Vessel segmentation method and computer device

Abstract

The application relates to a blood vessel segmentation method and a computer device, which performs segmentation processing on a medical image of a target region, obtains a plurality of sub-images corresponding to different local regions in the target region, performs blood vessel feature recognition on each sub-image, obtains a blood vessel graph corresponding to each sub-image, and performs splicing processing on each blood vessel graph to obtain a target blood vessel graph corresponding to the target region; that is, the method segments a complete medical image into a plurality of sub-images corresponding to different regions, respectively recognizes blood vessels in each sub-image, splices the blood vessel graph corresponding to each sub-image to obtain a complete target blood vessel graph corresponding to the medical image; in the application, the medical image is segmented into a plurality of different regions, and blood vessel recognition is performed on each region respectively; the blood vessel recognition algorithm used by each region is more targeted to the blood vessels in the region, the recognition effect of the blood vessels in the region is better, and the segmentation effect of the blood vessels and the integrity of the blood vessels are improved.

Description

Technical Field

[0001] This application relates to the field of image processing technology, and in particular to a method for segmenting blood vessels and a computer device. Background Technology

[0002] With the development of medical imaging equipment, the level of intelligence of medical imaging equipment is getting higher and higher. Based on the scanned images acquired by medical imaging equipment, the scanned images can be further processed. For example, the scanned images can be converted to generate target type images, or the target tissues in the scanned images can be extracted, such as segmenting the lower limb blood vessel images from the lower limb scan images. The lower limb scan images can be lower limb CT enhanced images, lower limb MR enhanced images, etc.

[0003] In traditional techniques, after acquiring enhanced CT images of the human lower limbs, medical imaging equipment uses a vascular segmentation model to segment the CT enhanced images into vascular segmentation images.

[0004] However, due to the wide distribution of blood vessels in the lower limbs of the human body, the diameter of the blood vessels decreases as they descend, and the imaging effect under CT enhancement images is also weaker. As a result, the blood vessels of the lower limbs segmented using the existing blood vessel segmentation model are incomplete, and the segmentation effect of the blood vessels is poor. Summary of the Invention

[0005] Therefore, it is necessary to provide a vascular segmentation method and computer equipment that can accurately segment complete lower limb blood vessels to address the aforementioned technical problems.

[0006] In a first aspect, this application provides a method for segmenting blood vessels. The method includes:

[0007] The medical image of the target region is segmented to obtain multiple sub-images, each sub-image corresponding to a different local region in the target region;

[0008] Perform vascular feature recognition on each sub-image to obtain the vascular map corresponding to each sub-image;

[0009] The various vascular images are stitched together to obtain the target vascular image corresponding to the target region.

[0010] In one embodiment, the medical image of the target region is segmented to obtain multiple sub-images, including:

[0011] Tissue recognition processing is performed on medical images to obtain the location of at least one target tissue in the medical image;

[0012] The medical image is segmented based on the location of the target tissue to obtain multiple sub-images.

[0013] In one embodiment, the medical image is segmented based on the location of the target tissue to obtain multiple sub-images, including:

[0014] Determine the dividing line based on the location of the target organization;

[0015] The medical image is segmented based on the dividing lines to obtain multiple sub-images.

[0016] In one embodiment, vascular feature recognition is performed on each sub-image to obtain a vascular map corresponding to each sub-image, including:

[0017] For each sub-image, the sub-image is input into the target blood vessel segmentation model to obtain the blood vessel map corresponding to the sub-image output by the target blood vessel segmentation model.

[0018] In one embodiment, the training process of the target blood vessel segmentation model includes:

[0019] Obtain a set of medical image samples of the target local region corresponding to the sub-image, and a blood vessel label image corresponding to each medical image sample in the set;

[0020] A target blood vessel segmentation model corresponding to a sub-image is obtained by training a medical image sample set and the blood vessel label images corresponding to each medical image sample.

[0021] In one embodiment, the various vascular maps are stitched together to obtain a target vascular map corresponding to the target human body region, including:

[0022] The stitching order of each vascular map is determined based on the position of the corresponding sub-image in the medical image.

[0023] Based on the stitching order, each vascular image is stitched together sequentially to obtain the target vascular image corresponding to the target region.

[0024] In one embodiment, based on the stitching order, the various vascular images are stitched together sequentially, including:

[0025] For any adjacent first and second vascular images in the splicing sequence, target vessels in the first and second vascular images are identified respectively, and the first and second vascular images are spliced ​​based on the identified target vessels. The target vessels include at least one of the aorta and the main vein.

[0026] In one embodiment, the first and second vascular maps are stitched together based on the identified target blood vessels, including:

[0027] Detect whether the positions of the target blood vessels in the first vascular image and the target blood vessels in the second vascular image match;

[0028] Detect whether the size of the target blood vessel in the first vascular image matches that of the target blood vessel in the second vascular image;

[0029] The first and second vascular images were spliced ​​together based on the test results.

[0030] In one embodiment, detecting whether the positions of target blood vessels in the first vascular map and the second vascular map match includes:

[0031] The first vascular map and the second vascular map are mapped onto the medical image, and the center line of the target blood vessel in the first vascular map is determined in the medical image, and the center line of the target blood vessel in the second vascular map is determined in the medical image.

[0032] The distance between the first endpoint of the first vessel centerline and the second endpoint of the second vessel centerline is less than or equal to a preset distance threshold. The first endpoint of the first vessel centerline is the endpoint of the first vessel centerline that is closer to the second vessel centerline, and the second endpoint of the second vessel centerline is the endpoint of the second vessel centerline that is closer to the first vessel centerline.

