CN116664592A

CN116664592A - Image-based arteriovenous blood vessel separation method and device, electronic equipment and medium

Info

Publication number: CN116664592A
Application number: CN202310475345.1A
Authority: CN
Inventors: 张潇月
Original assignee: Ping An Technology Shenzhen Co Ltd
Current assignee: Ping An Technology Shenzhen Co Ltd
Priority date: 2023-04-26
Filing date: 2023-04-26
Publication date: 2023-08-29

Abstract

The invention relates to artificial intelligence and digital medical treatment, and provides an image-based arteriovenous vascular separation method, an image-based arteriovenous vascular separation device, electronic equipment and a medium, wherein a vascular skeleton in a CT image is extracted; dividing the blood vessel into a plurality of blood vessel segments according to the blood vessel skeleton; constructing a vessel topology map based on the plurality of vessel segments; extracting a plurality of features of each of the vessel segments; training to obtain an arteriovenous vascular separation model based on a plurality of characteristics of each vascular segment; inputting the CT image to be processed into the training-completed arteriovenous vascular separation model to perform arteriovenous vascular separation. The invention can improve the accuracy of arteriovenous vessel separation.

Description

Image-based arteriovenous blood vessel separation method and device, electronic equipment and medium

Technical Field

The invention relates to the technical field of artificial intelligence, in particular to an image-based arteriovenous vessel separation method, an image-based arteriovenous vessel separation device, electronic equipment and a medium.

Background

The fundus color photograph is an imaging mode for non-intrusively observing the state of human eye microvasculature, not only can intuitively reflect the condition of fundus focus such as hemorrhage, hard exudation and the like, but also can be used for observing morphological changes of anatomical structures such as width, bending degree and the like of fundus blood vessels. With the development of medical imaging technology, a series of fundus artery and vein segmentation technologies based on deep learning are developed through fundus color photographic images.

Previous studies often used full convolution networks (Fully Convolutional Networks, FCN), U-net, etc. for performing vein segmentation on fundus images, but the disadvantage was that: (1) It is difficult to distinguish between arteries and veins, and the same vessel is prone to misleading and inconsistent veins (where a section of a vessel is part of a vein is part of an artery). (2) some fine blood vessels are easily missed. (3) the generalization capability of the image of different devices is weak.

Disclosure of Invention

In view of the foregoing, it is desirable to provide an image-based arteriovenous vessel separation method, apparatus, electronic device, and medium that can improve the accuracy of the image-based arteriovenous vessel separation.

A first aspect of the present invention provides an image-based arteriovenous vessel separation method, the method comprising:

extracting a vascular skeleton in the CT image;

dividing the blood vessel into a plurality of blood vessel segments according to the blood vessel skeleton;

constructing a vessel topology map based on the plurality of vessel segments;

extracting a plurality of features of each of the vessel segments;

training to obtain an arteriovenous vascular separation model based on a plurality of characteristics of each vascular segment;

inputting the CT image to be processed into the training-completed arteriovenous vascular separation model to perform arteriovenous vascular separation.

According to an alternative embodiment of the present invention, the extracting the vascular skeleton in the CT image includes:

dividing the CT image by using a preset blood vessel division model to obtain a binarized blood vessel mask image;

and refining the vascular mask image to obtain the vascular skeleton.

According to an optional embodiment of the invention, the refining the vessel mask image to obtain the vessel skeleton comprises:

performing filtering operation on the vascular mask image to obtain a filtered image;

extracting an initial vascular skeleton from the filtered image through a refinement algorithm;

fitting the initial vascular skeleton to obtain a continuous vascular skeleton;

and carrying out single pixelation treatment on the continuous vascular skeleton to obtain the vascular skeleton.

According to an alternative embodiment of the present invention, the dividing the blood vessel into a plurality of blood vessel segments according to the blood vessel skeleton comprises:

acquiring a vascular branch point in the vascular skeleton;

dividing the blood vessel skeleton by taking the blood vessel branch point as a dividing point to obtain a plurality of blood vessel segments; or (b)

Deleting the blood vessel branch points to split the blood vessel skeleton into a plurality of branches, and determining each branch as a blood vessel segment to obtain a plurality of blood vessel segments.

According to an alternative embodiment of the present invention, said extracting a plurality of features of each of said vessel segments comprises:

acquiring a tight packing frame of each blood vessel segment;

acquiring a feature map output by the preset vessel segmentation model, and extracting a first feature corresponding to the tight packing frame from the feature map;

normalizing the first feature to obtain a normalized feature;

and adopting a plurality of preset feature extraction models for each blood vessel segment, and calculating a plurality of second features according to the tight packing frames corresponding to the blood vessel segments in the feature map.

