Lesion segmentation method and device, electronic equipment and storage medium

Abstract

The application discloses a kind of focus segmentation method and device, electronic equipment and storage medium.Therein, focus segmentation method includes: obtaining medical image to be handled;The blood vessel segmentation processing is carried out to medical image to be handled, and first blood vessel segmentation result is obtained;According to the center line of blood vessel that first blood vessel segmentation result extracts;Along the vertical section of center line in medical image to be handled is intercepted multiple times, and target medical image is obtained, and along the vertical section of center line in first blood vessel segmentation result is intercepted multiple times, and target blood vessel segmentation result is obtained;Target medical image and target blood vessel segmentation result are input in focus segmentation model and carry out focus segmentation, and focus segmentation result can be obtained, similar to the accurate segmentation of the feature of blood vessel with focus can be realized, not only can the proportion of false positive sample in focus segmentation result be reduced, but also the phenomenon of focus over segmentation can be greatly relieved, help to improve the segmentation details of focus and blood vessel contact surface.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image processing, in particular to a lesion segmentation method and device, an electronic device and a storage medium. BACKGROUND

[0002] Intracranial aneurysm is a local bulge of intracranial blood vessels caused by endothelial injury, and the prevalence rate in adult population is about 5-8%. Subarachnoid hemorrhage (SAH) caused by rupture of intracranial aneurysm is one of the important causes of hemorrhagic stroke, and has a very high disability and mortality rate. The rupture of intracranial aneurysm is extremely related to its morphological parameters (such as aneurysm neck, maximum diameter, etc.). At present, the diagnosis of intracranial aneurysm in clinic is mainly through CTA (CT Angiography, CT angiography), MRA (Magnetic Resonance Angiography, Magnetic Resonance Angiography), DSA (Digital Subtraction Angiography, Digital Subtraction Angiography), DR (Digital Radiography, Digital X-ray photography) and other imaging methods, and then based on the prior knowledge of doctors, the aneurysm lesion area is manually outlined or painted to obtain aneurysm lesion mask, so as to measure the morphological parameters of aneurysm. This method is not only time-consuming and laborious, but also not accurate enough, which may lead to errors in measuring the morphological parameters of aneurysm. With the improvement of people's living standards and the development of imaging technology, a large amount of imaging data to be detected and diagnosed has been accumulated in clinic, and a clinical auxiliary diagnosis method is urgently needed to improve the accuracy and efficiency of doctors in segmenting intracranial aneurysm lesions.

[0003] At present, the target detection and image segmentation method based on deep learning method has been widely used in the detection and segmentation of medical image lesions, such as the detection of lung nodules and the segmentation of heart chambers. However, due to the small size of intracranial aneurysm, the target foreground and background are extremely unbalanced, which brings challenges to the lesion segmentation task based on deep learning method. In addition, the shape of intracranial aneurysm is variable, and similar to the features of surrounding aneurysm vessels, which greatly reduces the accuracy of lesion region segmentation. SUMMARY

[0004] The technical problem to be solved by the present application is to overcome the low segmentation accuracy of lesions similar to the features of blood vessels in the prior art, and to provide a lesion segmentation method and device, an electronic device and a storage medium.

[0005] The present application solves the above technical problems by the following technical solutions:

[0006] The first aspect of the present application provides a lesion segmentation method, comprising the following steps:

[0007] Obtaining a medical image to be processed;

[0008] performing a blood vessel segmentation process on the to-be-processed medical image to obtain a first blood vessel segmentation result;

[0009] extracting a center line of a blood vessel according to the first blood vessel segmentation result;

[0010] performing multiple times of intercepting along a vertical section of the center line in the to-be-processed medical image to obtain a target medical image composed of multiple intercepting results, and performing multiple times of intercepting along the vertical section of the center line in the first blood vessel segmentation result to obtain a target blood vessel segmentation result composed of multiple intercepting results;

[0011] inputting the target medical image and the target blood vessel segmentation result into a lesion segmentation model to perform lesion segmentation and obtain a lesion segmentation result.

[0012] Optionally, the step of extracting the center line of the blood vessel according to the first blood vessel segmentation result specifically comprises:

[0013] extracting the center line of the blood vessel according to positions of at least two points selected in the first blood vessel segmentation result.

[0014] Optionally, the step of performing multiple times of intercepting along the vertical section of the center line in the to-be-processed medical image specifically comprises: performing multiple times of intercepting along the vertical section of the center line in the to-be-processed medical image at a first step length.

