

S206, adjusting the blood vessel classification model according to the attention sequence.

According to the technical scheme of the embodiment of the invention, for each vascular point in at least one vascular point, the corresponding point corresponding to the vascular point is determined from the at least one vascular point based on the at least one vascular point and the direction vector corresponding to the at least one vascular point respectively; and determining the attention sequence according to the corresponding points corresponding to the at least one blood vessel point respectively. According to the technical scheme provided by the embodiment of the invention, the attention sequence capable of representing the relevance between the blood vessel points can be determined through the determination of the corresponding points, so that the blood vessel classification model can be conveniently adjusted to obtain the blood vessel classification model with higher accuracy according to the attention sequence.

Fig. 4 is a flowchart of a method for adjusting a blood vessel classification model according to another embodiment of the present invention. The present embodiment is optimized based on the above technical solutions. In this embodiment, optionally, the method for adjusting a blood vessel classification model further includes: for each vascular point in at least one vascular point, carrying out normalization processing on the direction vector of the vascular point to obtain a unit vector; constructing a target K-dimensional tree according to at least one vascular point and unit vectors corresponding to the vascular points respectively; for each of the at least one vascular point, determining a corresponding point corresponding to the vascular point from the at least one vascular point based on the at least one vascular point and the direction vector corresponding to the at least one vascular point, respectively, comprising: and determining a corresponding point corresponding to the blood vessel point from the at least one blood vessel point according to the target K-dimensional tree aiming at each blood vessel point in the at least one blood vessel point. Wherein, the explanation of the same or corresponding terms as the above embodiments is not repeated herein.

Referring to fig. 4, the method of this embodiment may specifically include the following steps:

s301, acquiring a blood vessel point cloud image, a blood vessel classification model and a trained feature extraction model, wherein the blood vessel point cloud image comprises at least one blood vessel point.

S302, inputting the vessel point cloud image into a feature extraction model, and obtaining point features corresponding to at least one vessel point respectively according to an output result of the feature extraction model.

S303, determining a direction vector of each vascular point in at least one vascular point according to the point characteristics corresponding to the vascular point.

S304, carrying out normalization processing on the direction vector of the vascular point to obtain a unit vector.

In the embodiment of the invention, the direction vector of the vascular point including offset information such as a direction offset angle can be normalized for each vascular point in at least one vascular point to obtain the unit vector which can more represent the direction of the blood vessel at the vascular point, for example, the direction vector of the vascular point can represent that the direction of the blood vessel at the vascular point is offset 5 degrees clockwise in the north-south direction, and the unit vector which represents that the direction of the blood vessel at the vascular point is in the north-south direction can be obtained after the normalization processing is performed on the direction vector.

S305, constructing a target K-dimensional tree according to at least one vascular point and unit vectors corresponding to the vascular points.

In the embodiment of the invention, at least one discrete vascular point and at least one unit vector corresponding to the vascular point are respectively corresponded in a mode of constructing the K-dimensional tree, so that the target K-dimensional tree is constructed.

S306, determining corresponding points corresponding to the blood vessel points from the at least one blood vessel point according to the target K-dimensional tree aiming at each blood vessel point in the at least one blood vessel point.

In the embodiment of the invention, for each vascular point in at least one vascular point, the vascular point with the association relationship with the vascular point can be searched in at least one vascular point in the target K-dimensional tree as the corresponding point.

S307, determining an attention sequence according to the corresponding points corresponding to the at least one blood vessel point.

S308, adjusting the blood vessel classification model according to the attention sequence.

According to the technical scheme, for each vascular point in at least one vascular point, the direction vector of the vascular point is normalized to obtain a unit vector; constructing a target K-dimensional tree according to at least one vascular point and unit vectors corresponding to the vascular points respectively; and determining a corresponding point corresponding to the blood vessel point from the at least one blood vessel point according to the target K-dimensional tree aiming at each blood vessel point in the at least one blood vessel point. According to the technical scheme provided by the embodiment of the invention, the dimension influence of the subsequent adjustment of the blood vessel classification model can be eliminated by obtaining the unit vector of normalization processing, and the speed of determining the corresponding point can be increased by constructing the target K-dimensional tree, so that the speed of determining the attention sequence is increased, and the time for adjusting the blood vessel classification model is saved.