According to an optional embodiment of the invention, the training to obtain the arteriovenous vascular separation model based on the vascular topological structure diagram and the plurality of features of each vascular segment comprises:

obtaining overall characteristics according to the normalized characteristics and the second characteristics corresponding to each blood vessel segment;

inputting the vessel topological structure diagram and a plurality of overall characteristics into a preset neural network, and acquiring a prediction label of each vessel segment output by the preset neural network;

calculating a first loss function value according to the preset label and the corresponding real label;

Calculating a second loss function value according to the blood vessel mask image and the gold standard image corresponding to the CT image;

and training the arteriovenous vascular separation model and the preset vascular segmentation model based on the first loss function value and the second loss function value by adopting a gradient descent algorithm to obtain a trained arteriovenous vascular separation model and a trained vascular segmentation model.

According to an alternative embodiment of the invention, after said obtaining arterial and venous vessels, the method further comprises:

obtaining a maximum arterial vessel and a minimum arterial vessel;

calculating the equivalent value of the central retinal artery diameter according to the caliber of the maximum arterial vessel and the caliber of the minimum arterial vessel;

obtaining a maximum vein and a minimum vein;

calculating the equivalent value of the central retinal vein caliber according to the caliber of the maximum vein vessel and the caliber of the minimum vein vessel;

and calculating to obtain an arteriovenous vessel caliber quantification value according to the equivalent value of the central retinal artery caliber and the equivalent value of the central retinal vein caliber.

A second aspect of the present invention provides an image-based arteriovenous vascular separation device, the device comprising:

The extraction module is used for extracting the vascular skeleton in the CT image;

a segmentation module for segmenting the blood vessel into a plurality of blood vessel segments according to the blood vessel skeleton;

a building module for building a vessel topology map based on the plurality of vessel segments;

a computing module for extracting a plurality of features of each of the vessel segments;

the training module is used for training to obtain an arteriovenous vascular separation model based on the multiple characteristics of each vascular segment;

and the separation module is used for inputting the CT image to be processed into the trained arteriovenous vascular separation model to perform arteriovenous vascular separation.

A third aspect of the invention provides an electronic device comprising a processor and a memory, the processor being arranged to implement the image-based arteriovenous vessel separation method when executing a computer program stored in the memory.

A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, implements the image-based arteriovenous vessel separation method.

According to the image-based arteriovenous blood vessel separation method, the device, the electronic equipment and the medium, the deep neural network and the graphic neural network (GCN) are combined for combined training, so that the classification of blood vessels is more accurate, and the classification between arteriovenous is simpler and more accurate; the blood vessel is classified based on the binary mask blood vessel image of the whole blood vessel, so that the problem of inconsistent artery and vein caused by the wrong division of the same blood vessel can be avoided, namely, one part of a blood vessel is a vein and the other part of the blood vessel is an artery, and the continuity of the arterial blood vessel and the venous blood vessel is ensured; second, some thin blood vessels can be avoided from leaking out.

Drawings

Fig. 1 is a flowchart of an image-based arteriovenous vessel separation method according to an embodiment of the present invention.

Fig. 2 is a block diagram of an image-based arteriovenous vascular separation device according to a second embodiment of the present invention.

Fig. 3 is a schematic structural diagram of an electronic device according to a third embodiment of the present invention.

Detailed Description

In order that the above-recited objects, features and advantages of the present invention will be more clearly understood, a more particular description of the invention will be rendered by reference to specific embodiments thereof which are illustrated in the appended drawings. Embodiments of the invention and features of the embodiments may be combined with each other without conflict.

Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. The terminology used herein in the description of the invention is for the purpose of describing embodiments only in an alternative embodiment only and is not intended to be limiting of the invention.

The image-based arteriovenous blood vessel separation method provided by the embodiment of the invention is executed by the electronic equipment, and correspondingly, the image-based arteriovenous blood vessel separation device runs in the electronic equipment.

The embodiment of the invention can perform standardized processing on the data based on the artificial intelligence technology. Among these, artificial intelligence (Artificial Intelligence, AI) is the theory, method, technique and application system that uses a digital computer or a digital computer-controlled machine to simulate, extend and extend human intelligence, sense the environment, acquire knowledge and use knowledge to obtain optimal results.

Artificial intelligence infrastructure technologies generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technologies, operation/interaction systems, mechatronics, and the like. The artificial intelligence software technology mainly comprises a computer vision technology, a robot technology, a biological recognition technology, a voice processing technology, a natural language processing technology, machine learning/deep learning and other directions.

Arteriovenous separation has unique advantages in different medical scenarios and in a variety of diseases or pathological conditions. In disease studies, features such as vessel density, structure and size can be used to assess disease, while pulmonary vascular disease may specifically affect arteries or veins through different physiological mechanisms. Therefore, the research on arteries and veins can show complex anatomical relations between lung lesions and vascular systems, which is beneficial to improving the accuracy of lung disease diagnosis. The artery and vein separation is beneficial to assisting in early screening and tracking of lung diseases, doctors can diagnose and track pathological states of patients, and effective references can be provided for preoperative planning, intra-operative navigation, postoperative evaluation and the like of lung disease operations, so that the method has great clinical significance. The fundus is the only part of the whole body which can directly observe arteries, veins and capillaries by naked eyes, and a large number of long-term follow-up researches show that quantitative indexes (central artery and vein equivalent, artery and vein fractal dimension, artery and vein ratio and the like) of fundus artery and vein blood vessels are obviously related to systemic chronic diseases (hypertension, diabetes, cardiovascular and cerebrovascular diseases and the like). The segmentation and extraction technology of the fundus artery and vein blood vessel is a basic stone for realizing the automatic quantification of fundus artery and vein. Therefore, an effective arteriovenous extraction method is an urgent need in the field of medical imaging.