[0015] Optionally, the step of performing multiple times of intercepting along the vertical section of the center line in the first blood vessel segmentation result specifically comprises: performing multiple times of intercepting along the vertical section of the center line in the first blood vessel segmentation result at a second step length.

[0016] Optionally, the step of inputting the target medical image and the target blood vessel segmentation result into the lesion segmentation model specifically comprises:

[0017] in response to a target point selected in the first blood vessel segmentation result, respectively intercepting an image block of a preset size from the target medical image and the target blood vessel segmentation result with the target point as a center;

[0018] inputting the two intercepted image blocks into the lesion segmentation model.

[0019] Optionally, after the step of inputting the target medical image and the target blood vessel segmentation result into the lesion segmentation model to perform lesion segmentation and obtain a lesion segmentation result, the method further comprises:

[0020] merging the lesion segmentation result and the first blood vessel segmentation result to obtain a final lesion segmentation result.

[0021] Optionally, the lesion segmentation model is trained according to sample data, and the sample data includes a sample medical image, a sample blood vessel segmentation result corresponding to the sample medical image, and a result of marking a lesion in the sample blood vessel segmentation result.

[0022] Optionally, the sample medical image and the sample blood vessel segmentation result are obtained according to the following steps:

[0023] performing blood vessel segmentation processing on the obtained first medical image to obtain a second blood vessel segmentation result;

[0024] extracting a center line of the blood vessel according to the second blood vessel segmentation result;

[0025] performing multiple times of intercepting along a vertical section of the center line in the first medical image to obtain the sample medical image synthesized by the multiple intercepting results, and performing multiple times of intercepting along the vertical section of the center line in the second blood vessel segmentation result to obtain the sample blood vessel segmentation result synthesized by the multiple intercepting results.

[0026] A second aspect of the present application provides a lesion segmentation device, comprising:

[0027] an image acquisition module configured to acquire a to-be-processed medical image;

[0028] a blood vessel segmentation module configured to perform blood vessel segmentation processing on the to-be-processed medical image to obtain a first blood vessel segmentation result;

[0029] a center line extraction module configured to extract a center line of the blood vessel according to the first blood vessel segmentation result;

[0030] an image intercepting module configured to perform multiple times of intercepting along a vertical section of the center line in the to-be-processed medical image to obtain a target medical image synthesized by the multiple intercepting results, and perform multiple times of intercepting along the vertical section of the center line in the first blood vessel segmentation result to obtain a target blood vessel segmentation result synthesized by the multiple intercepting results;

[0031] a lesion segmentation module configured to input the target medical image and the target blood vessel segmentation result into a lesion segmentation model to perform lesion segmentation and obtain a lesion segmentation result;

[0032] a result merging module configured to merge the lesion segmentation result and the first blood vessel segmentation result to obtain a final lesion segmentation result.

[0033] Optionally, the center line extraction module is specifically configured to extract the center line of the blood vessel according to positions of at least two points selected in the first blood vessel segmentation result.

[0034] Optionally, the image intercepting module is specifically configured to intercept multiple times in the to-be-processed medical image along the vertical section of the center line with a first step size.

[0035] Optionally, the image intercepting module is specifically configured to intercept multiple times in the first blood vessel segmentation result along the vertical section of the center line with a second step size.

[0036] Optionally, the lesion segmentation module is specifically configured to, in response to a target point selected in the first blood vessel segmentation result, intercept image blocks of a preset size from the target medical image and the target blood vessel segmentation result respectively with the target point as the center, and input the two intercepted image blocks into a lesion segmentation model.

[0037] Optionally, the lesion segmentation model is trained according to sample data, and the sample data includes a sample medical image, a sample blood vessel segmentation result corresponding to the sample medical image, and a result of labeling a lesion in the sample blood vessel segmentation result.

[0038] Optionally, the lesion segmentation device further includes a sample acquisition module, and the sample acquisition module includes:

[0039] a blood vessel segmentation unit configured to perform blood vessel segmentation processing on the acquired first medical image to obtain a second blood vessel segmentation result;

[0040] a center line extraction unit configured to extract a center line of the blood vessel according to the second blood vessel segmentation result;

[0041] an image intercepting unit configured to intercept multiple times in the first medical image along the vertical section of the center line to obtain the sample medical image composed of multiple interception results, and intercept multiple times in the second blood vessel segmentation result along the vertical section of the center line to obtain the sample blood vessel segmentation result composed of multiple interception results.

[0042] A third aspect of the present application provides an electronic device including a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the lesion segmentation method according to the first aspect when executing the computer program.