Fig. 5 is a flowchart of a method for adjusting a blood vessel classification model according to another embodiment of the present invention. The present embodiment is optimized based on the above technical solutions. In this embodiment, optionally, according to the target K-dimensional tree, determining a corresponding point corresponding to the vascular point from at least one vascular point includes: taking the vascular point as a current point; starting from the current point, moving a preset step length along the direction represented by the unit vector of the current point, determining a target alternative point from at least one vascular point in the target K-dimensional tree according to the obtained moving result, and updating the target alternative point to the current point; repeatedly executing the step of moving a preset step length along the direction represented by the unit vector of the current point from the current point; and under the condition that the number of the determined target candidate points is greater than or equal to the preset number, taking each determined target candidate point as a corresponding point corresponding to the vascular point. Wherein, the explanation of the same or corresponding terms as the above embodiments is not repeated herein.

Referring to fig. 5, the method of this embodiment may specifically include the following steps:

s401, acquiring a blood vessel point cloud image, a blood vessel classification model and a trained feature extraction model, wherein the blood vessel point cloud image comprises at least one blood vessel point.

S402, inputting the vessel point cloud image into a feature extraction model, and obtaining point features corresponding to at least one vessel point respectively according to an output result of the feature extraction model.

S403, determining a direction vector of each vascular point in the at least one vascular point according to the point characteristics corresponding to the vascular point.

S404, carrying out normalization processing on the direction vector of the vascular point to obtain a unit vector.

S405, constructing a target K-dimensional tree according to at least one vascular point and unit vectors corresponding to the vascular points.

S406, regarding each of the at least one vascular point, regarding the vascular point as a current point.

Wherein the current point is a vessel point of the at least one vessel point from which the current demand determines the target candidate point. The target candidate point is a vascular point of the at least one vascular point candidate as a corresponding point.

S407, starting from the current point, moving a preset step length along the direction represented by the unit vector of the current point, determining a target alternative point from at least one blood vessel point in the target K-dimensional tree according to the obtained movement result, and updating the target alternative point to the current point.

The preset step length is a preset moving step length of the current point; the preset step length can be set according to the requirements, for example, the classification accuracy of the blood vessel classification model according to the requirements can be set.

It should be noted that, starting from the current point, a movement result obtained by moving a preset step length along a direction represented by a unit vector of the current point cannot directly point to a target candidate point, and because a vessel point cloud exists discretely, a vessel point may exist or may not exist at a position pointed to by the movement result, in the embodiment of the present invention, instead of directly taking the obtained movement result as the target candidate point, the target candidate point may be determined from at least one vessel point in the target K-dimensional tree according to the obtained movement result, and the target candidate point is updated to the current point.

S408, returning to the step S407, and taking each determined target candidate point as a corresponding point corresponding to the vascular point when the number of the determined target candidate points is greater than or equal to the preset number.

The preset number is the number of corresponding points corresponding to the blood vessel points obtained by the preset requirement; the preset number may be set according to the requirement, for example, may be set according to the classification accuracy of the blood vessel classification model.

In the embodiment of the invention, when the number of the determined target candidate points is greater than or equal to the preset number, namely, the corresponding points corresponding to the blood vessel points which are required to be determined are all determined, the determined target candidate points can be used as the corresponding points corresponding to the blood vessel points.

S409, determining an attention sequence according to the corresponding points corresponding to the at least one blood vessel point respectively.

S410, adjusting the blood vessel classification model according to the attention sequence.

According to the technical scheme, the vascular point is used as the current point; starting from the current point, moving a preset step length along the direction represented by the unit vector of the current point, determining a target alternative point from at least one vascular point in the target K-dimensional tree according to the obtained moving result, and updating the target alternative point to the current point; repeatedly executing the step of moving a preset step length along the direction represented by the unit vector of the current point from the current point; and under the condition that the number of the determined target candidate points is greater than or equal to the preset number, taking each determined target candidate point as a corresponding point corresponding to the vascular point. According to the technical scheme provided by the embodiment of the invention, the target candidate point can be determined from at least one vascular point in the target K-dimensional tree according to the movement result obtained by moving the preset step length along the direction represented by the unit vector of the current point, so that the corresponding point corresponding to the vascular point can be determined more accurately.

An optional technical solution, after determining the target candidate point from at least one vascular point in the target K-dimensional tree, further includes: according to the obtained movement result, determining alternative information of a target alternative point from a target K-dimensional tree; after each determined target candidate point is used as a corresponding point corresponding to the vascular point, the method further comprises the following steps: according to the alternative information of each target alternative point, determining the corresponding information of the corresponding point corresponding to the vascular point; determining an attention sequence according to corresponding points corresponding to at least one blood vessel point respectively, wherein the attention sequence comprises the following steps: and determining the attention sequence according to the corresponding point corresponding to the at least one vascular point and the corresponding information of the corresponding point corresponding to the at least one vascular point.