Example 1

Fig. 1 is a flowchart of an image-based arteriovenous vessel separation method according to an embodiment of the present invention. The image-based arteriovenous blood vessel separation method specifically comprises the following steps, the sequence of the steps in the flow chart can be changed according to different requirements, and some of the steps can be omitted.

S11, extracting a vascular skeleton in the CT image.

The CT image is a fundus image obtained by scanning the fundus of the patient by using an electronic computed tomography (Computed Tomography, CT) technique, or may be a lung image obtained by scanning the lung of the patient by using an electronic computed tomography (Computed Tomography, CT) technique. The ocular disease is identified by separating retinal arteries and veins in the fundus CT image, or the pulmonary artery and veins in the lung CT image.

Wherein the CT image may be acquired by the electronic device from a digital medical database. The digital medical database may be a digital database storing patient cases in a certain hospital, and the preset medical database may also be a networking database of a plurality of hospitals, which is not limited by the present invention.

The electronics can acquire CT images of a plurality of patients, each of which can correspond to one or more CT images. And training an arteriovenous blood vessel separation model based on the acquired CT images, so that the trained arteriovenous blood vessel separation model is used for carrying out blood vessel segmentation, namely arteriovenous separation.

In an alternative embodiment, the extracting the vascular skeleton in the CT image includes:

and refining the vascular mask image to obtain the vascular skeleton.

The preset vessel segmentation model may be a model obtained by training based on a deep learning network, such as a U-net network, a full convolution network (Fully Convolutional Networks, FCN), and the like. The process of vessel segmentation model is prior art and the present invention is not described in detail here.

The electronic device uses a preset blood vessel segmentation model to segment the CT image, namely, blood vessels in the CT image are segmented from the background without distinguishing transfer veins. The gray value of the pixel point in the binarized vascular mask image is 0 or 255, namely, a black-and-white image. The white areas in the binarized vessel mask image represent vessels and the black areas represent the image background surrounding the vessels. The binarized vascular mask image obtained by the embodiment can reduce the data volume of the CT image, thereby facilitating extraction of a vascular skeleton, i.e. highlighting the outline of a blood vessel.

In an optional embodiment, the thinning the vascular mask image to obtain the vascular skeleton includes:

fitting the initial vascular skeleton to obtain a continuous vascular skeleton;

The vascular skeleton is a topological description of the geometric features of the blood vessel. The vascular skeleton can reflect the connection state, structural information and direction information of the blood vessel. The vascular skeleton is often located in the center of the blood vessel, also called the vessel centerline. The vascular skeleton is understood to mean the central axis of a blood vessel.

The electronic device can perform median filtering operation on the vascular mask image, and bifurcation possibly occurring at the tail end of the vascular skeleton in the vascular mask image can be removed after the median filtering operation. Median filtering is a nonlinear smoothing technique that sets the gray value of each pixel to the median of the gray values of all pixels within a certain neighborhood window of that point.

And extracting the vascular skeleton of the vascular mask image subjected to the filtering operation by using a thinning algorithm, so that redundant boundary points are removed, and important image nodes such as joints, endpoints and isolated points are reserved. The electronic device may refine the vessel mask image using a morphological refinement algorithm to extract a vessel skeleton, which is referred to as an initial vessel skeleton. In some examples, the morphological refinement algorithm may include, but is not limited to, a hildrich refinement algorithm, a pavilidis refinement algorithm, a Rosenfeld refinement algorithm, or the like. The morphological refinement algorithm is prior art and the present invention is not described in detail herein.

Because the initial vascular skeleton consists of discrete pixel points, the electronic device fits the initial vascular skeleton to obtain a continuous vascular skeleton. In some embodiments, the initial vessel skeleton may be fitted using a least squares cubic spline interpolation algorithm.

And because the morphological refinement algorithm cannot ensure that the extracted vascular skeleton with single pixel is not favorable for measuring the vascular diameter, the initial vascular skeleton extracted by the morphological refinement algorithm needs to be subjected to further single-pixelation treatment. In the specific implementation, the width of the blood vessel of the continuous blood vessel framework is thinned to be one pixel width along the central direction of the blood vessel, so that a single-pixel blood vessel framework is formed, and the basic topological structure of the blood vessel shape of the single-pixel blood vessel framework is kept unchanged.

S12, dividing the blood vessel into a plurality of blood vessel segments according to the blood vessel skeleton.