[0043] A fourth aspect of the present application provides a computer readable storage medium having a computer program stored thereon, and the computer program is executable on a processor to implement the lesion segmentation method according to the first aspect.

[0044] On the basis of common knowledge in the art, the above-mentioned optional conditions can be combined arbitrarily, that is, to obtain each preferred embodiment of the present application.

[0045] The positive progress effect of the present application is that: on the basis of the first blood vessel segmentation result of the to-be-processed medical image, the target medical image after the blood vessel center line straightening of the to-be-processed medical image and the target blood vessel segmentation result after the blood vessel center line straightening of the first blood vessel segmentation result are obtained, the target medical image and the target blood vessel segmentation result are jointly input into a lesion segmentation model for lesion segmentation, which can realize accurate segmentation of a lesion similar to a blood vessel feature, can not only reduce the proportion of false positive samples in the lesion segmentation result, but also can greatly alleviate the phenomenon of lesion over-segmentation, and is helpful to improve the segmentation details of the lesion and the blood vessel contact surface. BRIEF DESCRIPTION OF DRAWINGS

[0046] Figure 1 A flowchart of a lesion segmentation method provided for the embodiment 1 of the present application;

[0047] Figure 2 A schematic diagram of a first blood vessel segmentation result provided for the embodiment 1 of the present application;

[0048] Figure 3 A schematic diagram of a lesion position and a target point in a first blood vessel segmentation result provided for the embodiment 1 of the present application;

[0049] Figure 4 A schematic diagram of a center line of an extracted blood vessel provided for the embodiment 1 of the present application;

[0050] Figure 5 A schematic diagram of a plurality of vertical cross sections along the center line provided for the embodiment 1 of the present application;

[0051] Figure 6 A schematic diagram of a target blood vessel segmentation result provided for the embodiment 1 of the present application;

[0052] Figure 7 A schematic diagram of a lesion segmentation result provided for the embodiment 1 of the present application;

[0053] Figure 8 A method flowchart of obtaining a sample medical image and a sample blood vessel segmentation result provided for the embodiment 1 of the present application;

[0054] Figure 9 A flowchart of another lesion segmentation method provided for the embodiment 1 of the present application;

[0055] Figure 10 A structural block diagram of a lesion segmentation device provided for the embodiment 1 of the present application;

[0056] Figure 11 A structural schematic diagram of an electronic device provided for the embodiment 2 of the present application. DETAILED DESCRIPTION

[0057] The application will be further described below by way of examples without limiting the application to the examples described.

[0058] Example 1

[0059] Figure 1 A flowchart of a lesion segmentation method provided by the present embodiment is shown in FIG. 1. The lesion segmentation method can be executed by a lesion segmentation device, which can be implemented by software and / or hardware, and can be part or all of an electronic device. The electronic device in the present embodiment can be a personal computer (PC), such as a desktop computer, an all-in-one computer, a notebook computer, a tablet computer, and the like, and can also be a terminal device, such as a mobile phone, a wearable device, a personal digital assistant (PDA), and the like. The lesion segmentation method provided by the present embodiment is described below with the electronic device as the execution subject.

[0060] As shown in FIG. 1, the lesion segmentation method provided by the present embodiment can include the following steps S1-S5: Figure 1

[0061] In step S1, a medical image to be processed is obtained.

[0062] The medical image to be processed is usually a three-dimensional image, which can be a CT angiography (CTA), a magnetic resonance angiography (MRA), a three-dimensional digital subtraction angiography (3D-DSA), a three-dimensional digital radiography (3D-DR), or the like, and can be obtained by scanning a target object or downloaded from a server or a network. The target object can be a part of a patient, such as the head, the abdomen, the chest, or the like.

[0063] In step S2, a blood vessel segmentation process is performed on the medical image to be processed to obtain a first blood vessel segmentation result.

[0064] In a specific implementation, different methods can be used for blood vessel segmentation processing for different medical images to be processed. For example, for a 3D-DSA medical image to be processed, a threshold segmentation and a blood vessel enhancement algorithm based on a Hessian filter can be used to obtain a first blood vessel segmentation result. In some scenarios, the first blood vessel segmentation result can also be referred to as a first blood vessel mask. Specifically, the voxel value of the blood vessel region in the first blood vessel mask is 1, and the voxel value of the non-blood vessel region is 0.

[0065] Figure 2 A schematic diagram showing a first blood vessel segmentation result is shown in FIG. 2. As shown in FIG. 2, the white solid line part in the figure is a blood vessel. Figure 2 ​​

[0066] Step S3: extracting the centerline of the blood vessel according to the first blood vessel segmentation result.