[0033] If the distance between the first endpoint of the first vessel centerline and the second endpoint of the second vessel centerline is less than or equal to a preset distance threshold, the position of the target vessel in the first vessel map is determined to match that of the target vessel in the second vessel map.

[0034] In one embodiment, detecting whether the size of the target blood vessel in the first vascular image matches that of the target blood vessel in the second vascular image includes:

[0035] The difference between the size parameters of the cross-section of the target blood vessel at the first endpoint and the size parameters of the cross-section of the target blood vessel at the second endpoint is less than or equal to a preset size threshold.

[0036] If the difference between the cross-sectional dimensions of the target blood vessel at the first endpoint and the cross-sectional dimensions of the target blood vessel at the second endpoint is less than or equal to a preset size threshold, the target blood vessel in the first blood vessel map is determined to be size-matched with the target blood vessel in the second blood vessel map.

[0037] In one embodiment, the cross-section is elliptical, and the size parameters include the major axis and minor axis of the ellipse. Detecting whether the difference between the size parameters of the target blood vessel cross-section at the first endpoint and the size parameters of the target blood vessel cross-section at the second endpoint is less than or equal to a preset size threshold includes:

[0038] Detect whether the difference between the major axis of the ellipse of the cross-section of the target blood vessel at the first endpoint and the major axis of the ellipse of the cross-section of the target blood vessel at the second endpoint is less than or equal to a preset major axis threshold.

[0039] The difference between the minor axis of the cross-section of the target blood vessel at the first endpoint and the minor axis of the cross-section of the target blood vessel at the second endpoint is less than or equal to a preset minor axis threshold.

[0040] If the difference between the major axis of the cross-section of the target blood vessel at the first endpoint and the major axis of the cross-section of the target blood vessel at the second endpoint is less than or equal to a preset major axis threshold, and the difference between the minor axis of the cross-section of the target blood vessel at the first endpoint and the minor axis of the cross-section of the target blood vessel at the second endpoint is less than or equal to a preset minor axis threshold, then it is determined that the difference between the size parameter of the cross-section of the target blood vessel at the first endpoint and the size parameter of the cross-section of the target blood vessel at the second endpoint is less than or equal to a preset size threshold.

[0041] In one embodiment, the first vascular image and the second vascular image are stitched together based on the detection results, including:

[0042] If the detection results show that the location and size match, the first and second vascular images are stitched together.

[0043] In one embodiment, the first vascular image and the second vascular image are stitched together based on the detection results, including:

[0044] If the detection result is that the position matches but the size does not, the target blood vessel in the second blood vessel image is corrected based on the target blood vessel in the first blood vessel image to obtain the corrected second blood vessel image.

[0045] The first vascular image and the corrected second vascular image are stitched together.

[0046] In one embodiment, based on the target vessels in the first vascular map, the target vessels in the second vascular map are corrected to obtain a corrected second vascular map, including:

[0047] Determine the fitting curve for the first blood vessel; wherein one end of the fitting curve intersects the cross-section of the target blood vessel in the first blood vessel image at the first endpoint, and the other end intersects the outer edge of the target blood vessel in the second blood vessel image;

[0048] Based on the fitting curve of the first blood vessel, the portion of the target blood vessel near the first endpoint in the second blood vessel map is corrected to obtain the corrected second blood vessel map.

[0049] In one embodiment, the first vascular image and the second vascular image are stitched together based on the detection results, including:

[0050] In the case where the detection results show that the position and size do not match, the large-section vascular image and the small-section vascular image are determined from the first vascular image and the second vascular image. The vascular image with the larger size at the splicing position of the first vascular image and the second vascular image is the large-section vascular image, and the vascular image with the smaller size at the splicing position is the small-section vascular image.

[0051] Based on the cross-sectional dimensions of the target blood vessel in the large cross-sectional vascular image where it is close to the target blood vessel in the small cross-sectional vascular image, the target blood vessel in the small cross-sectional vascular image is corrected to obtain the corrected small cross-sectional vascular image.

[0052] The large-section vascular image and the corrected small-section vascular image are stitched together.

[0053] In one embodiment, based on the size of the cross-section of the target vessel in the large-section vascular image near the target vessel in the small-section vascular image, the target vessel in the small-section vascular image is corrected to obtain a corrected small-section vascular image, including:

[0054] Determine the fitting curve for the second blood vessel; wherein one end of the fitting curve intersects the cross section of the target blood vessel in the large cross section image where it is close to the target blood vessel in the small cross section image, and the other end intersects the outer edge of the target blood vessel in the small cross section image.

[0055] Based on the second vessel fitting curve, the portion of the target vessel in the small cross-section vessel image that is close to the target vessel in the large cross-section vessel image is corrected to obtain the corrected small cross-section vessel image.

[0056] Secondly, this application also provides a blood vessel segmentation device. The device includes:

[0057] The segmentation module is used to segment the medical image of the target region to obtain multiple sub-images, each of which corresponds to a different local region in the target region;

[0058] The recognition module is used to perform vascular feature recognition on each of the sub-images to obtain the vascular map corresponding to each of the sub-images;

[0059] The stitching module is used to stitch together the various vascular images to obtain the target vascular image corresponding to the target region.

[0060] Thirdly, this application also provides a computer device. The computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps of any of the methods described in the first aspect above.

[0061] Fourthly, this application also provides a computer-readable storage medium. This computer-readable storage medium stores a computer program thereon, which, when executed by a processor, implements the steps of any of the methods described in the first aspect above.

[0062] Fifthly, this application also provides a computer program product. This computer program product includes a computer program that, when executed by a processor, implements the steps of any of the methods described in the first aspect above.