The whole blood vessel skeleton is unfavorable for the separation of the arterial and venous blood vessels, so that the electronic equipment divides the target blood vessel skeleton after extracting the target blood vessel skeleton, thereby obtaining a plurality of blood vessel segments, and each blood vessel segment is either a venous blood vessel or an arterial blood vessel, thereby being favorable for the separation of the arterial and venous blood vessels.

After the electronic equipment extracts the blood vessel skeleton, a capturing tool of a calling point can be used for determining the crossing point of the blood vessel skeleton, and the blood vessel is divided into a plurality of blood vessel segments according to the crossing point, so that a plurality of blood vessel segment images are obtained. The crossing point may be a crossing point of an arterial blood vessel and an arterial blood vessel, a branching point of an arterial blood vessel, a crossing point of a venous blood vessel and a venous blood vessel, a branching point of a venous blood vessel, or a crossing point of an arterial blood vessel and a venous blood vessel, and the type of the crossing point is not particularly limited in the present application.

In an alternative embodiment, the segmenting the vessel into a plurality of vessel segments according to the vessel skeleton comprises:

acquiring a vascular branch point in the vascular skeleton;

The pixels located at the branches have a common feature: there must be three neighbors with the center pixel within the eight neighborhood. According to this feature, a blood vessel branch point in the blood vessel skeleton can be detected using eight neighborhood filtering. The electronic equipment performs eight-neighborhood filtering operation on the image corresponding to the vascular skeleton, then calculates the eight neighborhood number of each pixel point, determines the pixel point with the eight neighborhood number of 3 as a suspicious point, acquires the pixel value of the suspicious point, and judges whether the suspicious point is a vascular branch point or not according to the acquired pixel value. Specifically, when the acquired pixel value is 0, indicating that the suspicious point is a background point, determining that the suspicious point is not a vessel branch point; when the acquired pixel value is 1, which indicates that the suspected point is a central pixel point, the suspected point is determined to be a blood vessel branch point.

S13, constructing a blood vessel topological structure diagram based on the plurality of blood vessel segments.

In this embodiment, each vessel segment is used as a node, if the vessel segments are connected with each other, an undirected edge is established between the two corresponding nodes, and if the vessel segments are not connected with each other, an undirected edge is not established between the two corresponding nodes, so that a vessel topology structure is constructed.

S14, extracting a plurality of characteristics of each blood vessel segment.

After obtaining the plurality of vessel segments, the electronic device may extract a plurality of features of each vessel segment in order to determine whether each vessel segment is a venous vessel segment or an arterial vessel segment, thereby training the graph neural network based on the plurality of features of each vessel segment, and performing a bi-classification on each vessel segment through the graph neural network.

In an alternative embodiment, said extracting a plurality of features of each of said vessel segments comprises:

acquiring a tight packing frame of each blood vessel segment;

acquiring a feature map output by a preset vessel segmentation model, and extracting a first feature corresponding to the tight packing frame from the feature map;

normalizing the first feature to obtain a normalized feature;

Wherein, the tight box refers to the smallest circumscribed rectangle capable of framing each vessel segment in the vessel mask image.

The preset vessel segmentation model is a deep learning model for performing vessel segmentation on the CT image, and a feature map output by the penultimate layer of the preset vessel segmentation model can be acquired, wherein the feature map is consistent with the size of the vessel mask image. The electronic device may acquire first position coordinates of a tight packet frame corresponding to each vessel segment in the vessel mask image, for example, first position coordinates of four vertices of the tight packet frame, then acquire second position coordinates corresponding to the first position coordinates in the feature map, and use a plurality of first features framed by a rectangular frame corresponding to the second position coordinates in the feature map as a feature matrix of the vessel segment corresponding to the tight packet frame.

In order to facilitate the convergence rate and efficiency of the model to be accelerated in the subsequent model training, the electronic equipment performs normalization processing on the feature matrix of each blood vessel segment after obtaining the feature matrix of each blood vessel segment, so as to obtain the normalization feature of the corresponding blood vessel segment. In specific implementation, the average value of all the feature values in the feature matrix can be calculated, then the difference value between each feature value and the average value is calculated, then the square sum of each difference value is calculated, and finally the average value of the square sums is calculated based on all the square sums to be used as the normalized feature.

The plurality of preset feature extraction models are preset calculation models for extracting a plurality of features of the blood vessel segment, for example, a calculation model of length features, a calculation model of pipe diameter features and a calculation model of gray scale features. The length characteristic of the blood vessel segment is the first number of pixels corresponding to the blood vessel segment in the binarized blood vessel mask image, and the pipe diameter characteristic of the blood vessel segment is the ratio of the second number of pixels to the first number in the corresponding tight packing frame in the characteristic diagram.

And S15, training to obtain an arteriovenous blood vessel separation model based on the multiple characteristics of each blood vessel segment.