It should be noted that, because the target K-dimensional tree includes not only at least one vascular point, but also at least one vascular point corresponding to relevant information related to the vascular point, so that in order to make the accuracy of the follow-up determination of the attention sequence higher, the characteristic of the attention sequence characterization is more, and according to the obtained movement result, the candidate information of the target candidate point can be determined from the target K-dimensional tree, where the candidate information is relevant information related to the target candidate point; according to the alternative information of each target alternative point, corresponding information of a corresponding point corresponding to the vascular point is determined, wherein the corresponding information is related information and the like related to the corresponding point; and determining the attention sequence according to the corresponding point corresponding to the at least one vascular point and the corresponding information of the corresponding point corresponding to the at least one vascular point. The relevant information related to the self may include, for example, a unit vector, and information such as a label and/or a position set for the self based on blood vessel points adjacent to each other in front of and behind the unit vector, and in the embodiment of the present invention, the content of the relevant information of the self is not specifically limited.

In the embodiment of the invention, the attention sequence is determined according to the corresponding point corresponding to at least one blood vessel point and the corresponding information of the corresponding point corresponding to at least one blood vessel point, so that the attention sequence obtained by determination can represent more blood vessel point characteristics, such as the characteristics of the directional relation among the blood vessel points, the characteristics of the front-back corresponding relation among the blood vessel points, and the like, thereby improving the classification accuracy of the blood vessel classification model obtained by adjusting the attention sequence.

Another alternative solution, taking the vessel point as the current point, includes: taking the blood vessel point as a starting point and taking the starting point as a current point; moving a preset step length along the direction represented by the unit vector of the current point from the current point, and determining a target alternative point from at least one vascular point in the target K-dimensional tree according to the obtained moving result, wherein the method comprises the following steps: in the case that the current point is a starting point, determining a positive direction according to the unit vector of the current point, and taking a direction opposite to the positive direction as a negative direction; starting from the current point, moving a preset step length along the positive direction, and determining a first alternative point from at least one vascular point in the target K-dimensional tree according to the obtained movement result; starting from the current point, moving a preset step length along the negative direction, and determining a second alternative point from at least one vascular point in the target K-dimensional tree according to the obtained movement result; taking the first candidate point and the second candidate point as target candidate points; under the condition that the current point is not a starting point, determining the current direction of the current point according to the current direction of the history point and the unit vector of the current point, wherein the history point is a blood vessel point which is the nearest current point before the current point; starting from the current point, moving a preset step length along the current direction of the current point, and determining a third alternative point from at least one vascular point in the target K-dimensional tree according to the obtained moving result; and taking the third alternative point as a target alternative point.

The starting point is a blood vessel point which starts to determine a target alternative point by taking the point as the starting point.

It should be noted that, since the unit vector may only be able to characterize the trend of the blood vessel at the vascular point, but not necessarily point in a specific direction, for example, the unit vector characterizes the trend of the blood vessel at the vascular point as a north-south trend, but cannot refer to the blood vessel as a south direction, that is, it may be understood that the unit vector does not characterize the flow direction of the blood in the blood vessel, and thus, in the case that the current point is the starting point, there may be two directions that the unit vector can characterize to determine the target candidate point. According to the characteristics of the unit vector, in the embodiment of the present invention, the positive direction may be determined according to the unit vector of the current point, where the current point is the starting point, the manner of determining the positive direction according to the unit vector of the current point is not specifically limited, a fixed direction may be taken as the positive direction, or any one direction from two directions that can be represented by the unit vector may be taken as the positive direction, and a direction opposite to the positive direction may be taken as the negative direction, for example, where the unit vector represents that the direction of the blood vessel at the vascular point is a north-south direction, the positive direction may be a south direction and the negative direction may be a north direction, or the positive direction may be a north direction and the negative direction may be a south direction, that is, the positive direction and the negative direction may correspond to the two directions that can be represented by the unit vector, respectively.

In the embodiment of the invention, a preset step length can be moved from a current point along a positive direction, and according to an obtained movement result, for example, the positive direction and the preset step length can be multiplied, the movement is performed from the current point according to the multiplication result, a first alternative point is determined from at least one vascular point in a target K-dimensional tree, and the first alternative point is a target alternative point determined based on the positive direction; moving a preset step length along a negative direction from a current point, for example, multiplying the negative direction by the preset step length, moving according to a multiplication result from the current point, and determining a second alternative point from at least one vascular point in a target K-dimensional tree according to the obtained movement result, wherein the second alternative point is a target alternative point determined based on the negative direction; the first candidate point and the second candidate point are taken as target candidate points.