[0067] In the specific implementation of step S3, the centerline of the blood vessel may be extracted based on a model such as a neural network.

[0068] To improve the overall lesion segmentation effect, the vessel centerline can be extracted based on points selected by the user based on the lesion location in the first vessel segmentation result. In an optional embodiment of step S3, the vessel centerline is extracted based on the positions of at least two points selected in the first vessel segmentation result. The more points selected, the more accurate the extracted vessel centerline.

[0069] In a specific implementation, at least two points can be selected based on the user's, for example, doctor's, perceived lesion location. These selected points must cover both the proximal and distal ends of the lesion. The proximal end of the lesion, also known as the proximal end, refers to the end of the vascular segment where the lesion is located that is closer to the heart; the distal end of the lesion, also known as the distal end, refers to the end of the vascular segment where the lesion is located that is farther from the heart. In other words, based on the user's perceived lesion location, the first vessel segmentation result of the branch vessel where the lesion is located is refined to obtain the centerline of the branch vessel where the lesion is located.

[0070] In such Figure 3 In the example shown, the user can select a target point in the first vessel segmentation result based on the location of the lesion, and select a point at the proximal and distal ends of the lesion. The centerline of the vessel is extracted based on the two points selected by the user. Figure 4 The dashed curve segment is shown in the figure.

[0071] Step S4: performing multiple interceptions of vertical sections along the centerline in the medical image to be processed to obtain a target medical image synthesized by the multiple interception results, and performing multiple interceptions of vertical sections along the centerline in the first blood vessel segmentation result to obtain a target blood vessel segmentation result synthesized by the multiple interception results.

[0072] In a specific implementation, a single vertical cross-section along the centerline of the medical image to be processed can be captured to obtain a single result. Each result is essentially a two-dimensional image. Multiple cross-sections of the medical image to be processed can be performed to obtain multiple two-dimensional images. These multiple two-dimensional images can be synthesized to obtain a three-dimensional target medical image. In some scenarios, the target medical image can also be referred to as the medical image to be processed after the vascular centerline has been straightened.

[0073] In a specific implementation, a vertical section along the center line in the first blood vessel segmentation result can be taken once to obtain a section result, which is essentially a two-dimensional image. Multiple two-dimensional images can be obtained by taking multiple sections in the first blood vessel segmentation result. The target blood vessel segmentation result can be obtained by synthesizing the multiple two-dimensional images. In some scenarios, the target blood vessel segmentation result can also be referred to as the first blood vessel segmentation result after the center line of the blood vessel is straightened, and can also be referred to as a target blood vessel mask.

[0074] Figure 5 A schematic diagram for showing a plurality of vertical sections along a center line. Figure 6 A schematic diagram for showing a target blood vessel segmentation result. In a specific example, a plurality of vertical sections along the center line shown in the target medical image are taken in the target medical image to obtain the target medical image, and a plurality of vertical sections along the center line shown in the first blood vessel segmentation result are taken in the first blood vessel segmentation result to obtain the target blood vessel segmentation result shown in the target blood vessel segmentation result. Figure 5 Figure 6 A plurality of vertical sections along the center line shown in the target medical image are taken in the target medical image to obtain the target medical image, and a plurality of vertical sections along the center line shown in the first blood vessel segmentation result are taken in the first blood vessel segmentation result to obtain the target blood vessel segmentation result shown in the target blood vessel segmentation result.

[0075] In a specific implementation of step S4, a plurality of vertical sections along the center line can be taken in the target medical image at a first step length. The first step length can be set according to actual conditions, for example, the first step length can be set to 0.25 mm, that is, a vertical section along the center line is taken in the target medical image at an interval of 0.25 mm.

[0076] In a specific implementation of step S4, a plurality of vertical sections along the center line can also be taken in the first blood vessel segmentation result at a second step length. The second step length can be set according to actual conditions, and can be set to be the same as the first step length or different from the first step length.

[0077] Step S5: inputting the target medical image and the target blood vessel segmentation result into a lesion segmentation model to perform lesion segmentation and obtain a lesion segmentation result.

[0078] In an optional implementation, the above step S5 includes the following steps S51 and S52:

[0079] Step S51: in response to a target point selected in the first blood vessel segmentation result, respectively taking an image block of a preset size from the target medical image and the target blood vessel segmentation result with the target point as the center.

[0080] Step S52: inputting the two taken image blocks into a lesion segmentation model.