The electronic device initializes the network architecture of the arteriovenous blood vessel separation model, for example, a graph neural network can be employed as the network architecture of the arteriovenous blood vessel separation model. And simultaneously inputting the constructed vessel topological structure diagram and the extracted multiple characteristics of each vessel segment into a graph neural network for iterative training and performing two-class prediction, wherein the graph neural network predicts the class of each node in the vessel topological structure diagram, namely an arterial vessel or a venous vessel. And (3) obtaining the artery and vein classification of the vascular mask image through classifying the nodes, so as to output an artery and vein separation result.

In an optional embodiment, the training to obtain the arteriovenous vessel separation model based on the vessel topological structure diagram and the plurality of features of each vessel segment includes:

And splicing the normalized feature and the second features corresponding to each blood vessel segment to obtain an overall feature. And taking the vessel topological structure diagram and the plurality of overall characteristics as input of a preset neural network, and outputting a prediction label of each vessel segment through the preset neural network. The predictive label may be an arterial vessel or a venous vessel. The real label is obtained by labeling an expert based on professional medical knowledge. And calculating a first loss function value according to the preset label and the corresponding real label. The first loss function value is used for reflecting the approaching degree of the prediction label and the real label output by the arteriovenous vascular separation model. The larger the first loss function value, the more consistent the predicted tag is with the real tag, and the smaller the first loss function value, the more different the predicted tag is from the real tag. The first loss function value L1 may be calculated using a cross entropy loss function, l1= -SUM (y×log '), where y represents a real label and y' represents a predicted label.

The golden standard image refers to a pre-segmented blood vessel image, for example, an expert outlines the positions of blood vessels to be segmented from an untreated original CT image based on professional medical knowledge, namely the golden standard image is pre-marked with blood vessels. The second loss function value is used for reflecting the approaching degree of the vascular mask image output by the vascular separation model and the corresponding gold standard image. The larger the second loss function value, the more different the vessel mask image is from the corresponding gold standard image, and the smaller the second loss function value, the closer the vessel mask image is to the corresponding gold standard image. The second loss function value L2 may be a euclidean distance or a cosine angle between the vascular mask image and the corresponding gold standard image.

And obtaining a total loss function value based on the sum of the first loss function value and the second loss function value, and performing joint iterative training on the vessel segmentation network and the arteriovenous vessel separation network based on the total loss function value until the total loss function value is converged, and stopping iterative training. Iterative training is a model training mode in deep learning, and is used for optimizing a model. The iterative training implementation process in the step is as follows: in each cycle training process, all training samples are sequentially read in, the current total loss function value is calculated, a random gradient descent algorithm is adopted to determine the gradient descent direction, so that the total loss function value is gradually lowered and reaches a stable state, and the optimization of each parameter in the constructed network model is realized. The convergence of the total loss function value means that the total loss function value approaches 0, for example, less than 0.1.

S16, inputting the CT image to be processed into the trained arteriovenous vascular separation model to perform arteriovenous vascular separation.

The CT image to be processed refers to a fundus CT image or a lung CT image or other medical images which need to be subjected to arteriovenous vessel separation.

Inputting the CT image to be processed into an arteriovenous blood vessel separation model which is completed by training, and outputting classification labels of arterial blood vessel segments and venous blood vessel segments through the arteriovenous blood vessel separation model.

The accurate segmentation effect is required for the disease prediction based on the arteriovenous vessel quantification index, and the arteriovenous vessel segmentation based on deep learning in the past is mostly based on the CNN task, has strong limitation, and leads to insufficient classification precision. According to the embodiment of the invention, the deep neural network and the graphic neural network (GCN) are combined for combined training, so that the classification of blood vessels is more accurate, and the classification between the artery and the vein is simpler and more accurate; the blood vessel is classified based on the binary mask blood vessel image of the whole blood vessel, so that the problem of inconsistent artery and vein caused by the wrong division of the same blood vessel can be avoided, namely, one part of a blood vessel is a vein and the other part of the blood vessel is an artery, and the continuity of the arterial blood vessel and the venous blood vessel is ensured; second, some thin blood vessels can be avoided from leaking out. The embodiment of the invention converts the problem of classifying the veins of each pixel into the problem of classifying the veins of each blood vessel segment, thereby not only overcoming the problems that one blood vessel has arterial pixels and venous pixels respectively possibly caused by the error of a pixel-level classification algorithm, but also avoiding the problems of easily causing erroneous judgment areas of blood vessels such as arterial and venous crossing, blood vessel branching and the like. By extracting the blood vessel skeleton (blood vessel center line) and determining the blood vessel branch points of the blood vessel skeleton, the accuracy of segmenting the blood vessel image to obtain the blood vessel segment image can be improved.

In an alternative embodiment, prior to the extracting the vascular skeleton in the CT image, the method further comprises:

converting the CT image into an image with a preset specification size;

normalizing CT images with preset specification sizes to obtain normalized CT images;

and carrying out enhancement processing on the normalized CT image to obtain an enhanced CT image.

The preset specification size may be 512×512.