[0063] The aforementioned blood vessel segmentation method and computer device involve the computer device segmenting a medical image of a target region to obtain multiple sub-images corresponding to different local regions within the target region. Then, blood vessel feature recognition is performed on each sub-image to obtain a corresponding blood vessel map. Finally, these blood vessel maps are stitched together to obtain a target blood vessel map corresponding to the target region. In other words, the blood vessel segmentation method in this embodiment divides a complete medical image into multiple sub-images corresponding to different regions, identifies blood vessels in each sub-image, and finally stitches together the blood vessel maps corresponding to each sub-image to obtain a complete target blood vessel map corresponding to the medical image. Compared to… In existing technologies for directly identifying blood vessels in medical images, the blood vessel recognition algorithms used in these technologies are poor at recognizing blood vessels in areas with weak imaging in medical images. However, in this application, the medical image is segmented into multiple different regions, and blood vessel recognition is performed separately for each region. The blood vessel recognition algorithm used for each region is more targeted to the blood vessels in that region, resulting in better blood vessel recognition results in that region. Furthermore, the blood vessel images obtained from each region are stitched together, resulting in a more complete blood vessel image corresponding to the target blood vessel image of the medical image, which greatly improves the blood vessel segmentation effect and the integrity of the segmented blood vessels. Attached Figure Description

[0064] Figure 1 This is a flowchart illustrating a blood vessel segmentation method in one embodiment;

[0065] Figure 2 This is a flowchart illustrating a blood vessel segmentation method in another embodiment;

[0066] Figure 3 This is a schematic diagram of multiple dividing lines determined based on the target tissue in one embodiment;

[0067] Figure 4 This is a flowchart illustrating a blood vessel segmentation method in another embodiment;

[0068] Figure 5 This is a flowchart illustrating a blood vessel segmentation method in another embodiment;

[0069] Figure 6This is a schematic diagram of the blood vessel splicing structure in one embodiment;

[0070] Figure 7(a) is a schematic diagram of the vascular splicing structure in another embodiment;

[0071] Figure 7(b) is a schematic diagram of the vascular splicing structure in another embodiment;

[0072] Figure 8(a) is a schematic diagram of the blood vessel splicing structure in another embodiment;

[0073] Figure 8(b) is a schematic diagram of the vascular splicing structure in another embodiment;

[0074] Figure 8(c) is a schematic diagram of the blood vessel splicing structure in another embodiment;

[0075] Figure 9 This is a structural block diagram of a blood vessel segmentation device in one embodiment;

[0076] Figure 10 This is a structural block diagram of the blood vessel segmentation device in another embodiment;

[0077] Figure 11 This is a structural block diagram of the blood vessel segmentation device in another embodiment;

[0078] Figure 12 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0079] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0080] The blood vessel segmentation method provided in this application can be applied to computer equipment, which can be medical imaging equipment, a terminal device connected to the medical imaging equipment, or a server connected to the medical imaging equipment. The terminal device can be, but is not limited to, various personal computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. IoT devices can be smart speakers, smart TVs, smart air conditioners, smart vehicle devices, etc. Portable wearable devices can be smartwatches, smart bracelets, head-mounted devices, etc. The server can be a standalone server or a server cluster composed of multiple servers.

[0081] In one embodiment, such as Figure 1 As shown, a method for segmenting blood vessels is provided. Taking the application of this method to the aforementioned computer device as an example, the method includes the following steps:

[0082] Step 101: Segment the medical image of the target region to obtain multiple sub-images, each sub-image corresponding to a different local region in the target region.

[0083] The medical image can be a CT image, a CT-enhanced image, or an MR image, etc., and this application does not specifically limit the form of the medical image. Furthermore, the computer device can segment the medical image from top to bottom along its axis to obtain multiple sub-images corresponding to different local regions from top to bottom.

[0084] Optionally, the medical image of the target region can be input into a segmentation model for image segmentation processing to obtain multiple sub-images of different local regions corresponding to the medical image; alternatively, the medical image of the target region can be segmented into multiple sub-images corresponding to different local regions based on the image features of different local regions in the target region, wherein the image features of the local region can be features of blood vessels or tissues in the local region, etc.; alternatively, the medical image can be segmented according to preset segmentation rules corresponding to different local regions to obtain multiple sub-images, etc., such as bisection, trisection, or quartic division; it should be noted that the method of segmenting medical images in this application is not limited.

[0085] Optionally, the target region can be the lower limb region, which can be the region below the abdomen. The blood vessels in the lower limb region are divided into lower limb arteries and lower limb veins. The lower limb arteries originate from the common iliac artery and terminate at the toe artery, while the lower limb veins originate from the iliac vein and terminate at the toe vein. The lower limb region can be segmented into the upper abdomen, lower abdomen, thigh, knee, and calf. In other words, image segmentation of the medical image of the lower limb region can yield sub-images of the upper abdomen, lower abdomen, thigh, knee, and calf.

[0086] Step 102: Perform vascular feature recognition on each sub-image to obtain the vascular map corresponding to each sub-image.

[0087] The vascular image can be a mask image of blood vessels.

[0088] Optionally, each sub-image can be processed using preset image processing operations to obtain a vascular map containing only blood vessels for each sub-image. These preset image processing operations can include image filtering, denoising, background removal, enhancement, and restoration. Alternatively, pattern recognition technology can be used to extract vascular features from each sub-image to obtain a vascular map corresponding to each sub-image. Alternatively, each sub-image can be input into a target vascular segmentation model corresponding to each sub-image, and after processing by different target vascular segmentation models, vascular maps corresponding to each sub-image can be output. The target vascular segmentation model corresponds to the target local region corresponding to the sub-image; that is, the target vascular segmentation model corresponding to the sub-image is the vascular segmentation model corresponding to the target local region of the sub-image.