[0081] ​In this embodiment, a target point can be selected according to the position of a lesion considered by a user, such as a doctor, a first image block of a preset size can be cropped in the target medical image with the target point as the center, and a second image block of a preset size can be cropped in the target blood vessel segmentation result with the target point as the center. Finally, the first image block and the second image block are input into a lesion segmentation model for lesion segmentation. In a specific example, the lesion segmentation result output by the lesion segmentation model is as shown in Figure 7

[0082] The lesion segmentation model is trained according to sample data, and the sample data includes a sample medical image, a sample blood vessel segmentation result corresponding to the sample medical image, and a result of labeling a lesion in the sample blood vessel segmentation result. In a specific implementation, the lesion segmentation model can be a three-dimensional convolutional neural network model, a 3D-Unet network model, a VNet network model, etc.

[0083] As shown in Figure 8 The sample medical image and the sample blood vessel segmentation result can be obtained according to the following steps S01-S03:

[0084] Step S01, performing blood vessel segmentation processing on the obtained first medical image to obtain a second blood vessel segmentation result. The first medical image is usually a three-dimensional image, which can be a CTA, MRA, 3D-DSA, 3D-DR image, etc., and can be obtained by scanning different objects or downloaded from a server or network. Different objects can include the same part of different patients, different parts of the same patient, or different parts of different patients.

[0085] Step S02, extracting a center line of the blood vessel according to the second blood vessel segmentation result. The specific implementation of step S02 is similar to step S3 described above, and the extraction of the center line of the blood vessel can be performed in combination with the points selected by the user in the second blood vessel segmentation result according to the position of the lesion. Specifically, in response to at least two points selected in the second blood vessel segmentation result, a center line of the blood vessel is extracted according to the positions of the at least two points. The at least two points selected need to cover the proximal end and the distal end of the lesion.

[0086] Step S03, performing multiple cropping in the first medical image along the vertical section of the center line to obtain the sample medical image composed of multiple cropping results, and performing multiple cropping in the second blood vessel segmentation result along the vertical section of the center line to obtain the sample blood vessel segmentation result composed of multiple cropping results.

[0087] ​In specific implementations, one cross-section along the center line in the first medical image can obtain one cross-section result, which is essentially a two-dimensional image, multiple cross-sections in the first medical image can obtain multiple two-dimensional images, and the synthesis of the multiple two-dimensional images can obtain a three-dimensional sample medical image. In some scenarios, the sample medical image can also be referred to as the first medical image after the straightening of the vessel center line.

[0088] In specific implementations, one cross-section along the center line in the second blood vessel segmentation result can obtain one cross-section result, which is essentially a two-dimensional image, multiple cross-sections in the second blood vessel segmentation result can obtain multiple two-dimensional images, and the synthesis of the multiple two-dimensional images can obtain a three-dimensional sample blood vessel segmentation result. In some scenarios, the sample blood vessel segmentation result can also be referred to as the second blood vessel segmentation result after the straightening of the vessel center line, and can also be referred to as a sample blood vessel mask.

[0089] Then, the lesion segmentation model is trained by using the sample medical image, the sample blood vessel segmentation result, and the result of labeling the lesion in the sample blood vessel segmentation result. Specifically, the sample medical image and the sample blood vessel segmentation result are input into the lesion segmentation model for lesion segmentation to obtain a predicted lesion segmentation result. A loss is calculated by using a loss function according to the predicted lesion segmentation result and the result of labeling the lesion. Whether a convergence condition is met is determined according to the loss. For example, if the loss is less than a preset value, it is indicated that the convergence condition is met. If the loss is greater than or equal to the preset value, it is indicated that the convergence condition is not met. If the convergence condition is not met, the parameters of the lesion segmentation model are adjusted according to the loss, and the lesion segmentation model is trained by using a new first medical image obtained in step S01. If the convergence condition is met, it is indicated that the lesion segmentation model has been trained, and the trained lesion segmentation model can be output.

[0090] Different first medical images can have different sizes, and the sample medical images and the sample blood vessel segmentation results obtained based on different first medical images can also have different sizes. In order to improve the training efficiency of the lesion segmentation model, the sizes of the first medical image and the second blood vessel segmentation result can be unified before the sample medical image and the sample blood vessel segmentation result are input into the lesion segmentation model. The first medical image and the second blood vessel segmentation result are respectively cut into image blocks of a preset size, with the lesion labeled in the second blood vessel segmentation result as the center.

[0091] If, during the training of the lesion segmentation model, the sizes of the sample medical images and sample blood vessel segmentation results corresponding to different first medical images are unified to a preset size, then in order to improve the accuracy of the lesion segmentation results predicted by the lesion segmentation model, in the specific implementation of step S5, it is also necessary to adjust the sizes of the target medical image and the target blood vessel segmentation results to the same preset size. The specific value of the preset size can be set according to actual conditions. For example, the preset size can be set to 64 mm * 64 mm * 64 mm.