It should be noted that, in the case that the current point is not the starting point, that is, the current point is already the current point where the general direction trend is determined, if the same manner as that of determining the target candidate point by the starting point is the same, the situation that the target candidate point is repeatedly determined or erroneously determined may be caused, so in the embodiment of the present invention, in the case that the current point is not the starting point, the current direction of the current point may be determined according to the current direction of the history point which is the current point recently before the current point and the unit vector of the current point, for example, the history point is the starting point, the unit vector of the history point represents the direction of the blood vessel at the blood vessel point as the north-south trend, the current direction of the current point may be determined according to the current direction of the history point as the south direction and the unit vector of the current point, if the unit vector of the current point is also represented as the north-south trend. The current direction is the direction in which the current point needs to move.

In the embodiment of the invention, a preset step length is moved from a current point along the current direction of the current point, for example, the current direction and the preset step length can be multiplied, the movement is performed according to the multiplication result from the current point, and a third alternative point is determined from at least one vascular point in the target K-dimensional tree according to the obtained movement result, wherein the third alternative point is the target alternative point determined based on the current direction; and taking the third alternative point as a target alternative point.

In the embodiment of the invention, the situation that the target candidate points are repeatedly determined or erroneously determined can be avoided by setting different target candidate points for the current point serving as the starting point and for the current point not serving as the starting point, so that the accuracy of the attention sequence obtained by subsequent determination is improved.

Still another alternative solution, starting from the current point, moving a preset step length along a direction represented by a unit vector of the current point, and determining a target candidate point from at least one vascular point in the target K-dimensional tree according to the obtained movement result, including: starting from the current point, moving a preset step length along the direction represented by the unit vector of the current point, and determining expected information corresponding to the current point according to the obtained movement result; and taking the vascular point which is determined from the target K-dimensional tree and matches with the expected information as a target candidate point.

It should be noted that, since the vascular point cloud exists discretely, the moving result obtained by moving the preset step length along the direction represented by the unit vector of the current point from the current point cannot directly point to the target candidate point, and may point to a position where no vascular point exists, even point to an area outside the vascular point cloud, but the position to which the moving result points is a position of the target candidate point in the ideal case, so that the moving result may start from the current point, move the preset step length along the direction represented by the unit vector of the current point, and determine the expected information corresponding to the current point according to the obtained moving result, where the expected information is relevant information that may represent the target candidate point in the ideal case, for example, the expected information may include the position and/or distance of the target candidate point in the ideal case compared with the current point, the position and/or distance of the target candidate point in the ideal case, the minimum value obtained according to the filtering algorithm in the ideal case of the target candidate point, and/or the like.

In the embodiment of the present invention, a blood vessel point matching with the expected information may be determined from at least one blood vessel point in the target K-dimensional tree according to the expected information, for example, a blood vessel point having the highest matching degree with the expected information may be determined from at least one blood vessel point in the target K-dimensional tree according to the expected information as a target candidate point, and the criterion of the matching degree between the expected information and the blood vessel point may be, for example, the matching degree between the position represented by the expected information and the position of the blood vessel point, and further, the matching degree between the gray scale represented by the expected information and the gray scale of the blood vessel point may be, for example, and in the embodiment of the present invention, the criterion of the matching degree between the expected information and the blood vessel point is not limited specifically.

In the embodiment of the invention, the target candidate points can be determined based on the expected information, and the target candidate points with higher accuracy meeting the requirements can be determined.

For better understanding of the technical solution of the embodiment of the present invention described above, an alternative example is provided herein. Illustratively, referring to fig. 6, a vessel point cloud image is acquired along with a trained feature extraction model; inputting the vessel point cloud image into a feature extraction model, and obtaining at least one point feature corresponding to the vessel point according to the output result of the feature extraction model; inputting the point characteristics corresponding to at least one blood vessel point respectively to a filtering module adopting a Franagi filtering algorithm to obtain direction vectors corresponding to at least one blood vessel point respectively; an attention sequence is determined based on the direction vectors respectively corresponding to the at least one vessel point.