[0089] Optionally, when using a target vessel segmentation model to identify vessel features in different sub-images, the training process for the target vessel segmentation model corresponding to different sub-images can be as follows: First, obtain a set of medical image samples of the target local region corresponding to the sub-image and the vessel label images corresponding to each medical image sample in the set. Then, train the initial vessel segmentation model based on this set of medical image samples and the corresponding vessel label images to obtain the target vessel segmentation model for each sub-image. Since different sub-images correspond to different local regions, the image samples and labels corresponding to different local regions are also different. In other words, the set of medical image samples and the corresponding vessel label images used during the training of the target vessel segmentation model for each sub-image are also different.

[0090] Optionally, for the acquired set of medical image samples of the target local region, image preprocessing can be performed on each medical image sample in the set. This preprocessing may include window width and window level adjustment, grayscale value normalization, etc., and may also include resampling operations on each medical image sample so that each medical image sample can be resampled to the same resolution range, ensuring that the resolution of each medical image sample remains consistent, and also increasing the convergence of network model training. The process of adjusting the window width and window level may include: calculating the maximum CT value (or the second largest CT value), minimum CT value (or the second smallest CT value), and average CT value of the region of interest of the blood vessel based on the blood vessel label image corresponding to the medical image sample; half of the difference between the maximum CT value and the minimum CT value can be used as the target window width corresponding to the medical image sample; and the average CT value can be used as the target window level corresponding to the medical image sample; and the window width and window level of the medical image sample can be adjusted based on the target window width and target window level.

[0091] Step 103: The various vascular images are stitched together to obtain the target vascular image corresponding to the target region.

[0092] Optionally, after obtaining the sub-images of the medical image of the target region and the corresponding vascular images of each sub-image, the vascular images of each sub-image can be stitched together. Here, the vascular images of each sub-image can be stitched together based on the stitching order of each sub-image to obtain the stitched vascular image, which is the target vascular image corresponding to the target region. The vascular images obtained in this way may have misaligned vessels or inconsistent vessel diameters at the stitching points. In this case, further post-processing operations can be performed on the vascular images to correct the vessels in the stitched vascular images, and obtain complete vessels with smooth surfaces at each stitching position.

[0093] In the aforementioned blood vessel segmentation method, a computer device segments a medical image of the target region to obtain multiple sub-images corresponding to different local regions within the target region. Then, blood vessel feature recognition is performed on each sub-image to obtain a blood vessel map corresponding to each sub-image. Finally, the blood vessel maps are stitched together to obtain a target blood vessel map corresponding to the target region. In other words, the blood vessel segmentation method in this embodiment divides a complete medical image into multiple sub-images corresponding to different regions, identifies blood vessels in each sub-image, and finally stitches together the blood vessel maps corresponding to each sub-image to obtain a complete target blood vessel map corresponding to the medical image. Compared to existing methods... In the field of direct blood vessel identification in medical images, existing blood vessel identification algorithms are poorly designed for areas with weak imaging in medical images. This application, however, segments the medical image into multiple regions and performs blood vessel identification separately for each region. The blood vessel identification algorithm used for each region is highly targeted to the blood vessels within that region, resulting in better identification of the blood vessels within that region. Furthermore, the blood vessel images obtained from each region are stitched together, resulting in a more complete blood vessel image corresponding to the target blood vessel in the medical image, significantly improving the segmentation effect and the integrity of the segmented blood vessels.

[0094] Figure 2 This is a flowchart illustrating a blood vessel segmentation method in another embodiment. This embodiment relates to one optional implementation process of a computer device segmenting a medical image of a target region to obtain multiple sub-images. Based on the above embodiment, such as... Figure 2 As shown, step 101 above includes:

[0095] Step 201: Perform tissue recognition processing on the medical image to obtain the location of at least one target tissue in the medical image.

[0096] The tissue in the medical image can be an organ, such as the kidney, bladder, or patella in the lower limb image mentioned above.

[0097] Optionally, the medical image can be input into a tissue segmentation model to identify target tissues in the medical image and obtain the location of at least one target tissue. The tissue segmentation model can be a multi-tissue segmentation model, meaning it can identify multiple different tissues in the medical image and output segmentation results including multiple different tissues. Alternatively, the tissue segmentation model can also be multiple single-tissue segmentation models or multiple single-organ segmentation models, meaning a single-tissue segmentation model can identify one tissue in the medical image, or a single-organ segmentation model can identify one organ in the medical image and obtain the segmentation result of that tissue or organ. In practical applications, the medical image can be input into each single-tissue segmentation model and / or each single-organ segmentation model to obtain the locations of multiple target tissues.

[0098] Taking a single-tissue segmentation model as an example, the model training process is illustrated. For multiple initial single-tissue segmentation models, a training sample image set and label images of different tissues corresponding to each training sample image in the set can be obtained. Then, a single-tissue segmentation model corresponding to that tissue can be trained based on the training sample image set and the label images of the tissues corresponding to each training sample image. For example, kidney label images, bladder label images, and patella label images corresponding to each training sample image can be obtained. Further, a kidney segmentation model is obtained by training the model using each training sample image and its corresponding kidney label image; a bladder segmentation model is obtained by training the model using each training sample image and its corresponding bladder label image; and a patella segmentation model is obtained by training the model using each training sample image and its corresponding patella label image. Furthermore, this single-tissue segmentation model can be a neural network model based on V-net or a neural network model based on U-net; this application does not limit the specific form of the single-tissue segmentation model.

