CN117635950B

CN117635950B - Method, device, electronic equipment and storage medium for vessel segmentation correction processing

Info

Publication number: CN117635950B
Application number: CN202410012978.3A
Authority: CN
Inventors: 李舜禹; 谭文安; 姜祁翎; 梁江荣; 郭志飞; 黄智珩; 黄泽鑫; 安昕
Original assignee: Guangdong Oupu Mandi Technology Co ltd
Current assignee: Guangdong Oupu Mandi Technology Co ltd
Priority date: 2024-01-04
Filing date: 2024-01-04
Publication date: 2024-04-09
Anticipated expiration: 2044-01-04
Also published as: CN117635950A

Abstract

The application provides a method, a device, electronic equipment and a storage medium for blood vessel segmentation correction processing, which are applied to the technical field of image processing, wherein a morphological method is adopted in blood vessel segmentation post-processing of lung lobes, and the purpose of correction is realized by carrying out blood vessel post-processing correction by combining a method of using a three-dimensional space and a two-dimensional plane connected domain, so that the problems that more false recognition phenomena exist at three-level blood vessels and four-level blood vessels and more noise exists around the three-level blood vessels and the four-level blood vessels in blood vessel reconstruction are solved.

Description

Method, device, electronic equipment and storage medium for vessel segmentation correction processing

Technical Field

The present disclosure relates to the field of image processing technologies, and in particular, to a method, an apparatus, an electronic device, and a storage medium for vascular segmentation correction processing.

Background

Blood vessels are the most complex tissue in the human body, compared with massive tissue such as lung lobes, liver and gall. The complex structure of the blood vessels adds a number of difficulties to the data labelling personnel and there are phenomena of missed and false marks at some specific complex areas such as the level four blood vessels. These factors lead to the fact that deep learning models cannot develop their strong learning ability 100%, and eventually lead to the fact that the models have more misidentification phenomena at the tertiary and quaternary blood vessels and more noise around them when predicting the arteriovenous of the blood vessels.

Therefore, in the existing vascular reconstruction technique, there are many misidentification phenomena at the tertiary blood vessel and the quaternary blood vessel and many problems of noise around.

Disclosure of Invention

In view of the shortcomings of the prior art, the method, the device, the electronic equipment and the storage medium for blood vessel segmentation correction processing are applied to the technical field of image processing, a morphological method is adopted in blood vessel segmentation post-processing of lung lobes, the purpose of correction is achieved by carrying out blood vessel post-processing correction by combining a method of using a three-dimensional space and a two-dimensional plane connected domain, and the problems that more misidentification phenomena exist at three-level blood vessels and four-level blood vessels and more noise exists around the blood vessels in blood vessel reconstruction are solved.

In a first aspect, the present application provides a method of vascular segmentation correction processing, the method comprising the steps of:

s1: acquiring blood vessel data information of CT scanning lung lobes;

s2: the blood vessel data information is imported into a blood vessel segmentation model to obtain segmented blood vessel segmentation results, wherein the blood vessel segmentation results comprise vein blood vessels and artery blood vessels;

s3: acquiring a three-dimensional maximum connected domain of any one of the venous blood vessel and the arterial blood vessel in the blood vessel segmentation result, and correcting the other blood vessel connected with the three-dimensional maximum connected domain of any one blood vessel to obtain a first correction result;

S4: and acquiring a two-dimensional maximum connected domain of any one of the venous blood vessel and the arterial blood vessel in each slice of the first correction result, and correcting the other blood vessel connected with the two-dimensional maximum connected domain of any one blood vessel to obtain a second correction result.

According to the blood vessel segmentation correction processing method, the blood vessel CT image of the lung lobes can be presented by acquiring the blood vessel data information of the CT scanning lung lobes, so that the follow-up three-dimensional reconstruction processing is facilitated; after the blood vessel data information is obtained, the blood vessel data information is imported into a blood vessel segmentation model, blood vessels are segmented, blood vessel segmentation results are obtained, the blood vessel segmentation results comprise vein blood vessels and artery blood vessels, and because the blood vessel segmentation results are wrongly recognized and noisy due to the problem of data labeling during blood vessel segmentation, in order to solve the problem, noise filtering operation can be carried out to eliminate noise during blood vessel segmentation, blood vessel segmentation results with eliminated noise are obtained, then the part of the vein blood vessel in the blood vessel segmentation results, which is wrongly recognized as the artery blood vessel, can be corrected as the vein blood vessel, and the part of the artery blood vessel in the blood vessel segmentation results, which is wrongly recognized as the vein blood vessel, can be corrected as the artery blood vessel, so that the obtained first correction result can be eliminated to a certain extent, but the position relation among all the communicating domains can not be covered by using a three-dimensional space communicating domain method, further can be traversed through the vein blood vessel or the artery blood vessel in each layer of the first correction result, the part of the vein blood vessel, which is wrongly recognized as the artery blood vessel, the part in the vein blood vessel can be corrected as the vein blood vessel, and the second correction result can be corrected in a three-dimensional basis, and the three-dimensional correction result can be further reconstructed. Therefore, the method for correcting the segmentation of the blood vessel solves the problems that more misidentification exists at the three-level blood vessel and the four-level blood vessel in the reconstruction of the blood vessel and more noise exists around the blood vessel.

Further, the step S2 includes:

s21: the blood vessel data information is imported into a blood vessel segmentation model to obtain an initial blood vessel segmentation result;

s22: acquiring a lung lobe profile of the lung lobe;

s23: combining the initial blood vessel segmentation result with the lung lobe contour, and removing the part, outside the lung lobe contour, of the initial blood vessel segmentation result to obtain the blood vessel segmentation result.

According to the blood vessel segmentation correction processing method, blood vessel data information is imported into a blood vessel segmentation model to obtain an initial blood vessel segmentation result, the initial blood vessel segmentation result is not subjected to noise removal processing, more noise is generated, the three-dimensional reconstruction result is easy to influence, the blood vessel segmentation correction processing method can be combined with the lung lobe profile of a lung lobe, and the part, located outside the lung lobe profile, of the initial blood vessel segmentation result is removed to obtain a blood vessel segmentation result, so that the problem of noise interference in the three-dimensional reconstruction of the blood vessel is solved.

Further, the step S22 includes:

s221: acquiring lung lobe data information of CT scanning lung lobes;

s222: importing the lung lobe data information into a lightweight semantic segmentation model to obtain segmented lung lobe segmentation results;

S223: traversing each slice of the lung lobe segmentation result, and extracting the lung lobe profile.