Fig. 7 is a block diagram of a device for adjusting a blood vessel classification model according to an embodiment of the present invention, where the device is configured to execute the method for adjusting a blood vessel classification model according to any of the above embodiments. The device and the method for adjusting the blood vessel classification model in the above embodiments belong to the same inventive concept, and reference may be made to the embodiment of the method for adjusting the blood vessel classification model for details which are not described in detail in the embodiment of the device for adjusting the blood vessel classification model. Referring to fig. 7, the apparatus may specifically include: the feature extraction model acquisition module 510, the point feature acquisition module 520, the direction vector determination module 530, and the blood vessel classification model adjustment module 540.

The feature extraction model obtaining module 510 is configured to obtain a vessel point cloud image, a vessel classification model, and a trained feature extraction model, where the vessel point cloud image includes at least one vessel point;

the point feature obtaining module 520 is configured to input the vessel point cloud image into a feature extraction model, and obtain point features corresponding to at least one vessel point respectively according to an output result of the feature extraction model;

a direction vector determining module 530, configured to determine, for each of the at least one vascular point, a direction vector of the vascular point according to a point feature corresponding to the vascular point;

The blood vessel classification model adjustment module 540 is configured to determine an attention sequence based on at least one blood vessel point and a direction vector corresponding to the at least one blood vessel point, and adjust the blood vessel classification model according to the attention sequence.

Optionally, the blood vessel classification model adjustment module 540 may include:

a corresponding point determining sub-module, configured to determine, for each of the at least one vascular point, a corresponding point corresponding to the vascular point from the at least one vascular point based on the at least one vascular point and a direction vector corresponding to the at least one vascular point, respectively;

and the attention sequence determination submodule is used for determining an attention sequence according to corresponding points corresponding to at least one blood vessel point respectively.

Optionally, on the basis of the above device, the device may further include:

the unit vector obtaining module is used for carrying out normalization processing on the direction vector of the vascular points aiming at each vascular point in the at least one vascular point to obtain a unit vector;

the target K-dimensional tree construction module is used for constructing a target K-dimensional tree according to at least one vascular point and unit vectors corresponding to the vascular points respectively;

the corresponding point determination submodule may include:

And the corresponding point determining unit is used for determining corresponding points corresponding to the blood vessel points from the at least one blood vessel points according to the target K-dimensional tree aiming at each of the at least one blood vessel points.

Optionally, on the basis of the above apparatus, the corresponding point determining unit may include:

the current point is used as a subunit, and is used for taking the vascular point as the current point;

the current point updating subunit is used for moving a preset step length along the direction represented by the unit vector of the current point from the current point, determining a target alternative point from at least one vascular point in the target K-dimensional tree according to the obtained moving result, and updating the target alternative point to the current point;

a repeated execution subunit, configured to repeatedly execute a step of moving a preset step length along a direction represented by a unit vector of a current point from the current point;

the corresponding points are used as sub-units and used for taking the determined target candidate points as corresponding points corresponding to the blood vessel points under the condition that the number of the determined target candidate points is larger than or equal to the preset number.

Optionally, on the basis of the above device, the device may further include:

the candidate information determining module is used for determining candidate information of the target candidate point from the target K-dimensional tree according to the obtained moving result after determining the target candidate point from at least one vascular point in the target K-dimensional tree;

The corresponding information determining module is used for determining the corresponding information of the corresponding points corresponding to the blood vessel points according to the alternative information of each target alternative point after the determined target alternative points are used as the corresponding points corresponding to the blood vessel points;

the attention sequence determination submodule may include:

and the attention sequence determining unit is used for determining an attention sequence according to the corresponding points respectively corresponding to the at least one vascular point and the corresponding information of the corresponding points respectively corresponding to the at least one vascular point.

Optionally, on the basis of the above device, the current point serves as a subunit, and may specifically be used for:

taking the blood vessel point as a starting point and taking the starting point as a current point;

the current point updating subunit may specifically be configured to:

in the case that the current point is a starting point, determining a positive direction according to the unit vector of the current point, and taking a direction opposite to the positive direction as a negative direction;

starting from the current point, moving a preset step length along the positive direction, and determining a first alternative point from at least one vascular point in the target K-dimensional tree according to the obtained movement result;

starting from the current point, moving a preset step length along the negative direction, and determining a second alternative point from at least one vascular point in the target K-dimensional tree according to the obtained movement result;

Taking the first candidate point and the second candidate point as target candidate points;

under the condition that the current point is not a starting point, determining the current direction of the current point according to the current direction of the history point and the unit vector of the current point, wherein the history point is a blood vessel point which is the nearest current point before the current point;

starting from the current point, moving a preset step length along the current direction of the current point, and determining a third alternative point from at least one vascular point in the target K-dimensional tree according to the obtained moving result;

and taking the third alternative point as a target alternative point.