[0099] In practical applications, the medical image can be input into a kidney segmentation model to obtain kidney segmentation results, and the location of the kidney can be determined based on these results. Similarly, the medical image can be input into a bladder segmentation model to obtain bladder segmentation results, and the location of the bladder can be determined based on these results. Furthermore, the medical image can be input into a patellar segmentation model to obtain patellar segmentation results, and the location of the patella can be determined based on these results. Optionally, the tissue segmentation results, such as the kidney, bladder, and patellar segmentation results, can be predicted mask images of the tissues output by the tissue segmentation model. Based on these predicted mask images, the location of the tissues within the predicted mask images can be directly determined, or the predicted mask images can be mapped onto the medical image, and the location of the tissues can be determined through the medical image. This embodiment does not limit the method of determining the location of tissues based on the tissue segmentation results.

[0100] Step 202: Segment the medical image according to the location of the target tissue to obtain multiple sub-images.

[0101] Optionally, segmentation lines can be determined based on the location of the target tissue, and the medical image can be segmented according to these lines to obtain multiple sub-images. For example, based on the above example, following the axial order of the medical image from top to bottom, the lowest layer of the kidney can be determined as the segmentation line between the upper and lower abdomen, based on the location of the kidney; the lowest layer of the bladder can be determined as the segmentation line between the lower abdomen and the thigh, based on the location of the patella; the highest layer of the patella can be determined as the segmentation line between the thigh and the knee, and the lowest layer of the patella can be determined as the segmentation line between the knee and the lower leg, and so on. Figure 3 As shown; furthermore, the medical image can be segmented into an upper abdomen sub-image, a lower abdomen sub-image, a thigh sub-image, a knee sub-image, and a calf sub-image based on the segmentation lines of the upper abdomen and lower abdomen, the lower abdomen and thigh, the thigh and knee, and the knee and calf.

[0102] In this embodiment, the computer device performs tissue recognition processing on the medical image to obtain the location of at least one target tissue in the medical image, and performs axial segmentation processing on the medical image according to the location of the target tissue to obtain multiple sub-images. Since the imaging effect of blood vessels in different parts varies greatly, the medical image is axially segmented using the axial coordinates corresponding to different tissues to obtain multiple sub-images corresponding to different parts. For each sub-image, since a sub-image only includes the image of one part, the imaging effect of blood vessels in the sub-image is roughly the same. Therefore, when the algorithm for the sub-image is used to segment the blood vessels in the sub-image, the blood vessel segmentation effect is better and the segmented blood vessels are more complete than when the algorithm for the entire medical image is used to segment the blood vessels in the entire medical image, which can improve the segmentation effect of blood vessels in medical images.

[0103] Figure 4 This is a flowchart illustrating a blood vessel segmentation method in another embodiment. This embodiment involves a computer device stitching together various blood vessel images to obtain a target blood vessel image corresponding to a target human body region—one of the optional implementation processes. Based on the above embodiments, as... Figure 4 As shown, step 103 above includes:

[0104] Step 401: Determine the stitching order of each vascular map based on the position of the sub-image corresponding to each vascular map in the medical image.

[0105] The stitching order can be either from top to bottom along the axis of the medical image or from bottom to top. For example, based on the above example, the stitching order could be: vascular image corresponding to the upper abdomen sub-image → vascular image corresponding to the lower abdomen sub-image → vascular image corresponding to the thigh sub-image → vascular image corresponding to the knee sub-image → vascular image corresponding to the calf sub-image.

[0106] Step 402: Based on the stitching order, each vascular image is stitched together sequentially to obtain the target vascular image corresponding to the target region.

[0107] Specifically, after determining the stitching order of each vascular image, the computer device can sequentially stitch two adjacent vascular images according to this stitching order to obtain the target vascular image corresponding to the target region. Optionally, during the stitching process of two adjacent vascular images, for any adjacent first and second vascular images in the stitching order, the target vessels in the first and second vascular images can be identified respectively, and the first and second vascular images can be stitched based on the identified target vessels. The target vessels include at least one of the aorta and the main vein, that is, the target vessels in the first vascular image can be the aorta and / or the main vein in the first vascular image, and the target vessels in the second vascular image can be the aorta and / or the main vein in the second vascular image. In other words, the aorta and main vein in the first vascular image and the aorta and main vein in the second vascular image can be identified. Based on the aorta in the first vascular image and the aorta in the second vascular image, as well as the main vein in the first vascular image and the main vein in the second vascular image, the first vascular image and the second vascular image are spliced ​​together. By analogy, the target vascular image corresponding to the target region after the vascular images are spliced ​​together in sequence can be obtained. The target vascular image includes the complete aorta and the complete main vein in the target region.

[0108] In this embodiment, the computer device determines the stitching order of each vascular image based on the position of the sub-image corresponding to each vascular image in the medical image, and performs stitching processing on each vascular image in sequence based on the stitching order to obtain the target vascular image corresponding to the target region, which can improve the accuracy of vascular image stitching.

[0109] Figure 5 This is a flowchart illustrating a blood vessel segmentation method in another embodiment. This embodiment relates to one optional implementation process of a computer device stitching together a first blood vessel image and a second blood vessel image based on identified target blood vessels. Based on the above embodiments, such as... Figure 5 As shown, step 302 above, "stabbing the first and second vascular images based on the identified target vessels," includes:

[0110] Step 501: Detect whether the positions of the target blood vessels in the first vascular image and the target blood vessels in the second vascular image match.

[0111] Specifically, the positions of the aorta in the first vascular image and the aorta in the second vascular image are checked to see if they match, and the positions of the main vein in the first vascular image and the main vein in the second vascular image are checked to see if they match.