The CT image with the preset specification is normalized, so as to obtain the CT image with uniform pixel color. In specific implementation, a CT image with a preset specification size can be subjected to Gaussian blur processing to obtain a Gaussian image, and the obtained Gaussian image and the CT image are reversely overlapped.

Enhancement processing of the normalized CT image may include: performing horizontal overturning according to a preset probability; transpose with a preset probability; random gamma transformation; randomly changing the saturation value of the normalized CT image according to a preset probability; carrying out histogram equalization processing on the normalized CT image; random brightness and contrast adjustment.

Through the optional implementation manner, CT images with different formats and different sizes can be converted into images with preset specification sizes, and meanwhile, the CT images with the preset specification sizes are normalized, so that CT images with uniform pixel colors can be obtained; and (3) enhancing the CT image with uniform pixel color to obtain the CT image with brightness, definition and saturation value reaching the standard.

It should be understood that extracting the vascular skeleton in the CT image includes extracting the vascular skeleton in the enhanced CT image.

In an alternative embodiment, after obtaining the arterial vessel and the venous vessel, the method further comprises:

obtaining a maximum arterial vessel and a minimum arterial vessel;

obtaining a maximum vein and a minimum vein;

The equivalent value (Central Retinal Artery Equivalent, CRAE) of the central retinal artery diameter in this example is calculated using the following formula:

CRAE＝(Ai ² +Aj ² ) ^1/2 ，

wherein Ai is the tube diameter of the largest arterial vessel obtained by iteration, and Aj is the tube diameter of the smallest arterial vessel obtained by iteration.

The equivalent value (Central Retinal Vein Equivalent, CRVE) of the central retinal vein diameter in this example is calculated using the following formula:

CRVE＝(Vi ² +Vj ² ) ^1/2 ，

Wherein Vi is the diameter of the maximum vein obtained by iteration, and Vj is the diameter of the minimum vein obtained by iteration.

The quantitative value of the arteriovenous vessel caliber in the embodiment can be the ratio of the equivalent value of the central retinal artery caliber to the equivalent value of the central retinal vein caliber.

According to the alternative embodiment, the equivalent value of the central retinal artery caliber and the equivalent value of the central retinal vein caliber are obtained by carrying out quantization treatment on the caliber of the arterial vessel and the caliber of the venous vessel, so that the quantized value of the arteriovenous vessel caliber is obtained. Because the separation accuracy of the arterial blood vessel and the venous blood vessel is higher, the accuracy of the obtained caliber of the arterial blood vessel and the caliber of the venous blood vessel is higher, and the accuracy of the obtained quantized value of the caliber of the arterial blood vessel is higher.

Medical science proves that the user suffering from hypertension and the user not suffering from hypertension obviously have differences of microvasculature in fundus images, such as: under repeated high pressure stimulation, the early retinal arterioles in hypertension can be slightly stenosed and slightly hardened; if the blood pressure is increased for a long time and the blood pressure is developed to a certain stage, the retina can be further changed, the artery is continuously narrowed, the retinal arteriosclerosis is obvious, the artery can have silver line reaction, the artery diameter is not narrow, and the artery and vein cross-pressing phenomenon exists; from the above, high blood pressure can affect or reduce the vessel diameter of arteriovenous vessels. That is, the risk of hypertension of the object to be measured can be predicted according to the quantitative value of the arteriovenous vessel diameter.

Example two

Fig. 2 is a block diagram of an image-based arteriovenous vascular separation device according to an embodiment of the present invention.

In some embodiments, the image-based arteriovenous blood vessel separation device 20 can include a plurality of functional modules composed of computer program segments. The computer program of the individual program segments in the image-based arteriovenous vessel separation device 20 can be stored in a memory of an electronic device and executed by at least one processor to perform (see fig. 1 for details) the image-based arteriovenous vessel separation function.

In this embodiment, the image-based arteriovenous blood vessel separation device 20 can be divided into a plurality of functional modules according to the functions performed thereby. The functional module may include: extraction module 201, segmentation module 202, construction module 203, calculation module 204, training module 205, separation module 206, enhancement module 207, and quantization module 208. The module referred to in the present invention refers to a series of computer program segments capable of being executed by at least one processor and of performing a fixed function, stored in a memory. In the present embodiment, the functions of the respective modules will be described in detail in the following embodiments.

The extraction module 201 is configured to extract a vascular skeleton in the CT image.

and refining the vascular mask image to obtain the vascular skeleton.

fitting the initial vascular skeleton to obtain a continuous vascular skeleton;

The segmentation module 202 is configured to segment the blood vessel into a plurality of blood vessel segments according to the blood vessel skeleton.

acquiring a vascular branch point in the vascular skeleton;

The construction module 203 constructs a vessel topology map based on the plurality of vessel segments.

The computing module 204 is configured to extract a plurality of features of each of the vessel segments.

acquiring a tight packing frame of each blood vessel segment;

Normalizing the first feature to obtain a normalized feature;

The training module 205 is configured to train to obtain an arteriovenous vascular separation model based on the vascular topological structure diagram and the multiple features of each of the vascular segments.