[0092] It should be noted that data augmentation can be performed on the sample medical images and sample blood vessel segmentation results, such as random rotation, to expand the sample data for training the lesion segmentation model. Therefore, if the convergence condition is not met, after adjusting the model parameters, there is no need to return to step S01 to obtain new sample images. The lesion segmentation model can be trained directly using the data-augmented sample data, which can improve training efficiency.

[0093] In the lesion segmentation method provided in this embodiment, based on the first blood vessel segmentation result of the medical image to be processed, a target medical image after the blood vessel centerline of the medical image to be processed is obtained, and a target blood vessel segmentation result after the blood vessel centerline of the first blood vessel segmentation result is straightened is obtained. The target medical image and the target blood vessel segmentation result are jointly input into the lesion segmentation model for processing. This can achieve accurate segmentation of lesions with similar blood vessel characteristics, not only reducing the proportion of false positive samples in the lesion segmentation results, but also significantly alleviating the phenomenon of lesion over-segmentation, which helps to improve the segmentation details of the interface between the lesion and the blood vessel.

[0094] There are many lesions with similar characteristics to blood vessels, including aneurysms, venous aneurysms, plaques, etc. In a specific example, the lesion is an intracranial aneurysm. The lesion segmentation method provided in this embodiment can accurately segment the intracranial aneurysm and improve the segmentation details of the aneurysm neck.

[0095] In an optional embodiment, as Figure 9 As shown, after step S5, step S6 is further included to merge the lesion segmentation result with the first blood vessel segmentation result to obtain a final lesion segmentation result. In a specific implementation, the lesion segmentation result can be filtered and then restored to the first blood vessel segmentation result. In a specific example, the filtering process can be to filter out small noise in the lesion segmentation result.

[0096] In this embodiment, the lesion segmentation result output by the lesion segmentation model is obtained based on the straightened to-be-processed medical image and the straightened first blood vessel segmentation result. In order to facilitate the doctor to diagnose the actual position of the lesion, the lesion segmentation result is restored to the original first blood vessel segmentation result, which can better assist medical diagnosis.

[0097] This embodiment also provides a lesion segmentation device 30, as shown in the accompanying drawings, which comprises an image acquisition module 31, a blood vessel segmentation module 32, a center line extraction module 33, an image intercepting module 34, a lesion segmentation module 35, and a result merging module 36. Figure 10

[0098] The image acquisition module 31 is configured to acquire a to-be-processed medical image. The blood vessel segmentation module 32 is configured to perform blood vessel segmentation processing on the to-be-processed medical image to obtain a first blood vessel segmentation result. The center line extraction module 33 is configured to extract a center line of a blood vessel according to the first blood vessel segmentation result. The image intercepting module 34 is configured to intercept the to-be-processed medical image along a vertical section of the center line multiple times to obtain a target medical image composed of multiple intercepting results, and intercept the first blood vessel segmentation result along the vertical section of the center line multiple times to obtain a target blood vessel segmentation result composed of multiple intercepting results. The lesion segmentation module 35 is configured to input the target medical image and the target blood vessel segmentation result into a lesion segmentation model to perform lesion segmentation and obtain a lesion segmentation result. The result merging module 36 is configured to merge the lesion segmentation result and the first blood vessel segmentation result to obtain a final lesion segmentation result.

[0099] In an optional embodiment, the center line extraction module is specifically configured to extract the center line of the blood vessel according to the positions of at least two points selected in the first blood vessel segmentation result.

[0100] In an optional embodiment, the image intercepting module is specifically configured to intercept the to-be-processed medical image along the vertical section of the center line multiple times at a first step length.

[0101] In another optional embodiment, the image intercepting module is specifically configured to intercept the first blood vessel segmentation result along the vertical section of the center line multiple times at a second step length.

[0102] In an optional embodiment, the lesion segmentation module is specifically configured to, in response to a target point selected in the first blood vessel segmentation result, intercept an image block of a preset size from the target medical image and the target blood vessel segmentation result respectively with the target point as the center, and input the two intercepted image blocks into the lesion segmentation model.

[0103] ​In an optional implementation, the lesion segmentation model is trained according to sample data, and the sample data includes a sample medical image, a sample blood vessel segmentation result corresponding to the sample medical image, and a result of marking a lesion in the sample blood vessel segmentation result.