[0112] Optionally, the location of the target blood vessel can be the endpoint of the centerline of the target blood vessel, that is, referring to... Figure 6As shown, assuming the above top-to-bottom splicing order, it is determined whether the lower endpoint (point a) of the center line of the target vessel in the first vascular diagram (i.e., vascular diagram A) matches the upper endpoint (point b) of the center line of the target vessel in the second vascular diagram (i.e., vascular diagram B). Optionally, if the lower endpoint of the center line of the target vessel in the first vascular diagram is the same as the upper endpoint of the center line of the target vessel in the second vascular diagram, it can be determined that the target vessel in the first vascular diagram matches the target vessel in the second vascular diagram. Alternatively, if the distance between the lower endpoint of the center line of the target vessel in the first vascular diagram and the upper endpoint of the center line of the target vessel in the second vascular diagram is within a preset range, it can be determined that the target vessel in the first vascular diagram matches the target vessel in the second vascular diagram. Correspondingly, if the opposite is true, it can be determined that the target vessel in the first vascular diagram does not match the target vessel in the second vascular diagram.

[0113] Optionally, the first and second vascular maps can be mapped onto a medical image, and the center line of the target vessel in the first vascular map and the center line of the target vessel in the second vascular map can be determined in the medical image. Then, it is detected whether the distance between the first endpoint of the first vascular center line and the second endpoint of the second vascular center line is less than or equal to a preset distance threshold. If the distance between the first endpoint of the first vascular center line and the second endpoint of the second vascular center line is less than or equal to the preset distance threshold, it can be determined that the target vessel in the first vascular map and the target vessel in the second vascular map are matched. The first endpoint of the first vascular center line is the endpoint of the first vascular center line closest to the second vascular center line (based on the above example, if the first vascular map is an upper vascular map, then the first endpoint is the lower endpoint of the first vascular center line, for example, as described above). Figure 6 Point a) in the diagram, the second endpoint of the second vessel centerline is the endpoint of the second vessel centerline closest to the first vessel centerline (based on the above example, if the second vessel diagram is the one below, then the second endpoint is the upper endpoint of the second vessel centerline, for example, as mentioned above). Figure 6 Point b in the middle).

[0114] Optionally, when determining the center line of the target blood vessel in the medical image, a blood vessel thinning algorithm can be used to obtain the center line of the target blood vessel in the medical image. Other methods for obtaining the center line of the blood vessel in the prior art can also be used. This application embodiment does not limit this.

[0115] Step 502: Detect whether the size of the target blood vessel in the first vascular image matches that of the target blood vessel in the second vascular image.

[0116] Specifically, the size of the aorta in the first vascular image and the aorta in the second vascular image are checked to see if they match, and the size of the main vein in the first vascular image and the main vein in the second vascular image are checked to see if they match.

[0117] Optionally, the size of the target blood vessel can be the size of the cross-section of the blood vessel at its endpoint. That is, based on the above example, the size of the cross-section of the lowest end of the target blood vessel in the first blood vessel diagram (e.g., as described above). Figure 6 The cross-section of the blood vessel where point a is located in the diagram, and the dimensions of the uppermost cross-section of the target blood vessel in the second blood vessel diagram below (e.g., as mentioned above). Figure 6 Check if the cross-section of the blood vessel where point b is located matches.

[0118] Optionally, it can be detected whether the difference between the size parameter of the cross-section of the target blood vessel at the first endpoint and the size parameter of the cross-section of the target blood vessel at the second endpoint is less than or equal to a preset size threshold. If the difference between the size parameter of the cross-section of the target blood vessel at the first endpoint and the size parameter of the cross-section of the target blood vessel at the second endpoint is less than or equal to the preset size threshold, it can be determined that the target blood vessel in the first blood vessel image matches the size of the target blood vessel in the second blood vessel image. Optionally, the size parameter can be the radius, diameter, area, etc. of the blood vessel cross-section. When the size parameter is the area of ​​the blood vessel cross-section, it can be calculated that the difference between the area of ​​the cross-section of the target blood vessel at the first endpoint and the area of ​​the cross-section of the target blood vessel at the second endpoint is less than or equal to a preset area threshold. If so, it can be determined that the target blood vessel in the first blood vessel image matches the size of the target blood vessel in the second blood vessel image.

[0119] Optionally, the cross-section of the blood vessel may also be elliptical, and the aforementioned size parameters may also include the major axis and minor axis of the ellipse of the cross-section. The computer device can detect whether the difference between the major axis of the cross-section of the target blood vessel at the first endpoint and the major axis of the cross-section of the target blood vessel at the second endpoint is less than or equal to a preset major axis threshold; and can detect whether the difference between the minor axis of the cross-section of the target blood vessel at the first endpoint and the minor axis of the cross-section of the target blood vessel at the second endpoint is less than or equal to a preset minor axis threshold. If the difference between the major axis of the cross-section of the target blood vessel at the first endpoint and the major axis of the cross-section of the target blood vessel at the second endpoint is less than or equal to the preset major axis threshold, and the difference between the minor axis of the cross-section of the target blood vessel at the first endpoint and the minor axis of the cross-section of the target blood vessel at the second endpoint is less than or equal to the preset minor axis threshold, the device determines that the difference between the size parameters of the cross-section of the target blood vessel at the first endpoint and the size parameters of the cross-section of the target blood vessel at the second endpoint is less than or equal to the preset size threshold.

[0120] Step 503: The first and second vascular images are spliced ​​together based on the detection results.