The separation module 206 is configured to input the CT image to be processed into a trained arteriovenous blood vessel separation model for arteriovenous blood vessel separation.

In an alternative embodiment, the enhancing module 207 is configured to convert the CT image into an image with a preset specification size before the extracting the vascular skeleton in the CT image; normalizing CT images with preset specification sizes to obtain normalized CT images; and carrying out enhancement processing on the normalized CT image to obtain an enhanced CT image.

The preset specification size may be 512×512.

The quantification module 208 is configured to obtain an arteriovenous vessel caliber quantification value after obtaining an arterial vessel and a venous vessel.

In an alternative embodiment, the obtaining the quantified value of the arteriovenous vessel diameter includes:

obtaining a maximum arterial vessel and a minimum arterial vessel;

obtaining a maximum vein and a minimum vein;

CRAE＝(Ai ² +Aj ² ) ^1/2 ，

CRVE＝(Vi ² +Vj ² ) ^1/2 ，

Example III

The present embodiment provides a computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the steps in the above-described embodiment of an image-based arteriovenous vessel separation method, such as S11-S16 shown in fig. 1:

s11, extracting a vascular skeleton in a CT image;

s12, dividing the blood vessel into a plurality of blood vessel segments according to the blood vessel skeleton;

s13, constructing a blood vessel topological structure diagram based on the plurality of blood vessel segments;

s14, extracting a plurality of characteristics of each blood vessel segment;

s15, training to obtain an arteriovenous blood vessel separation model based on a plurality of characteristics of each blood vessel segment;

Alternatively, the computer program, when executed by a processor, performs the functions of the modules/units in the above-described apparatus embodiments, e.g., modules 201-206 in fig. 2:

the extraction module 201 is configured to extract a vascular skeleton in the CT image;

the segmentation module 202 is configured to segment the blood vessel into a plurality of blood vessel segments according to the blood vessel skeleton;

The constructing module 203 is configured to construct a vascular topological structure diagram based on the plurality of vascular segments;

the computing module 204 is configured to extract a plurality of features of each of the vessel segments;

the training module 205 is configured to train to obtain an arteriovenous vascular separation model based on the vascular topological structure diagram and a plurality of features of each of the vascular segments;

The computer program, when executed by the processor, further implements the enhancement module 207 and the quantization module 208 in the above-described apparatus embodiments, and particularly please refer to fig. 2 and the related description.

Example IV

Fig. 3 is a schematic structural diagram of an electronic device according to an embodiment of the present invention. In the preferred embodiment of the invention, the electronic device 3 comprises a memory 31, at least one processor 32, at least one communication bus 33 and a transceiver 34.

It will be appreciated by those skilled in the art that the configuration of the electronic device shown in fig. 3 is not limiting of the embodiments of the present invention, and that either a bus-type configuration or a star-type configuration is possible, and that the electronic device 3 may also include more or less other hardware or software than that shown, or a different arrangement of components.

In some embodiments, the electronic device 3 is a device capable of automatically performing numerical calculation and/or information processing according to a preset or stored instruction, and its hardware includes, but is not limited to, a microprocessor, an application specific integrated circuit, a programmable gate array, a digital processor, an embedded device, and the like. The electronic device 3 may further include a client device, where the client device includes, but is not limited to, any electronic product that can interact with a client by way of a keyboard, a mouse, a remote control, a touch pad, or a voice control device, such as a personal computer, a tablet computer, a smart phone, a digital camera, etc.

The electronic device 3 is only an example, and other electronic products that may be present in the present invention or may be present in the future, such as those that may be adapted to the present invention, are also included in the scope of the present invention and are incorporated herein by reference.

In some embodiments, the memory 31 has stored therein a computer program which, when executed by the at least one processor 32, performs all or part of the steps in the image-based arteriovenous vessel separation method as described. The Memory 31 includes Read-Only Memory (ROM), programmable Read-Only Memory (PROM), erasable programmable Read-Only Memory (EPROM), one-time programmable Read-Only Memory (One-time Programmable Read-Only Memory, OTPROM), electrically erasable rewritable Read-Only Memory (EEPROM), compact disc Read-Only Memory (Compact Disc Read-Only Memory, CD-ROM) or other optical disc Memory, magnetic tape Memory, or any other medium that can be used for computer-readable carrying or storing data.

Further, the computer-readable storage medium may mainly include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application program required for at least one function, and the like; the storage data area may store data created from the use of blockchain nodes, and the like.

The blockchain is a novel application mode of computer technologies such as distributed data storage, point-to-point transmission, consensus mechanism, encryption algorithm and the like. The Blockchain (Blockchain), which is essentially a decentralised database, is a string of data blocks that are generated by cryptographic means in association, each data block containing a batch of information of network transactions for verifying the validity of the information (anti-counterfeiting) and generating the next block. The blockchain may include a blockchain underlying platform, a platform product services layer, an application services layer, and the like.