[0104] In an optional implementation, the lesion segmentation apparatus further includes a sample acquisition module, and the sample acquisition module includes a blood vessel segmentation unit, a center line extraction unit, and an image intercepting unit. The blood vessel segmentation unit is configured to perform blood vessel segmentation processing on the acquired first medical image to obtain a second blood vessel segmentation result. The center line extraction unit is configured to extract a center line of the blood vessel according to the second blood vessel segmentation result. The image intercepting unit is configured to intercept the first medical image along a vertical section of the center line multiple times to obtain the sample medical image composed of multiple intercepting results, and intercept the second blood vessel segmentation result along the vertical section of the center line multiple times to obtain the sample blood vessel segmentation result composed of multiple intercepting results.

[0105] It should be noted that the lesion segmentation apparatus in this embodiment can be a separate chip, a chip module, or an electronic device, or can be a chip or a chip module integrated in an electronic device.

[0106] As to each module / unit included in the lesion segmentation apparatus described in this embodiment, it can be a software module / unit, or a hardware module / unit, or part of it is a software module / unit and part of it is a hardware module / unit.

[0107] Embodiment 2

[0108] Figure 11 An electronic device provided in this embodiment is shown in a structural schematic diagram. The electronic device includes at least one processor and a memory in communication with the at least one processor. The memory stores a computer program that can be executed by the at least one processor. The computer program is executed by the at least one processor to enable the at least one processor to perform the lesion segmentation method of Embodiment 1. The electronic device provided in this embodiment can be a personal computer, such as a desktop computer, an all-in-one computer, a notebook computer, a tablet computer, and the like, and can also be a terminal device such as a mobile phone, a wearable device, a palm computer, and the like. Figure 11 The electronic device 3 shown is merely an example and should not impose any limitation on the functions and use range of the embodiments of the present application.

[0109] The components of the electronic device 3 can include, but are not limited to, the at least one processor 4, the at least one memory 5, and a bus 6 connecting different system components, including the memory 5 and the processor 4.

[0110] The bus 6 includes a data bus, an address bus, and a control bus.

[0111] The memory 5 can include volatile memory, such as random access memory (RAM) 51 and / or cache memory 52, and can further include non-volatile memory, such as read-only memory (ROM) 53.

[0112] The memory 5 can also include a program / utility 55 having a set (at least one) of program modules 54, including but not limited to an operating system, one or more application programs, other program modules, and program data, each of which

[0113] The processor 4, through running the computer program stored in the memory 5, executes various function applications and data processing, such as the lesion segmentation method of the above-mentioned embodiment 1.

[0114] The electronic device 3 can also communicate with one or more external devices 7, such as a keyboard, a pointing device, etc. through an input / output (I / O) interface 8. Also, the electronic device 3 can communicate with one or more networks, such as a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet, through a network adapter 9. As Figure 11 illustrated, the network adapter 9 communicates with the other modules of the electronic device 3 through the bus 6. It should be appreciated that the network adapter 9 can be communicatively coupled to the bus 6 through a network link that is connected to a hardwired, wireless, or optical fiber communication link. Figure 11 It should be appreciated that other hardware and / or software modules can be used in conjunction with the electronic device 3, as is known in the art, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID (redundant array of independent disks) systems, tape drives, and data backup storage systems, etc.

[0115] It should be noted that although several units / modules or sub-units / modules of the electronic device are mentioned in the above detailed description, such division is merely exemplary and not mandatory. In fact, according to the embodiments of the present application, the features and functions of two or more units / modules described above can be embodied in one unit / module. Conversely, the features and functions of one unit / module described above can be further divided into multiple units / modules.

[0116] Embodiment 3

[0117] The present embodiment provides a computer readable storage medium storing a computer program, which, when executed by a processor, implements the lesion segmentation method of embodiment 1.

[0118] More specifically, the readable storage medium can include, but is not limited to, a portable disc, a hard disk, a random access memory, a read-only memory, an erasable programmable read-only memory, an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0119] In a possible implementation, the present application can also be implemented in the form of a program product, which includes program codes for causing an electronic device to perform the lesion segmentation method of embodiment 1 when the program product is run on the electronic device.

[0120] The program codes for implementing the present application can be written in any combination of one or more programming languages, and can be executed entirely on the electronic device, partially on the electronic device, as a stand-alone software package, partially on the electronic device and partially on a remote device, or entirely on a remote device.

[0121] Although the above describes specific implementations of the present application, those skilled in the art should understand that this is only an example, and the protection scope of the present application is defined by the appended claims. Those skilled in the art can make various changes or modifications to these implementations without departing from the principles and essence of the present application, and such changes and modifications fall within the protection scope of the present application.