Optionally, on the basis of the above device, the current point updating subunit may be specifically configured to:

starting from the current point, moving a preset step length along the direction represented by the unit vector of the current point, and determining expected information corresponding to the current point according to the obtained movement result;

and taking the vascular point which is determined from the target K-dimensional tree and matches with the expected information as a target candidate point.

Optionally, the direction vector determining module 530 may include:

the feature vector obtaining sub-module is used for processing the point features corresponding to the blood vessel points based on a target filtering algorithm to obtain at least one feature value and feature vectors corresponding to the at least one feature value respectively;

The direction vector is used as a sub-module and is used for taking the feature vector corresponding to the feature value with the smallest value in at least one feature value as the direction vector of the vascular point.

Optionally, the blood vessel classification model comprises a blood vessel classification module and a blood vessel staining module;

the apparatus may further include:

the blood vessel classification model updating module is used for updating the blood vessel classification model according to the obtained adjustment result after the blood vessel classification model is adjusted according to the attention sequence;

the target blood vessel image input module is used for acquiring a target blood vessel image and inputting the target blood vessel image into the blood vessel classification module;

the blood vessel dyeing result obtaining module is used for inputting the output result of the blood vessel classifying module into the blood vessel dyeing module to obtain the blood vessel dyeing result aiming at the target blood vessel image.

Optionally, the feature extraction model is implemented based on a Unet network, which at least includes a sparse convolution layer.

The adjusting device of the blood vessel classification model provided by the embodiment of the invention is used for acquiring a blood vessel point cloud image, a blood vessel classification model and a trained characteristic extraction model through a characteristic extraction model acquisition module, wherein the blood vessel point cloud image comprises at least one blood vessel point; inputting the vessel point cloud image into a feature extraction model through a point feature obtaining module, and obtaining at least one point feature respectively corresponding to the vessel points according to the output result of the feature extraction model; determining, by a direction vector determining module, for each of the at least one vessel point, a direction vector of the vessel point according to a point feature corresponding to the vessel point; and determining an attention sequence based on at least one blood vessel point and a direction vector corresponding to the at least one blood vessel point respectively through a blood vessel classification model adjusting module, and adjusting the blood vessel classification model according to the attention sequence. According to the device, the blood vessel classification model is adjusted according to the attention sequence of the direction of the blood vessel, so that the attention of the direction of the blood vessel can be introduced into the blood vessel classification model, and the classification accuracy of the blood vessel classification model can be improved.

The device for adjusting the blood vessel classification model provided by the embodiment of the invention can execute the method for adjusting the blood vessel classification model provided by any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the execution method.

It should be noted that, in the embodiment of the adjusting device of the blood vessel classification model, each included unit and module are only divided according to the functional logic, but not limited to the above-mentioned division, so long as the corresponding functions can be implemented; in addition, the specific names of the functional units are also only for distinguishing from each other, and are not used to limit the protection scope of the present invention.

Fig. 8 shows a schematic diagram of the structure of an electronic device 10 that may be used to implement an embodiment of the invention. Electronic devices are intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. Electronic equipment may also represent various forms of mobile devices, such as personal digital processing, cellular telephones, 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 meant to be exemplary only, and are not meant to limit implementations of the inventions described and/or claimed herein.

As shown in fig. 8, the electronic device 10 includes at least one processor 11, and a memory, such as a Read Only Memory (ROM) 12, a Random Access Memory (RAM) 13, etc., communicatively connected to the at least one processor 11, in which the memory stores a computer program executable by the at least one processor, and the processor 11 may perform various appropriate actions and processes according to the computer program stored in the Read Only Memory (ROM) 12 or the computer program loaded from the storage unit 18 into the Random Access Memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 may also be stored. The processor 11, the ROM 12 and the RAM 13 are connected to each other via a bus 14. An input/output (I/O) interface 15 is also connected to bus 14.

Various components in the electronic device 10 are connected to the I/O interface 15, including: an input unit 16 such as a keyboard, a mouse, etc.; an output unit 17 such as various types of displays, speakers, and the like; a storage unit 18 such as a magnetic disk, an optical disk, or the like; and a communication unit 19 such as a network card, modem, wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information/data with other devices via a computer network, such as the internet, and/or various telecommunication networks.