In some embodiments, the at least one processor 32 is a Control Unit (Control Unit) of the electronic device 3, connects the various components of the entire electronic device 3 using various interfaces and lines, and performs various functions of the electronic device 3 and processes data by running or executing programs or modules stored in the memory 31, and invoking data stored in the memory 31. For example, the at least one processor 32, when executing the computer program stored in the memory, implements all or part of the steps of the image-based arteriovenous vessel separation method described in embodiments of the present invention; or to perform all or part of the functions of the image-based arteriovenous vascular separation device. The at least one processor 32 may be comprised of integrated circuits, such as a single packaged integrated circuit, or may be comprised of multiple integrated circuits packaged with the same or different functionality, including one or more central processing units (Central Processing unit, CPU), microprocessors, digital processing chips, graphics processors, combinations of various control chips, and the like.

In some embodiments, the at least one communication bus 33 is arranged to enable connected communication between the memory 31 and the at least one processor 32 or the like.

Although not shown, the electronic device 3 may further comprise a power source (such as a battery) for powering the various components, which may preferably be logically connected to the at least one processor 32 via a power management device, such that functions of managing charging, discharging, and power consumption are performed by the power management device. The power supply may also include one or more of any of a direct current or alternating current power supply, recharging device, power failure detection circuit, power converter or inverter, power status indicator, etc. The electronic device 3 may further include various sensors, bluetooth modules, wi-Fi modules, etc., which will not be described herein.

The integrated units implemented in the form of software functional modules described above may be stored in a computer readable storage medium. The software functional modules described above are stored in a storage medium and include instructions for causing a computer device (which may be a personal computer, an electronic device, or a network device, etc.) or a processor (processor) to perform portions of the methods described in the various embodiments of the invention.

In the several embodiments provided by the present invention, it should be understood that the disclosed apparatus and method may be implemented in other manners. For example, the above-described apparatus embodiments are merely illustrative, and for example, the division of the modules is merely a logical function division, and there may be other manners of division when actually implemented.

The modules described as separate components may or may not be physically separate, and components shown as modules may or may not be physical units, may be located in one place, or may be distributed over multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

In addition, each functional module in the embodiments of the present invention may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit. The integrated units can be realized in a form of hardware or a form of hardware and a form of software functional modules.

It will be evident to those skilled in the art that the invention is not limited to the details of the foregoing illustrative embodiments, and that the present invention may be embodied in other specific forms without departing from the spirit or essential characteristics thereof. The present embodiments are, therefore, to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein. Any reference sign in a claim should not be construed as limiting the claim concerned. Furthermore, it will be obvious that the term "comprising" does not exclude other elements or that the singular does not exclude a plurality. Several of the elements or devices recited in the specification may be embodied by one and the same item of software or hardware. The terms first, second, etc. are used to denote a name, but not any particular order.

Finally, it should be noted that the above-mentioned embodiments are merely for illustrating the technical solution of the present invention and not for limiting the same, and although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that modifications and equivalents may be made to the technical solution of the present invention without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. An image-based arteriovenous vessel separation method, the method comprising:

extracting a vascular skeleton in the CT image;

constructing a vessel topology map based on the plurality of vessel segments;

extracting a plurality of features of each of the vessel segments;

2. The image-based arteriovenous blood vessel separation method of claim 1, wherein the extracting the blood vessel skeleton in the CT image comprises:

And refining the vascular mask image to obtain the vascular skeleton.

3. The image-based arteriovenous vessel separation method as set forth in claim 2, wherein the refining the vessel mask image to obtain the vessel skeleton includes:

fitting the initial vascular skeleton to obtain a continuous vascular skeleton;

4. The image-based arteriovenous vessel separation method as set forth in claim 3, wherein the segmenting the vessel into a plurality of vessel segments according to the vessel skeleton comprises:

acquiring a vascular branch point in the vascular skeleton;

5. The image-based arteriovenous blood vessel separation method as set forth in claim 3 wherein said extracting a plurality of features of each of said blood vessel segments comprises:

acquiring a tight packing frame of each blood vessel segment;

normalizing the first feature to obtain a normalized feature;

6. The image-based arteriovenous blood vessel separation method as set forth in claim 5, wherein the training to obtain an arteriovenous blood vessel separation model based on the vessel topology map and the plurality of features of each vessel segment comprises:

7. The image-based arteriovenous vessel separation method as set forth in any one of claims 1 to 6, wherein after the obtaining arterial and venous vessels, the method further comprises:

obtaining a maximum arterial vessel and a minimum arterial vessel;

obtaining a maximum vein and a minimum vein;

8. An image-based arteriovenous vascular separation device, the device comprising:

9. An electronic device comprising a processor and a memory, wherein the processor is configured to implement the image-based arteriovenous vessel separation method as set forth in any one of claims 1 to 7 when executing a computer program stored in the memory.

10. A computer-readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the image-based arteriovenous vessel separation method according to any one of claims 1 to 7.