Claims

1. A lesion segmentation method, characterized in that: The method comprises the following steps: obtaining a to-be-processed medical image; performing blood vessel segmentation processing on the to-be-processed medical image to obtain a first blood vessel segmentation result; the first blood vessel segmentation result is a first blood vessel mask; extracting a center line of a blood vessel according to the first blood vessel segmentation result; performing multiple times of intercepting along a vertical section of the center line in the to-be-processed medical image to obtain a target medical image synthesized by multiple intercepting results, and performing multiple times of intercepting along the vertical section of the center line in the first blood vessel segmentation result to obtain a target blood vessel segmentation result synthesized by multiple intercepting results; inputting the target medical image and the target blood vessel segmentation result into a lesion segmentation model to perform lesion segmentation and obtain a lesion segmentation result; the step of inputting the target medical image and the target blood vessel segmentation result into the lesion segmentation model specifically comprises: in response to a target point selected in the first blood vessel segmentation result, intercepting an image block of a preset size from the target medical image and the target blood vessel segmentation result respectively with the target point as the center; inputting the two intercepted image blocks into the lesion segmentation model.

2. The lesion segmentation method of claim 1, wherein, the step of extracting the center line of the blood vessel according to the first blood vessel segmentation result specifically comprises: in response to at least two points selected in the first blood vessel segmentation result, extracting the center line of the blood vessel according to the positions of the at least two points.

3. The lesion segmentation method of claim 1, wherein, the step of performing multiple times of intercepting along the vertical section of the center line in the to-be-processed medical image specifically comprises: performing multiple times of intercepting along the vertical section of the center line in the to-be-processed medical image at a first step length; and / or, the step of performing multiple times of intercepting along the vertical section of the center line in the first blood vessel segmentation result specifically comprises: performing multiple times of intercepting along the vertical section of the center line in the first blood vessel segmentation result at a second step length.

4. The lesion segmentation method of any one of claims 1-3, wherein, after the step of inputting the target medical image and the target blood vessel segmentation result into the lesion segmentation model to perform lesion segmentation and obtain a lesion segmentation result, the method further comprises: merging the lesion segmentation result and the first blood vessel segmentation result to obtain a final lesion segmentation result.

5. The lesion segmentation method of claim 1, wherein, The lesion segmentation model is obtained by training sample data, and the sample data comprises a sample medical image, a sample blood vessel segmentation result corresponding to the sample medical image, and a result of labeling a lesion in the sample blood vessel segmentation result.

6. The lesion segmentation method of claim 5, wherein, The sample medical image and the sample blood vessel segmentation result are obtained according to the following steps: performing blood vessel segmentation processing on a first medical image to obtain a second blood vessel segmentation result; extracting a center line of a blood vessel according to the second blood vessel segmentation result; performing multiple times of intercepting along a vertical section of the center line in the first medical image to obtain the sample medical image synthesized by multiple intercepting results, and performing multiple times of intercepting along the vertical section of the center line in the second blood vessel segmentation result to obtain the sample blood vessel segmentation result synthesized by multiple intercepting results.

7. A lesion segmentation apparatus characterized by comprising: comprises: an image acquisition module configured to acquire a to-be-processed medical image; a blood vessel segmentation module configured to perform blood vessel segmentation on the to-be-processed medical image to obtain a first blood vessel segmentation result, the first blood vessel segmentation result being a first blood vessel mask; a center line extraction module configured to extract a center line of the blood vessel according to the first blood vessel segmentation result; an image intercepting module configured to intercept the to-be-processed medical image along a vertical section of the center line multiple times to obtain a target medical image composed of multiple intercepted results, and intercept the first blood vessel segmentation result along the vertical section of the center line multiple times to obtain a target blood vessel segmentation result composed of multiple intercepted results; a lesion segmentation module configured to input the target medical image and the target blood vessel segmentation result into a lesion segmentation model to perform lesion segmentation and obtain a lesion segmentation result; a result merging module configured to merge the lesion segmentation result and the first blood vessel segmentation result to obtain a final lesion segmentation result; the lesion segmentation module is specifically configured to, in response to a target point selected in the first blood vessel segmentation result, intercept image blocks of a preset size from the target medical image and the target blood vessel segmentation result respectively with the target point as a center, and input the two intercepted image blocks into a lesion segmentation model.

8. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the computer program to implement the lesion segmentation method in any one of claims 1-6.

9. A computer readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the lesion segmentation method in any one of claims 1-6.

Images