The processor 11 may be a variety of general and/or special purpose processing components having 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 specialized Artificial Intelligence (AI) computing chips, various processors running machine learning model algorithms, digital Signal Processors (DSPs), and any suitable processor, controller, microcontroller, etc. The processor 11 performs the various methods and processes described above, such as the adjustment method of the blood vessel classification model.

In some embodiments, the method of adapting the vessel classification model may be implemented as a computer program, which is tangibly embodied on a computer-readable storage medium, such as the storage unit 18. In some embodiments, part or all of the computer program may be loaded and/or installed onto the electronic device 10 via the ROM 12 and/or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the above-described adjustment method of the blood vessel classification model may be performed. Alternatively, in other embodiments, the processor 11 may be configured to perform the adjustment method of the vessel classification model by any other suitable means (e.g. by means of firmware).

Various implementations of the systems and techniques described here above may be implemented in digital electronic circuitry, integrated circuit systems, field Programmable Gate Arrays (FPGAs), application Specific Integrated Circuits (ASICs), application Specific Standard Products (ASSPs), systems On Chip (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and/or combinations thereof. These various embodiments may include: implemented in one or more computer programs, the one or more computer programs may be executed and/or interpreted on a programmable system including at least one programmable processor, which may be a special purpose or general-purpose programmable processor, that may receive data and instructions from, and transmit data and instructions to, a storage system, at least one input device, and at least one output device.

A computer program for carrying out 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, special purpose computer, or other programmable data processing apparatus, such that the computer programs, when executed by the processor, cause the functions/acts specified in the flowchart and/or block diagram block or blocks to be implemented. The computer program may execute entirely on the machine, partly on the machine, as a stand-alone software package, partly on the machine and partly on a remote machine or entirely on the remote machine or server.

In the context of the present invention, a computer-readable storage medium may be a tangible medium that can contain, or store a computer program for use by or in connection with an instruction execution system, apparatus, or device. The computer readable storage medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. Alternatively, the computer readable storage medium may be a machine readable signal medium. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

To provide for interaction with a user, the systems and techniques described here 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 a user; and a keyboard and a pointing device (e.g., a mouse or a trackball) through which a user can provide input to the electronic device. Other kinds of devices may also be used to provide for interaction with a user; for example, feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form, including acoustic input, speech input, or tactile input.

The systems and techniques described here can be implemented in a computing system that includes a background component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front-end component (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such background, middleware, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: local Area Networks (LANs), wide Area Networks (WANs), blockchain networks, and the internet.

The computing system may include clients and servers. The client and server are typically remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. The server can be a cloud server, also called a cloud computing server or a cloud host, and is a host product in a cloud computing service system, so that the defects of high management difficulty and weak service expansibility in the traditional physical hosts and VPS service are overcome.

It should be appreciated that various forms of the flows shown above may be used to reorder, add, or delete steps. For example, the steps described in the present invention may be performed in parallel, sequentially, or in a different order, so long as the desired results of the technical solution of the present invention are achieved, and the present invention is not limited herein.

The above embodiments do not limit the scope of the present invention. It will be apparent to those skilled in the art that various modifications, combinations, sub-combinations and alternatives are possible, depending on design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the scope of the present invention.

Claims

1. A method for adjusting a blood vessel classification model, comprising:

inputting the vessel point cloud image into the feature extraction model, and obtaining point features corresponding to the at least one vessel point respectively according to the output result of the feature extraction model;

For each vascular point in the at least one vascular point, determining a direction vector of the vascular point according to a point characteristic corresponding to the vascular point;

and determining an attention sequence based on the at least one blood vessel point and the direction vector corresponding to the at least one blood vessel point respectively, and adjusting the blood vessel classification model according to the attention sequence.

2. The method of claim 1, wherein the determining an attention sequence based on the at least one vascular point and the direction vector to which the at least one vascular point corresponds, respectively, comprises:

for each vascular point in the at least one vascular point, determining a corresponding point corresponding to the vascular point from the at least one vascular point based on the at least one vascular point and a direction vector corresponding to the at least one vascular point respectively;

and determining an attention sequence according to the corresponding points corresponding to the at least one blood vessel point respectively.

3. The method as recited in claim 2, further comprising:

for each vascular point in the at least one vascular point, carrying out normalization processing on the direction vector of the vascular point to obtain a unit vector;

Constructing a target K-dimensional tree according to the at least one vascular point and the unit vectors corresponding to the at least one vascular point respectively;

the determining, for each of the at least one vascular point, a corresponding point corresponding to the vascular point from the at least one vascular point based on the at least one vascular point and a direction vector corresponding to the at least one vascular point, respectively, includes:

and determining a corresponding point corresponding to each vascular point in the at least one vascular point from the at least one vascular point according to the target K-dimensional tree.