[0121] The detection result can include: the target blood vessel in the first vascular image and the target blood vessel in the second vascular image are matched in both position and size; the target blood vessel in the first vascular image and the target blood vessel in the second vascular image are matched in both position but not in size; and the target blood vessel in the first vascular image and the target blood vessel in the second vascular image are not matched in both position and size. It should be noted that since the target blood vessel can include the aorta and the main vein, the detection result can actually include nine different combinations: ① aorta position and size match, main vein position and size match; ② aorta position and size match, main vein position and size match; ③ aorta position and size match, main vein position and size match; ④ the target blood vessel in the first vascular image and the target blood vessel in the second vascular image are not matched in both position and size. The following are the possible outcomes of the splicing process: ④ Aortic position matches but size does not match, and main vein position matches and size matches; ⑤ Aortic position matches but size does not match, and main vein position matches but size does not match; ⑥ Aortic position matches but size does not match, and main vein position does not match and size does not match; ⑦ Aortic position does not match and size does not match, and main vein position matches and size matches; ⑧ Aortic position does not match and size does not match, and main vein position matches but size does not match; ⑨ Aortic position does not match and size does not match, and main vein position does not match and size does not match. When performing splicing, the aorta and main vein need to be spliced ​​separately, with the splicing process being the same for both. Next, a detailed analysis of the splicing process for the target vessel, the aorta, will be provided. For each of the three detection results corresponding to the aorta, the splicing process will be discussed in detail.

[0122] The first scenario: The detection results show that the target blood vessels in the first vascular image and the target blood vessels in the second vascular image are matched in both position and size, as described above. Figure 6 As shown, the lower endpoint 'a' of the target blood vessel's centerline in the first vascular image and the upper endpoint 'b' of the target blood vessel's centerline in the second vascular image are located at the same position, i.e., the two endpoints 'a' and 'b' coincide. Therefore, the positions of the target blood vessels in the first and second vascular images are matched. In addition, the diameter of the target blood vessel at the lower endpoint 'a' in the first vascular image (which can be the major and minor axes of the ellipse of the blood vessel's cross-section at the lower endpoint 'a') is the same as the diameter of the target blood vessel at the upper endpoint 'b' in the second vascular image (which can be the major and minor axes of the ellipse of the blood vessel's cross-section at the upper endpoint 'b'). Therefore, the dimensions of the target blood vessels in the first and second vascular images are matched. In other words, the lower endpoint of the target blood vessel in the first vascular image and the upper endpoint of the target blood vessel in the second vascular image are completely matched. In this case, the first and second vascular images can be directly spliced ​​together. The resulting spliced ​​vascular image has complete blood vessel connections, a smooth surface, and high integrity.

[0123] The second scenario is that the target blood vessel in the first vascular image and the target blood vessel in the second vascular image are matched in position but not in size, as shown in Figure 7(a). The endpoint a of the center line of the target blood vessel in vascular image A and the endpoint b of the center line of the target blood vessel in vascular image B are in the same position, but the cross-sectional size of the target blood vessel at endpoint a in vascular image A is different from that of the target blood vessel at endpoint b in vascular image B.

[0124] In this case, the target vessels in either the first or second vascular image can be corrected to obtain first and second vascular images with matching positions and sizes. Optionally, the target vessels in the smaller-sized vessel image in the first or second vascular image can be corrected, or the target vessels in the larger-sized vessel image in the first or second vascular image can be corrected. Optionally, if the detection result shows that the target vessels in the first and second vascular images are matched in position but mismatched in size, the target vessels in the second vascular image can be corrected based on the target vessels in the first vascular image to obtain a corrected second vascular image. Then, the first vascular image and the corrected second vascular image can be stitched together. Alternatively, the target vessels in the first vascular image can be corrected based on the target vessels in the second vascular image to obtain a corrected first vascular image. Then, the corrected first and second vascular images can be stitched together.

[0125] The following is a detailed description of a specific embodiment. Based on the above example, when the medical image is a lower limb image and the stitching order is from top to bottom along the axis of the medical image, since the blood vessels in the lower limb gradually narrow from top to bottom, that is, the stitching order is based on the arrangement order of blood vessel diameters in the target area from large to small, the first blood vessel image can be the blood vessel image with a large blood vessel diameter that appears earlier in the stitching order (i.e., the blood vessel image at the top corresponds to blood vessel image A), and the second blood vessel image can be the blood vessel image with a small blood vessel diameter that appears later in the stitching order (i.e., the blood vessel image at the bottom corresponds to blood vessel image B). Then, based on the target blood vessels in the first blood vessel image, the target blood vessels in the second blood vessel image can be corrected to obtain the corrected second blood vessel image, and the first blood vessel image and the corrected second blood vessel image can be stitched together. Because blood vessels in the lower limbs become thinner towards the bottom, their imaging effect in medical images becomes worse. Therefore, the blood vessel segmentation model used on the upper blood vessel image produces relatively better segmentation results than the lower blood vessel image. Thus, the upper blood vessel image is used to correct the blood vessels in the lower blood vessel image, resulting in more accurate blood vessels. Furthermore, the upper blood vessel image and the corrected lower blood vessel image are used to stitch the blood vessels together. The resulting stitched blood vessels have a higher degree of matching with the actual blood vessels, and the blood vessel segmentation effect is better.

[0126] Optionally, the process of modifying the target blood vessel in the second vascular map to obtain a modified second vascular map based on the target blood vessel in the first vascular map may include: determining a first blood vessel fitting curve based on the cross-sectional coordinates at the first endpoint of the first blood vessel centerline of the target blood vessel in the first vascular map and the cross-sectional coordinates at different points along the axial direction of the second blood vessel centerline of the target blood vessel in the second vascular map; and modifying a portion of the target blood vessel near the first endpoint in the second vascular map based on the first blood vessel fitting curve to obtain the modified second vascular map; wherein one end of the first blood vessel fitting curve intersects the cross-section of the target blood vessel in the first vascular map at the first endpoint, and the other end intersects the outer edge of the target blood vessel in the second vascular map. Optionally, the first blood vessel fitting curve can be determined by a layer-by-layer decreasing method, wherein the layer-by-layer decreasing can be a layer-by-layer equal-interval decreasing method, resulting in a straight line; or the layer-by-layer decreasing can be a decreasing method with a certain slope, resulting in a curve, as shown in Figure 7(b).