According to the blood vessel segmentation correction processing method, the lung lobe data information of CT scanning lung lobes is imported into the lightweight semantic segmentation model, and as the lung lobes are not complex tissues, the effect of rapidly segmenting lung areas can be achieved by using the lightweight semantic segmentation model, and then lung impeller contours can be extracted by using the lightweight semantic segmentation model to traverse each slice of lung lobe segmentation results.

Further, the step S3 includes:

s31: acquiring a first three-dimensional maximum connected domain of the venous blood vessel and a set of first non-maximum connected domains of the arterial blood vessel in the blood vessel segmentation result;

s32: correcting the part connected with the first three-dimensional maximum communicating domain in the first non-maximum communicating domain set into the vein to obtain a corrected second non-maximum communicating domain set;

s33: combining the first three-dimensional maximum connected domain, the second non-maximum connected domain and other vein blood vessels to obtain three-dimensional corrected vein blood vessels;

s34: and obtaining the first correction result according to the corrected venous blood vessel.

According to the method for blood vessel segmentation correction processing, the first three-dimensional maximum connected domain of the venous blood vessel and the first non-maximum connected domain of the arterial blood vessel in the blood vessel segmentation result are obtained, the part, connected with the first three-dimensional maximum connected domain, of the first non-maximum connected domain is corrected to be the venous blood vessel, namely the part, which is mistakenly identified as the arterial blood vessel, of the venous blood vessel can be corrected to be the venous blood vessel, so that the corrected second non-maximum connected domain is obtained, the first three-dimensional maximum connected domain, the second non-maximum connected domain and other venous blood vessels are combined, the three-dimensional corrected venous blood vessel can be obtained, and the first correction result can be obtained according to the three-dimensional corrected venous blood vessel. The aim of eliminating the false recognition problem in the three-dimensional reconstruction of the blood vessel is fulfilled by adopting a method of communicating the three-dimensional space.

Further, the step S34 includes:

s341: acquiring a second three-dimensional maximum connected domain of the arterial blood vessel and a collection of a third non-maximum connected domain of the venous blood vessel in the blood vessel segmentation result;

s342: correcting the part connected with the second three-dimensional maximum communicating domain in the third non-maximum communicating domain set into the arterial vessel to obtain a corrected fourth non-maximum communicating domain set;

S343: combining the second three-dimensional maximum connected domain, the fourth non-maximum connected domain and other arterial blood vessels to obtain three-dimensional corrected arterial blood vessels;

s344: and obtaining the first correction result according to the three-dimensional corrected venous blood vessel and the three-dimensional corrected arterial blood vessel.

Further, the step S4 includes:

s41: acquiring a set of first two-dimensional maximum connected domains of the venous blood vessel and fifth non-maximum connected domains of the arterial blood vessel in each slice of the first correction result;

s42: correcting the part connected with the first two-dimensional maximum connected domain in the set of the fifth non-maximum connected domain into the venous blood vessel to obtain a corrected set of the sixth non-maximum connected domain;

s43: combining the first two-dimensional maximum connected domain, the set of sixth non-maximum connected domain and other vein blood vessels to obtain two-dimensional corrected vein blood vessels;

s44: and obtaining the second correction result according to the two-dimensional corrected venous blood vessel.

Further, the step S44 includes:

s441: acquiring a second two-dimensional maximum connected domain of the arterial blood vessel and a set of seventh non-maximum connected domains of the venous blood vessel in each slice of the first correction result;

S442: correcting the part connected with the second two-dimensional maximum communicating domain in the set of the seventh non-maximum communicating domain into the arterial vessel to obtain a corrected set of eighth non-maximum communicating domain;

s443: combining the second two-dimensional maximum connected domain, the set of eighth non-maximum connected domain and other arterial blood vessels to obtain two-dimensional corrected arterial blood vessels;

s444: and obtaining the second correction result according to the two-dimensional corrected venous blood vessel and the two-dimensional corrected arterial blood vessel.

In a second aspect, the present application proposes an apparatus for a vessel segmentation correction procedure, the apparatus comprising:

the acquisition module is used for: the method comprises the steps of acquiring blood vessel data information of CT scanning lung lobes;

and a segmentation module: the blood vessel data information is used for guiding the blood vessel data information into a blood vessel segmentation model to obtain a segmented blood vessel segmentation result, wherein the blood vessel segmentation result comprises a venous blood vessel and an arterial blood vessel;

a first correction module: the method comprises the steps of obtaining a three-dimensional maximum connected domain of any one of the venous blood vessel and the arterial blood vessel in a blood vessel segmentation result, and correcting the other blood vessel connected with the three-dimensional maximum connected domain of any one blood vessel to obtain a first correction result;

A second correction module: and the two-dimensional maximum connected domain of any one of the vein blood vessel and the artery blood vessel in each slice used for obtaining the first correction result is used for correcting the other blood vessel connected with the two-dimensional maximum connected domain of any one of the blood vessel, so as to obtain a second correction result.

In a third aspect, the present application provides an electronic device comprising a processor and a memory storing computer readable instructions which, when executed by the processor, perform the steps of any of the methods described above.

In a fourth aspect, the present application provides a computer readable storage medium having stored thereon a computer program which when executed by a processor performs the steps of any of the methods described above.

The beneficial effects are that: according to the blood vessel segmentation correction processing method, device, electronic equipment and storage medium, the blood vessel CT image of the lung lobe can be presented by acquiring the blood vessel data information of the CT scanning lung lobe, so that the follow-up three-dimensional reconstruction processing is facilitated; after the blood vessel data information is obtained, the blood vessel data information is imported into a blood vessel segmentation model, blood vessels are segmented, blood vessel segmentation results are obtained, the blood vessel segmentation results comprise vein blood vessels and artery blood vessels, and because the blood vessel segmentation results are wrongly recognized and noisy due to the problem of data labeling during blood vessel segmentation, in order to solve the problem, noise filtering operation can be carried out to eliminate noise during blood vessel segmentation, blood vessel segmentation results with eliminated noise are obtained, then the part of the vein blood vessel in the blood vessel segmentation results, which is wrongly recognized as the artery blood vessel, can be corrected as the vein blood vessel, and the part of the artery blood vessel in the blood vessel segmentation results, which is wrongly recognized as the vein blood vessel, can be corrected as the artery blood vessel, so that the obtained first correction result can be eliminated to a certain extent, but the position relation among all the communicating domains cannot be covered by using a three-dimensional space communicating domain method, the vein blood vessel or the artery blood vessel in each layer of the first correction result can be traversed, the part of the vein blood vessel, which is wrongly recognized as the artery blood vessel can be corrected, the second correction result can be corrected as the vein blood vessel in a three-dimensional basis, and the three-dimensional correction result can be further reconstructed. Therefore, the method for correcting the segmentation of the blood vessel solves the problems that more misidentification exists at the three-level blood vessel and the four-level blood vessel in the reconstruction of the blood vessel and more noise exists around the blood vessel.