4. A method according to claim 3, wherein said determining, from said at least one vascular point, a corresponding point corresponding to said vascular point according to said target K-dimensional tree, comprises:

taking the vascular point as a current point;

starting from the current point, moving a preset step length along the direction represented by the unit vector of the current point, determining a target alternative point from at least one vascular point in the target K-dimensional tree according to the obtained movement result, and updating the target alternative point to the current point;

repeatedly executing the step of moving a preset step length along the direction represented by the unit vector of the current point from the current point;

And under the condition that the number of the determined target candidate points is greater than or equal to the preset number, taking each determined target candidate point as a corresponding point corresponding to the blood vessel point.

5. The method of claim 4, further comprising, after said determining a target candidate point from at least one vessel point in said target K-dimensional tree:

according to the obtained movement result, determining alternative information of the target alternative points from the target K-dimensional tree;

after each determined target candidate point is taken as the corresponding point corresponding to the blood vessel point, the method further comprises the following steps:

according to the alternative information of each target alternative point, corresponding information of corresponding points corresponding to the vascular points is determined;

the determining the attention sequence according to the corresponding points corresponding to the at least one vascular point respectively comprises the following steps:

and determining an attention sequence according to the corresponding points respectively corresponding to the at least one vascular point and the corresponding information of the corresponding points respectively corresponding to the at least one vascular point.

6. The method of claim 4, wherein said regarding said vessel point as a current point comprises:

The step of moving a preset step length along the direction represented by the unit vector of the current point from the current point, and determining a target alternative point from at least one vascular point in the target K-dimensional tree according to the obtained movement result comprises the following steps:

determining a positive direction according to a unit vector of the current point and taking a direction opposite to the positive direction as a negative direction in the case that the current point is the starting point;

determining the current direction of the current point according to the current direction of a history point and the unit vector of the current point under the condition that the current point is not the starting point, wherein the history point is a blood vessel point which is the nearest current point before the current point;

Starting from the current point, moving a preset step length along the current direction of the current point, and determining a third alternative point from at least one vascular point in the target K-dimensional tree according to the obtained movement result;

and taking the third alternative point as a target alternative point.

7. The method of claim 4, wherein moving from the current point by a preset step length along a direction represented by a unit vector of the current point, and determining a target candidate point from at least one vessel point in the target K-dimensional tree according to the obtained movement result, comprises:

8. The method of claim 1, wherein determining the direction vector of the vessel point based on the point feature corresponding to the vessel point comprises:

processing point features corresponding to the vascular points based on a target filtering algorithm to obtain at least one feature value and feature vectors corresponding to the at least one feature value respectively;

And taking the feature vector corresponding to the feature value with the smallest value in the at least one feature value as the direction vector of the vascular point.

9. The method of claim 1, wherein the vessel classification model comprises a vessel classification module and a vessel staining module;

after said adapting said vessel classification model according to said attention sequence, further comprising:

updating the blood vessel classification model according to the obtained adjustment result;

acquiring a target blood vessel image, and inputting the target blood vessel image into the blood vessel classification module;

and inputting the output result of the blood vessel classification module into the blood vessel staining module to obtain a blood vessel staining result aiming at the target blood vessel image.

10. The method according to any of claims 1-9, wherein the feature extraction model is implemented based on a Unet network comprising at least a sparse convolution layer.

11. An adjustment device for a blood vessel classification model, comprising:

The point feature obtaining module is used for inputting the blood vessel point cloud image into the feature extraction model, and obtaining point features corresponding to the at least one blood vessel point respectively according to the output result of the feature extraction model;

a direction vector determining module, configured to determine, for each vascular point in the at least one vascular point, a direction vector of the vascular point according to a point feature corresponding to the vascular point;

and the blood vessel classification model adjustment module is used for determining an attention sequence based on the at least one blood vessel point and the direction vector corresponding to the at least one blood vessel point respectively, and adjusting the blood vessel classification model according to the attention sequence.

12. An electronic device, comprising:

at least one processor; and

a memory communicatively coupled to the at least one processor; wherein,,

the memory stores a computer program executable by the at least one processor to cause the at least one processor to perform the method of adapting a blood vessel classification model according to any one of claims 1-10.

13. A computer-readable storage medium, characterized in that the computer-readable storage medium stores computer instructions for causing a processor to execute a method for adapting a blood vessel classification model according to any one of claims 1-10.