Drawings

Fig. 1 is a flowchart of a method of a vessel segmentation correction process according to the present application.

Fig. 2 is a block diagram showing a method of vascular segmentation correction processing according to the present application.

Fig. 3 is a schematic structural diagram of an electronic device provided in the present application.

Fig. 4 is a comparison diagram before and after denoising the blood vessel segmentation result in the method of blood vessel segmentation correction processing proposed in the present application.

Fig. 5 is a comparison chart of the blood vessel segmentation result before and after three-dimensional space connected domain correction in the blood vessel segmentation correction processing method provided by the application.

Fig. 6 is a comparison chart of the first correction result before and after two-dimensional spatial connected domain correction in the method for blood vessel segmentation correction processing provided by the application.

Description of the reference numerals: 201. an acquisition module; 202. a segmentation module; 203. a first correction module; 204. a second correction module; 301. a processor; 302. a memory; 303. a communication bus; 3. an electronic device.

Detailed Description

The following description of the embodiments of the present application will be made clearly and completely with reference to the drawings in the embodiments of the present application, and it is apparent that the described embodiments are only some embodiments of the present application, not all embodiments. The components of the embodiments of the present application, as generally described and illustrated in the figures herein, may be arranged and designed in a wide variety of different configurations. Thus, the following detailed description of the embodiments of the present application, as provided in the accompanying drawings, is not intended to limit the scope of the application, as claimed, but is merely representative of selected embodiments of the application. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present application without making any inventive effort, are intended to be within the scope of the present application.

It should be noted that: like reference numerals and letters denote like items in the following figures, and thus once an item is defined in one figure, no further definition or explanation thereof is necessary in the following figures. Meanwhile, in the description of the present application, the terms "first", "second", and the like are used only for distinguishing the description, and are not to be construed as indicating or implying relative importance.

The following disclosure provides many different embodiments or examples for achieving the objects of the present invention, solving the problems of the existing three-dimensional reconstruction model of blood vessels that there are more misidentification phenomena at the tertiary and quaternary blood vessels and more noise around them when predicting the arteriovenous of the blood vessel.

Referring to fig. 1, the present application provides a method for correcting a blood vessel segmentation, the method comprising the steps of:

s1: acquiring blood vessel data information of CT scanning lung lobes;

s3: acquiring a three-dimensional maximum connected domain of any one of a venous vessel and an arterial vessel in a vessel segmentation result, and correcting the other vessel connected with the three-dimensional maximum connected domain of any one vessel to obtain a first correction result;

S4: and obtaining a two-dimensional maximum connected domain of any one of the vein blood vessel and the artery blood vessel in each slice of the first correction result, and correcting the other blood vessel connected with the two-dimensional maximum connected domain of any one of the blood vessels to obtain a second correction result.

In practical application, in step S1, the lung lobe includes two sets of vascular systems, i.e. a venous blood vessel and an arterial blood vessel, and the blood vessel data information of the CT scan is the image information of the projection of the venous blood vessel and the arterial blood vessel at the lung lobe position on the two-dimensional plane.

In practical application, in step S2, the vessel segmentation model may realize segmentation of the medical image scanned by CT, and may be established in advance for a technician, where the vessel segmentation model has a plurality of algorithms, including a segmentation algorithm based on a threshold, a segmentation algorithm based on a clustering technique, a segmentation algorithm based on a deformable model, a segmentation algorithm based on a neural network, and the like, and the technician may establish a model with highest information efficiency for segmenting vessel data according to the actual requirement. The blood vessel data information is imported into a blood vessel segmentation model, so that the venous blood vessel and the arterial blood vessel are segmented, specifically, the segmented venous blood vessel and arterial blood vessel can be endowed with different colors for differentiation, a blood vessel segmentation result is obtained, and a follow-up technician can conveniently reconstruct the blood vessel in three dimensions according to the segmented blood vessel segmentation result. In an ideal situation, a technician hopes that the blood vessel segmented through the blood vessel segmentation model is high in accuracy and clear in boundary, but in practical situations, due to the complex blood vessel structure, a data labeling person is difficult to label the blood vessel completely and accurately, especially at a four-level blood vessel (the blood vessel can be classified into four levels according to the stenosis degree of the blood vessel, and the four-level blood vessel is the narrowest), noise is easy to generate during segmentation, the accuracy of a blood vessel segmentation result is not high, and adhesion and interference of tiny branch blood vessels (the tiny branch blood vessels are blood vessels which do not need to be segmented for three-dimensional reconstruction) exist at the boundary. In order to solve this problem, noise removal processing may be performed during segmentation of the blood vessel. Thus, further, step S2 includes:

S21: leading the blood vessel data information into a blood vessel segmentation model to obtain an initial blood vessel segmentation result;

s22: acquiring a lung lobe profile of a lung lobe;

s23: combining the initial blood vessel segmentation result with the lung lobe profile, and removing the part of the initial blood vessel segmentation result outside the lung lobe profile to obtain a blood vessel segmentation result.

In practical application, the blood vessel data information is imported into a blood vessel segmentation model to obtain an initial blood vessel segmentation result, wherein the initial blood vessel segmentation result is a blood vessel segmentation result without noise removal treatment, and more noise interference exists in an image, so that medical three-dimensional reconstruction is not facilitated. By analyzing the initial blood vessel segmentation result, the region where most of noise around the venous blood vessel and the arterial blood vessel is located outside the lung lobes can be easily obtained, so in order to rapidly perform noise filtering on the initial blood vessel segmentation result, the tiny branch blood vessels at the tail end of the initial blood vessel segmentation result can be eliminated by considering the shape of the lung lobes, and the specific operation mode is as follows: firstly, obtaining the lung lobe outline of a lung lobe, then combining the initial blood vessel segmentation result with the lung lobe outline, and removing the part of the initial blood vessel segmentation result outside the lung lobe outline to obtain the blood vessel segmentation result. Referring to fig. 4, the left image in fig. 4 is an initial blood vessel segmentation result, the white frame marked in the left image is noise, and the right image is a blood vessel segmentation result after the noise removal operation.

Among them, how to quickly obtain an accurate lung lobe profile in the step of obtaining the lung lobe profile of the lung lobe is a practical problem faced by the technician. Typically, the prior art approach to obtaining accurate lung lobe profiles is to reconstruct the lung lobes in three dimensions and then obtain the lung lobe profiles from the three-dimensionally reconstructed lung lobes. However, the method wastes more time, the three-dimensional reconstruction process of the lung lobes is more complex, and the traditional lung lobe contour acquisition method greatly increases the three-dimensional reconstruction time of the blood vessel when in actual clinical use, so that the three-dimensional reconstruction efficiency of the blood vessel is difficult to improve. In the face of this problem, the present application proposes that the lightweight semantic segmentation model may be adopted to directly segment the lung lobes, and then directly extract the lung lobe contours of the segmented lung lobes, without three-dimensional reconstruction of the lung lobes, so as to greatly improve the efficiency of extracting the lung lobe contours, specifically, in some preferred embodiments, step S22 includes:

s221: acquiring lung lobe data information of CT scanning lung lobes;

s223: traversing each slice of the lobe segmentation result to extract lobe contours.

In practical application, the lung lobe data information of the CT scan lung lobe is image information of the lung lobe projected on the two-dimensional plane. Because the lung lobes are not complex tissues, a lightweight semantic segmentation model can be selected to segment the lung lobes, and specifically, the model can be a BiSeNetV2 semantic segmentation model, so that a segmented lung lobe segmentation result is obtained, and the lung lobe segmentation result at the moment is that: the method comprises the steps of obtaining the image information of the lung lobes projected on a two-dimensional plane at different observation angles, segmenting the image information of the lung lobes, obtaining accurate lung lobe contours by the lung lobe image information, traversing each slice of a lung lobe segmentation result by continuously using a lightweight semantic segmentation model, extracting contour information of each slice, and combining to form complete lung lobe contours. According to the method, accurate lung lobe contours can be rapidly extracted without three-dimensional reconstruction of lung lobes, so that the denoising efficiency of an initial blood vessel segmentation result is greatly improved, and the efficiency of three-dimensional reconstruction of blood vessels is further improved.

In practical applications, in step S3, because the vascular tissue structure is complex, there are many misidentification problems in the blood vessel segmentation result obtained after the blood vessel segmentation, for example, identifying a part of the venous blood vessels as arterial blood vessels and identifying a part of the arterial blood vessels as venous blood vessels, which may mislead the doctor to judge when observing the blood vessel condition, and cause serious consequences. Therefore, after the blood vessel is segmented, the blood vessel segmentation result needs to be subjected to an optimized correction process. In the traditional technology of three-dimensional reconstruction of blood vessels, a deep learning technology is required to be adopted to continuously train a blood vessel segmentation model to optimize the blood vessel segmentation model, and the data marking of the blood vessels is perfected as much as possible, so that the false recognition of the blood vessel segmentation model can be reduced as much as possible when the blood vessels are segmented, and an accurate blood vessel segmentation result is obtained. However, the segmentation of blood vessels using deep learning techniques takes a long time and requires the acquisition of a large number of vessel data labels, which is also a great challenge for the data label personnel. In order to solve the problem, the method adopts a morphological mode to correct and optimize the blood vessel segmentation result segmented by the easily-obtained blood vessel segmentation model, and saves time, manpower and material resources for training the blood vessel segmentation model and acquiring data labels.

The specific correction mode is as follows: obtaining a three-dimensional maximum connected domain of any one of a vein blood vessel and an artery blood vessel in a blood vessel segmentation result, wherein the three-dimensional maximum connected domain actually refers to traversing pixels of any one of the vein blood vessel and the artery blood vessel in the blood vessel segmentation result, finding adjacent pixels and connecting the adjacent pixels to form a connected domain, wherein the largest region in all the connected domains is the three-dimensional maximum connected domain. And then correcting the other blood vessel connected with the three-dimensional maximum communication domain of any blood vessel to obtain a first correction result, thereby correcting the part of the vein blood vessel which is mistakenly identified as the arterial blood vessel and correcting the part of the artery blood vessel which is mistakenly identified as the venous blood vessel. In practical applications, two kinds of blood vessels exist at the position of the lung lobe, and only one of the blood vessels (vein blood vessel or arterial blood vessel) can be corrected only to achieve the purpose of partial correction, and in order to achieve more comprehensive correction, further, in some preferred embodiments, step S3 includes:

s31: acquiring a first three-dimensional maximum connected domain of a vein blood vessel and a set of first non-maximum connected domains of an artery blood vessel in a blood vessel segmentation result;

S32: correcting the part connected with the first three-dimensional maximum communicating domain in the first non-maximum communicating domain set into a vein to obtain a corrected second non-maximum communicating domain set;

s33: combining the first three-dimensional maximum connected domain, the second non-maximum connected domain and other venous blood vessels to obtain three-dimensional corrected venous blood vessels;

s34: and obtaining a first correction result according to the corrected vein blood vessel.

In practical application, the specific method for correcting the part of the venous blood vessel which is mistakenly identified as the arterial blood vessel comprises the following steps: acquiring a first three-dimensional maximum connected domain of a vein blood vessel and a set of first non-maximum connected domains of an artery blood vessel in a blood vessel segmentation result; wherein the first three-dimensional maximum connected domain is contained in the three-dimensional maximum connected domain, representing a portion of the connected domain of the vein vessel of the CT scan; the first non-maximum connected domain set represents a set of the connected domains identified as arterial blood vessels in the blood vessel segmentation result, except for the connected domain with the largest area, and includes arterial blood vessels and venous blood vessels erroneously identified as arterial blood vessels. Since the arterial blood vessel and the venous blood vessel are two unconnected space connected domains, the part connected with the first three-dimensional maximum connected domain in the first non-maximum connected domain is the venous blood vessel which is mistakenly identified as the arterial blood vessel, the part is corrected as the venous blood vessel, the corrected second non-maximum connected domain set (the second non-maximum connected domain set is the part which is mistakenly identified as the arterial blood vessel in the first non-maximum connected domain set and is the corrected as the venous blood vessel rear region set and only comprises the venous blood vessel) is combined with the first three-dimensional maximum connected domain and other venous blood vessels (the part which is identified as the venous blood vessel except the first three-dimensional maximum connected domain in the blood vessel segmentation result), and the three-dimensional corrected venous blood vessel can be obtained. The three-dimensional corrected venous blood vessel solves the problem of false recognition of a blood vessel segmentation model to a certain extent. However, in step S33, when the first three-dimensional maximum connected domain, the second non-maximum connected domain, and the other venous blood vessels are combined, the arterial blood vessels that are mistakenly identified as venous blood vessels are further included in the other venous blood vessels, and in order to solve the problem, further, in some preferred embodiments, step S34 includes:

S341: acquiring a second three-dimensional maximum connected domain of an arterial vessel and a collection of a third non-maximum connected domain of a venous vessel in a vessel segmentation result;

s342: correcting the part connected with the second three-dimensional maximum communicating domain in the third non-maximum communicating domain set into an arterial vessel to obtain a corrected fourth non-maximum communicating domain set;

s344: and obtaining a first correction result according to the three-dimensional corrected venous blood vessel and the three-dimensional corrected arterial blood vessel.

In practical application, a second three-dimensional maximum connected domain of an arterial vessel and a collection of a third non-maximum connected domain of a venous vessel in a vessel segmentation result can be obtained; wherein the second three-dimensional maximum connected domain is contained in the three-dimensional maximum connected domain, representing a portion of the connected domain of the arterial vessel of the CT scan; and the set of third non-maximum connected domains, in which the venous blood vessel and the arterial blood vessel erroneously recognized as the venous blood vessel are included, represents a set of other parts than the connected domain having the largest area among the connected domains recognized as the venous blood vessels in the blood vessel division result. The corrected arterial vessel portion is corrected to obtain a corrected fourth non-maximum connected domain set (the fourth non-maximum connected domain set is a set obtained by correcting a portion of the third non-maximum connected domain set, which is mistakenly recognized as a venous vessel, into a post-arterial vessel region set including only arterial vessels), and combining the second three-dimensional maximum connected domain, the fourth non-maximum connected domain set and other arterial vessels to obtain the three-dimensional corrected arterial vessel. By the method, the arterial blood vessel is corrected on the basis of correcting the venous blood vessel, the three-dimensional corrected venous blood vessel and the three-dimensional corrected arterial blood vessel are respectively obtained, and the three-dimensional corrected venous blood vessel and the three-dimensional corrected arterial blood vessel are combined to obtain a first correction result. Referring to fig. 5, the upper and lower images on the left side of fig. 5 are the blood vessel segmentation results before the correction of the three-dimensional connected domain, wherein the white frame is a blood vessel with false recognition, and the upper and lower images on the right side are the first correction results after the correction of the three-dimensional connected domain, wherein the white frame is a blood vessel with false recognition correction.

However, the first correction result is based on correction of three-dimensional connected domains, and in the three-dimensional space, because tissues among blood vessels are staggered, the structure is very complex, and the problem of false recognition at dead angles can not be corrected easily, so that the subsequent three-dimensional reconstruction of the blood vessels is influenced.

Therefore, in order to solve this problem, the false recognition problem existing in the blood vessel of the first correction result may be corrected by combining the two-dimensional connected domain. That is, in step S4, the two-dimensional maximum connected domain of any one of the vein blood vessel and the artery blood vessel in each slice of the first correction result may be obtained, and each slice of the first correction result actually refers to a slice in which the first correction result in the three-dimensional space is imported into the blood vessel segmentation model and output as a two-dimensional form, and the sum of all the two-dimensional forms of slices may constitute the first correction result in the three-dimensional space. The second correction result is obtained by correcting the other blood vessel connected with the two-dimensional maximum connected domain of any blood vessel, and the second correction result is a result of correcting the two-dimensional plane connected domain again on the basis of correcting the three-dimensional space connected domain, so that the problem that false recognition of dead angle positions is not easy to find and correct due to excessively complex blood vessel tissues is solved. Wherein, only one blood vessel (vein or artery) in each slice of the first correction result is corrected only to achieve the purpose of partial correction, and further, in some preferred embodiments, step S4 includes:

S41: acquiring a first two-dimensional maximum connected domain of a venous blood vessel and a set of fifth non-maximum connected domains of an arterial blood vessel in each slice of the first correction result;

s42: correcting the part connected with the first two-dimensional maximum connected domain in the set of the fifth non-maximum connected domain into a venous vessel to obtain a corrected set of the sixth non-maximum connected domain;

s43: combining the first two-dimensional maximum connected domain, the collection of the sixth non-maximum connected domain and other venous blood vessels to obtain two-dimensional corrected venous blood vessels;

s44: and obtaining a second correction result according to the vein blood vessel after the two-dimensional correction.

In practical application, the part of the vein blood vessel in each slice of the first correction result, which is mistakenly identified as the artery blood vessel, is corrected in the following specific manner: acquiring a first two-dimensional maximum connected domain of a venous blood vessel and a set of fifth non-maximum connected domains of an arterial blood vessel in each slice of the first correction result; the first two-dimensional maximum connected domain is contained in the two-dimensional maximum connected domain, the mode of acquiring the first two-dimensional maximum connected domain and the fifth non-maximum connected domain is the same as that of acquiring the three-dimensional space connected domain, the first two-dimensional maximum connected domain refers to the connected domain with the largest area of the identified venous blood vessel in each layer of slice of the first correction result, the set of the fifth non-maximum connected domain refers to the set of the identified arterial blood vessel except for the connected domain with the largest area, the set of the fifth non-maximum connected domain comprises the arterial blood vessel in each layer of slice and the venous blood vessel which is mistakenly identified as the arterial blood vessel, the part connected with the first two-dimensional maximum connected domain is the mistakenly identified part, and the corrected part is corrected, so that the corrected set of the sixth non-maximum connected domain can be easily obtained (the set of the sixth non-maximum connected domain is the set of the mistakenly identified arterial blood vessel is the venous blood vessel which is the set of the post-venous blood vessel, and only the venous blood vessel is included). Combining the first two-dimensional maximum connected domain, the set of sixth non-maximum connected domain and other venous blood vessels to obtain a venous blood vessel after two-dimensional correction, however, in step S43, when the first two-dimensional maximum connected domain, the set of sixth non-maximum connected domain and other venous blood vessels are combined, the other venous blood vessels further include arterial blood vessels that are misidentified as venous blood vessels, so as to solve the problem, further, in some preferred embodiments, step S44 includes:

S441: acquiring a second two-dimensional maximum connected domain of an arterial vessel and a set of seventh non-maximum connected domains of a venous vessel in each slice of the first correction result;

s442: correcting the part connected with the second two-dimensional maximum connected domain in the set of the seventh non-maximum connected domain into an arterial vessel to obtain a corrected set of the eighth non-maximum connected domain;

s443: combining the second two-dimensional maximum connected domain, the eighth non-maximum connected domain and other arterial blood vessels to obtain two-dimensional corrected arterial blood vessels;

s444: and obtaining a second correction result according to the two-dimensional corrected venous blood vessel and the two-dimensional corrected arterial blood vessel.

In practical application, a second two-dimensional maximum connected domain of an arterial vessel and a set of seventh non-maximum connected domains of a venous vessel in each slice of the first correction result can be obtained; wherein the second two-dimensional maximum connected domain is contained in the two-dimensional maximum connected domain, representing a portion of the connected domain of the arterial vessel of the CT scan; and the set of seventh non-maximum connected domains, in which the venous blood vessel and the arterial blood vessel erroneously recognized as the venous blood vessel are included, represents a set of other parts than the connected domain having the largest area among the connected domains recognized as the venous blood vessels in the blood vessel division result. The corrected arterial vessel portion is corrected to obtain a corrected eighth non-maximum connected domain set (the eighth non-maximum connected domain set is a set obtained by correcting a portion of the seventh non-maximum connected domain set, which is mistakenly recognized as a venous vessel, into a post-arterial vessel region set including only arterial vessels), and combining the second two-dimensional maximum connected domain, the eighth non-maximum connected domain set, and other arterial vessels to obtain two-dimensional corrected arterial vessels. By the method, the arterial blood vessel is corrected on the basis of correcting the venous blood vessel, the two-dimensional corrected venous blood vessel and the two-dimensional corrected arterial blood vessel are respectively obtained, and the two-dimensional corrected venous blood vessel and the two-dimensional corrected arterial blood vessel are combined to obtain a second correction result. Referring to fig. 6 specifically, the left-most image in fig. 6 is a slice of one layer in the first correction result, a blood vessel of the false recognition part is marked in the white frame, the middle image is an enlarged image of the false recognition blood vessel marked in the white frame, and the right-most image is a blood vessel image corrected by the two-dimensional plane connected domain. The second correction result is used for further optimizing and correcting the first correction result after the three-dimensional space connected domain correction, so that the problem that more erroneous identification exists at the three-level blood vessel and the four-level blood vessel in the blood vessel reconstruction is solved.

From the above, according to the method for blood vessel segmentation correction processing provided by the application, the blood vessel CT image of the lung lobe can be presented by acquiring the blood vessel data information of the CT scanning lung lobe, so that the follow-up three-dimensional reconstruction processing is facilitated; after the blood vessel data information is obtained, the blood vessel data information is imported into a blood vessel segmentation model, blood vessels are segmented, blood vessel segmentation results are obtained, the blood vessel segmentation results comprise vein blood vessels and artery blood vessels, and because the blood vessel segmentation results are wrongly recognized and noisy due to the problem of data labeling during blood vessel segmentation, in order to solve the problem, noise filtering operation can be carried out to eliminate noise during blood vessel segmentation, blood vessel segmentation results with eliminated noise are obtained, then the part of the vein blood vessel in the blood vessel segmentation results, which is wrongly recognized as the artery blood vessel, can be corrected as the vein blood vessel, and the part of the artery blood vessel in the blood vessel segmentation results, which is wrongly recognized as the vein blood vessel, can be corrected as the artery blood vessel, so that the obtained first correction result can be eliminated to a certain extent, but the position relation among all the communicating domains can not be covered by using a three-dimensional space communicating domain method, further can be traversed through the vein blood vessel or the artery blood vessel in each layer of the first correction result, the part of the vein blood vessel, which is wrongly recognized as the artery blood vessel, the part in the vein blood vessel can be corrected as the vein blood vessel, and the second correction result can be corrected in a three-dimensional basis, and the three-dimensional correction result can be further reconstructed. Therefore, the method for correcting the segmentation of the blood vessel solves the problems that more misidentification exists at the three-level blood vessel and the four-level blood vessel in the reconstruction of the blood vessel and more noise exists around the blood vessel.

Referring to fig. 2, the present application proposes a device for vascular segmentation correction, the device comprising:

the acquisition module 201: the method comprises the steps of acquiring blood vessel data information of CT scanning lung lobes;

the segmentation module 202: the method comprises the steps of importing blood vessel data information into a blood vessel segmentation model to obtain segmented blood vessel segmentation results, wherein the blood vessel segmentation results comprise vein blood vessels and artery blood vessels;

the first correction module 203: the method comprises the steps of obtaining a three-dimensional maximum connected domain of any one of a vein blood vessel and an artery blood vessel in a blood vessel segmentation result, and correcting the other blood vessel connected with the three-dimensional maximum connected domain of any one blood vessel to obtain a first correction result;

the second correction module 204: and the two-dimensional maximum connected domain of any one of the vein blood vessel and the artery blood vessel in each slice for obtaining the first correction result, and correcting the other blood vessel connected with the two-dimensional maximum connected domain of any one of the blood vessels to obtain the second correction result.

In practical application, the acquisition module 201 is an interface module connected to the CT scanner, the acquisition module 201 may be connected to the CT scanner through an electric wire or a radio, the CT scanner may guide the blood vessel data information into the segmentation module 202 through an electric wire after scanning out the blood vessel data information of the lung lobes, where the segmentation module 202 may be a data guiding algorithm which is set by a technician in advance and is connected to a blood vessel segmentation model, the first correction module 203 and the second correction module 204 may be morphological graphic processing libraries which are connected to the segmentation module 202, after the segmentation module 202 outputs the blood vessel segmentation result, the blood vessel segmentation result is guided into the morphological image processing libraries, the algorithm of the three-dimensional space connected domain is used first to perform first correction to obtain a first correction result, and then the algorithm of the two-dimensional space connected domain is used to perform second correction to obtain a final second correction result, thereby achieving the purpose of eliminating the problem that there are more false identifications at the three-stage blood vessels and four-stage blood vessels in the blood vessel reconstruction and there are more noise around.

In practical application, the blood vessel segmentation model connected with the segmentation module 202 may realize segmentation of the medical image scanned by CT, and may be established in advance for a technician, where the blood vessel segmentation model has a plurality of algorithms, including a segmentation algorithm based on a threshold value, a segmentation algorithm based on a clustering technology, a segmentation algorithm based on a deformable model, a segmentation algorithm based on a neural network, and the like, and the technician may establish a model with the highest information efficiency for segmenting blood vessel data according to the actual requirement. The blood vessel data information is imported into a blood vessel segmentation model, so that the venous blood vessel and the arterial blood vessel are segmented, specifically, the segmented venous blood vessel and arterial blood vessel can be endowed with different colors for differentiation, a blood vessel segmentation result is obtained, and a follow-up technician can conveniently reconstruct the blood vessel in three dimensions according to the segmented blood vessel segmentation result. In an ideal situation, a technician hopes that the blood vessel segmented through the blood vessel segmentation model is high in accuracy and clear in boundary, but in practical situations, due to the complex blood vessel structure, a data labeling person is difficult to label the blood vessel completely and accurately, especially at a four-level blood vessel (the blood vessel can be classified into four levels according to the stenosis degree of the blood vessel, and the four-level blood vessel is the narrowest), noise is easy to generate during segmentation, the accuracy of a blood vessel segmentation result is not high, and adhesion and interference of tiny branch blood vessels (the tiny branch blood vessels are blood vessels which do not need to be segmented for three-dimensional reconstruction) exist at the boundary. In order to solve this problem, noise removal processing may be performed during segmentation of the blood vessel.

In practical application, a specific denoising mode is to introduce blood vessel data information into a blood vessel segmentation model to obtain an initial blood vessel segmentation result, wherein the initial blood vessel segmentation result is a blood vessel segmentation result without denoising treatment, and more noise interference exists in an image, so that medical three-dimensional reconstruction is not facilitated. By analyzing the initial blood vessel segmentation result, the region where most of noise around the venous blood vessel and the arterial blood vessel is located outside the lung lobes can be easily obtained, so in order to rapidly perform noise filtering on the initial blood vessel segmentation result, the tiny branch blood vessels at the tail end of the initial blood vessel segmentation result can be eliminated by considering the shape of the lung lobes, and the specific operation mode is as follows: firstly, obtaining the lung lobe outline of a lung lobe, then combining the initial blood vessel segmentation result with the lung lobe outline, and removing the part of the initial blood vessel segmentation result outside the lung lobe outline to obtain the blood vessel segmentation result.

In practical application, the mode of quickly acquiring the lung lobe outline can select to divide the lung lobe by adopting a lightweight semantic division model, specifically, the model which can be selected to be adopted can be a BiSeNetV2 semantic division model, so as to obtain a divided lung lobe division result, and the lung lobe division result at the moment is as follows: the method comprises the steps of obtaining the image information of the lung lobes projected on a two-dimensional plane at different observation angles, segmenting the image information of the lung lobes, obtaining accurate lung lobe contours by the lung lobe image information, traversing each slice of a lung lobe segmentation result by continuously using a lightweight semantic segmentation model, extracting contour information of each slice, and combining to form complete lung lobe contours.

In practical application, the specific correction manner of the first correction module 203 is: and acquiring a three-dimensional maximum connected domain of any one of the vein blood vessel and the artery blood vessel in the blood vessel segmentation result, wherein the three-dimensional maximum connected domain actually refers to traversing pixels of the vein blood vessel or the artery blood vessel in the blood vessel segmentation result, finding adjacent pixels and connecting the adjacent pixels to form a connected domain, wherein the largest region in all the connected domains is the three-dimensional maximum connected domain. And then correcting the other blood vessel connected with the three-dimensional maximum communication domain of any blood vessel to obtain a first correction result, thereby correcting the part of the vein blood vessel which is mistakenly identified as the arterial blood vessel and correcting the part of the artery blood vessel which is mistakenly identified as the venous blood vessel.

In practical application, the specific method for correcting the part of the venous blood vessel which is mistakenly identified as the arterial blood vessel comprises the following steps: acquiring a first three-dimensional maximum connected domain of a vein blood vessel and a set of first non-maximum connected domains of an artery blood vessel in a blood vessel segmentation result; wherein the first three-dimensional maximum connected domain is contained in the three-dimensional maximum connected domain, representing a portion of the connected domain of the vein vessel of the CT scan; the first non-maximum connected domain set represents a set of the connected domains identified as arterial blood vessels in the blood vessel segmentation result, except for the connected domain with the largest area, and includes arterial blood vessels and venous blood vessels erroneously identified as arterial blood vessels. Since the arterial blood vessel and the venous blood vessel are two unconnected space connected domains, the part connected with the first three-dimensional maximum connected domain in the collection of the first non-maximum connected domain is the venous blood vessel which is mistakenly identified as the arterial blood vessel, the part is corrected as the venous blood vessel, the collection of the corrected second non-maximum connected domain is obtained, and the corrected second non-maximum connected domain is combined with the first three-dimensional maximum connected domain and other venous blood vessels (the part which is identified as the venous blood vessel except the first three-dimensional maximum connected domain in the blood vessel segmentation result), and the three-dimensional corrected venous blood vessel can be obtained. The three-dimensional corrected venous blood vessel solves the problem of false recognition of a blood vessel segmentation model to a certain extent.

In practical application, in order to more comprehensively solve the problem of false recognition of a blood vessel segmentation model, a set of a second three-dimensional maximum connected domain of an arterial blood vessel and a third non-maximum connected domain of a venous blood vessel in a blood vessel segmentation result can be obtained; wherein the second three-dimensional maximum connected domain is contained in the three-dimensional maximum connected domain, representing a portion of the connected domain of the arterial vessel of the CT scan; and the set of third non-maximum connected domains, in which the venous blood vessel and the arterial blood vessel erroneously recognized as the venous blood vessel are included, represents a set of other parts than the connected domain having the largest area among the connected domains recognized as the venous blood vessels in the blood vessel division result. And correcting the misidentified arterial vessel part to obtain a corrected fourth non-maximum connected domain set, and combining the second three-dimensional maximum connected domain, the fourth non-maximum connected domain set and other arterial vessels to obtain the three-dimensional corrected arterial vessel. By the method, the arterial blood vessel is corrected on the basis of correcting the venous blood vessel, the three-dimensional corrected venous blood vessel and the three-dimensional corrected arterial blood vessel are respectively obtained, and the three-dimensional corrected venous blood vessel and the three-dimensional corrected arterial blood vessel are combined to obtain a first correction result.

In practical application, because the first correction result is based on correction of the three-dimensional connected domain, in the three-dimensional space, because tissues among blood vessels are staggered, the structure is very complex, and the false recognition problem at dead angles can not be corrected easily, thereby affecting the follow-up three-dimensional reconstruction of the blood vessels. In order to solve the problem, a two-dimensional maximum connected domain of any one of a vein blood vessel and an artery blood vessel in each slice of the first correction result can be obtained, wherein each slice of the first correction result actually refers to a slice in which the first correction result in the three-dimensional space is led into a blood vessel segmentation model and is output as a two-dimensional slice, and the sum of all the two-dimensional slices can form the first correction result in the three-dimensional space. The second correction result is obtained by correcting the other blood vessel connected with the two-dimensional maximum connected domain of any blood vessel, and the second correction result is a result of correcting the two-dimensional plane connected domain again on the basis of correcting the three-dimensional space connected domain, so that the problem that false recognition of dead angle positions is not easy to find and correct due to excessively complex blood vessel tissues is solved.

From the above, the device for blood vessel segmentation and correction processing provided by the application can present the blood vessel CT image of the lung lobes by acquiring the blood vessel data information of the CT scanning lung lobes so as to facilitate the follow-up three-dimensional reconstruction processing; after the blood vessel data information is obtained, the blood vessel data information is imported into a blood vessel segmentation model, blood vessels are segmented, blood vessel segmentation results are obtained, the blood vessel segmentation results comprise vein blood vessels and artery blood vessels, and because the blood vessel segmentation results are wrongly recognized and noisy due to the problem of data labeling during blood vessel segmentation, in order to solve the problem, noise filtering operation can be carried out to eliminate noise during blood vessel segmentation, blood vessel segmentation results with eliminated noise are obtained, then the part of the vein blood vessel in the blood vessel segmentation results, which is wrongly recognized as the artery blood vessel, can be corrected as the vein blood vessel, and the part of the artery blood vessel in the blood vessel segmentation results, which is wrongly recognized as the vein blood vessel, can be corrected as the artery blood vessel, so that the obtained first correction result can be eliminated to a certain extent, but the position relation among all the communicating domains can not be covered by using a three-dimensional space communicating domain method, further can be traversed through the vein blood vessel or the artery blood vessel in each layer of the first correction result, the part of the vein blood vessel, which is wrongly recognized as the artery blood vessel, the part in the vein blood vessel can be corrected as the vein blood vessel, and the second correction result can be corrected in a three-dimensional basis, and the three-dimensional correction result can be further reconstructed. Therefore, the method for correcting the segmentation of the blood vessel solves the problems that more misidentification exists at the three-level blood vessel and the four-level blood vessel in the reconstruction of the blood vessel and more noise exists around the blood vessel.

Referring to fig. 3, fig. 3 is a schematic structural diagram of an electronic device according to an embodiment of the present application, where the electronic device 3 includes: processor 301 and memory 302, the processor 301 and memory 302 being interconnected and in communication with each other by a communication bus 303 and/or other form of connection mechanism (not shown), the memory 302 storing computer readable instructions executable by the processor 301, which when executed by an electronic device, the processor 301 executes the computer readable instructions to perform the methods in any of the alternative implementations of the above embodiments to perform the functions of: acquiring blood vessel data information of CT scanning lung lobes; the blood vessel data information is imported into a blood vessel segmentation model to obtain segmented blood vessel segmentation results, wherein the blood vessel segmentation results comprise vein blood vessels and artery blood vessels; acquiring a three-dimensional maximum connected domain of any one of a venous vessel and an arterial vessel in a vessel segmentation result, and correcting the other vessel connected with the three-dimensional maximum connected domain of any one vessel to obtain a first correction result; and obtaining a two-dimensional maximum connected domain of any one of the vein blood vessel and the artery blood vessel in each slice of the first correction result, and correcting the other blood vessel connected with the two-dimensional maximum connected domain of any one of the blood vessels to obtain a second correction result.

The present application provides a computer readable storage medium having stored thereon a computer program which, when executed by a processor, performs the method of any of the alternative implementations of the above embodiments to implement the following functions: acquiring blood vessel data information of CT scanning lung lobes; the blood vessel data information is imported into a blood vessel segmentation model to obtain segmented blood vessel segmentation results, wherein the blood vessel segmentation results comprise vein blood vessels and artery blood vessels; acquiring a three-dimensional maximum connected domain of any one of a venous vessel and an arterial vessel in a vessel segmentation result, and correcting the other vessel connected with the three-dimensional maximum connected domain of any one vessel to obtain a first correction result; and obtaining a two-dimensional maximum connected domain of any one of the vein blood vessel and the artery blood vessel in each slice of the first correction result, and correcting the other blood vessel connected with the two-dimensional maximum connected domain of any one of the blood vessels to obtain a second correction result.

In the embodiments provided in the present application, it should be understood that the disclosed apparatus and method may be implemented in other manners. The above-described apparatus embodiments are merely illustrative, for example, the division of the units is merely a logical function division, and there may be other manners of division in actual implementation, and for example, multiple units or components may be combined or integrated into another system, or some features may be omitted, or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed with each other may be through some communication interface, device or unit indirect coupling or communication connection, which may be in electrical, mechanical or other form.

Further, the units described as separate units may or may not be physically separate, and units displayed as units may or may not be physical units, may be located in one place, or may be distributed over a plurality of network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

Furthermore, functional modules in various embodiments of the present application may be integrated together to form a single portion, or each module may exist alone, or two or more modules may be integrated to form a single portion.

In this document, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions.

The foregoing is merely exemplary embodiments of the present application and is not intended to limit the scope of the present application, and various modifications and variations may be suggested to one skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application should be included in the protection scope of the present application.

Claims

1. A method of vessel segmentation correction processing, the method comprising the steps of:

s1: acquiring blood vessel data information of CT scanning lung lobes;

the step S3 includes:

S34: obtaining the first correction result according to the three-dimensional corrected venous blood vessel;

s4: acquiring a two-dimensional maximum connected domain of any one of the venous blood vessel and the arterial blood vessel in each slice of the first correction result, and correcting the other blood vessel connected with the two-dimensional maximum connected domain of any one blood vessel to obtain a second correction result;

the step S4 includes:

2. A method of vessel segmentation correction procedure according to claim 1, characterized in that said step S2 comprises:

s22: acquiring a lung lobe profile of the lung lobe;

3. The method of vascular segmentation correction according to claim 2, wherein the step S22 includes:

s221: acquiring lung lobe data information of CT scanning lung lobes;

4. A method of vessel segmentation correction procedure as set forth in claim 3, wherein the step S34 includes:

5. The method of vascular segmentation correction according to claim 1, wherein the step S44 includes:

6. An apparatus for a vessel segmentation correction procedure, the apparatus comprising:

the first correction module is further configured to: acquiring a first three-dimensional maximum connected domain of the venous blood vessel and a set of first non-maximum connected domains of the arterial blood vessel in the blood vessel segmentation result;

correcting the part connected with the first three-dimensional maximum communicating domain in the first non-maximum communicating domain set into the vein to obtain a corrected second non-maximum communicating domain set;

combining the first three-dimensional maximum connected domain, the second non-maximum connected domain and other vein blood vessels to obtain three-dimensional corrected vein blood vessels;

Obtaining the first correction result according to the three-dimensional corrected venous blood vessel;

a second correction module: the two-dimensional maximum connected domain of any one of the vein blood vessel and the artery blood vessel in each slice used for obtaining the first correction result is used for correcting the other blood vessel connected with the two-dimensional maximum connected domain of any one blood vessel to obtain a second correction result;

the second correction module is further configured to: acquiring a set of first two-dimensional maximum connected domains of the venous blood vessel and fifth non-maximum connected domains of the arterial blood vessel in each slice of the first correction result;

correcting the part connected with the first two-dimensional maximum connected domain in the set of the fifth non-maximum connected domain into the venous blood vessel to obtain a corrected set of the sixth non-maximum connected domain;

combining the first two-dimensional maximum connected domain, the set of sixth non-maximum connected domain and other vein blood vessels to obtain two-dimensional corrected vein blood vessels;

and obtaining the second correction result according to the two-dimensional corrected venous blood vessel.

7. An electronic device comprising a processor and a memory storing computer readable instructions which, when executed by the processor, perform the steps of the method of any of claims 1-5.

8. A computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, performs the steps of the method according to any of claims 1-5.