[0127] The third scenario: The detection results show that the target blood vessel in the first vascular image and the target blood vessel in the second vascular image are mismatched in position and size, as shown in Figure 8(a). The endpoint a of the center line of the target blood vessel in vascular image A and the endpoint b of the center line of the target blood vessel in vascular image B are not in the same position, and the cross-sectional size of the target blood vessel at endpoint a in vascular image A is also different from the cross-sectional size of the target blood vessel at endpoint b in vascular image B.

[0128] In this case, the target blood vessel in the first vascular image or the target blood vessel in the second vascular image can be modified to obtain a first vascular image and a second vascular image that match in position and size. For example, the target blood vessel in the smaller vascular image in the first vascular image and the second vascular image can be modified.

[0129] Optionally, a large-section vascular image and a small-section vascular image can be determined from the first vascular image and the second vascular image. Next, based on the size of the cross-section of the target vessel in the large-section vascular image that is close to the target vessel in the small-section vascular image, the target vessel in the small-section vascular image can be corrected to obtain a corrected small-section vascular image. Then, the large-section vascular image and the corrected small-section vascular image can be spliced ​​together. The vascular image with the larger cross-section size at the splicing position between the first and second vascular images is the large-section vascular image, and the vascular image with the smaller cross-section size at the splicing position is the small-section vascular image. For two adjacent vascular images, when segmenting the sub-image using the target vascular segmentation model, the resulting vascular image may contain segmentation errors, leading to incomplete segmented vessels and deviations in the endpoint positions of the vessels from the actual endpoint positions. This results in significant discrepancies in endpoint positions and cross-sectional dimensions between the segmented vessels and the actual vessels. Therefore, when stitching together the images, it can be assumed that the segmentation effect of the larger cross-section vessel is relatively better than that of the smaller cross-section vessel image. Thus, in this embodiment, the larger cross-section vessel image is used to correct some vessels in the smaller cross-section vessel image, and the images are then stitched together based on the larger cross-section vessel image and the corrected smaller cross-section vessel image. The resulting stitched vascular image has higher integrity and accuracy, and the segmentation effect is better.

[0130] In other words, a large-section vascular map and a small-section vascular map can be determined based on the cross-sectional dimensions at the first endpoint of the first vessel centerline in the first vascular map and the cross-sectional dimensions at the second endpoint of the second vessel centerline in the second vascular map. Here, it is assumed that the first vascular map is a large-section vascular map and the second vascular map is a small-section vascular map, as shown in Figure 8(a). The second vessel fitting curve can be determined based on the cross-sectional coordinates at the first endpoint of the first vessel centerline of the target vessel in the first vascular map and the cross-sectional coordinates at different points on the second vessel centerline of the target vessel in the second vascular map. Based on the second vessel fitting curve, the portion of the target vessel in the second vascular map that is close to the target vessel in the first vascular map can be corrected to obtain the corrected second vascular map, as shown in Figure 8(b). One end of the second vessel fitting curve intersects the cross-section at the lower endpoint of the target vessel in the first vascular map (i.e., the large-section vascular map), and the other end intersects the outer edge of the target vessel in the second vascular map (i.e., the small-section vascular map). Optionally, the second vessel fitting curve can be determined using a vessel end tracking algorithm or by a layer-by-layer decreasing method. This embodiment of the application does not limit this.

[0131] Optionally, before determining the second blood vessel fitting curve, a centerline curve can be determined based on the positions of the first endpoint and the second endpoint, and then the second blood vessel fitting curve can be determined based on the centerline curve; wherein, one end of the centerline curve connects the first endpoint and the other end connects the second endpoint; optionally, an optimal path algorithm can be used to determine the centerline curve between the positions of the first endpoint and the second endpoint, and this application embodiment does not limit this.

[0132] In another optional embodiment, the target vessel in the larger of the first and second vascular maps can be modified. Specifically, the target vessel in the larger vascular map can be modified based on the cross-sectional dimensions of the target vessel in the smaller vascular map where it is close to the target vessel in the larger vascular map, resulting in a modified larger vascular map. Then, the smaller vascular map and the modified larger vascular map can be stitched together. Here, a third vascular fitting curve can be determined using a layer-by-layer incremental approach. Based on this third vascular fitting curve, the portion of the target vessel in the larger vascular map that is close to the target vessel in the smaller vascular map is modified, resulting in a modified larger vascular map, as shown in Figure 8(c). One end of the third vascular fitting curve intersects the cross-section at the upper endpoint of the target vessel in the smaller vascular map, and the other end intersects the outer edge of the target vessel in the larger vascular map.

[0133] In this embodiment, the computer device detects whether the positions of the target blood vessels in the first vascular image and the target blood vessels in the second vascular image match, and detects whether the sizes of the target blood vessels in the first vascular image and the target blood vessels in the second vascular image match, and performs splicing processing on the first vascular image and the second vascular image based on the detection results; by judging the position and size of the target blood vessels in the two adjacent vascular images during the splicing process, the integrity and smoothness of the spliced ​​blood vessels can be improved, and the segmentation effect of the blood vessels can be improved.

[0134] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0135] Based on the same inventive concept, this application also provides a blood vessel segmentation apparatus for implementing the aforementioned blood vessel segmentation method. The solution provided by this apparatus is similar to the implementation described in the above method; therefore, the specific limitations in one or more embodiments of the blood vessel segmentation apparatus provided below can be found in the limitations of the blood vessel segmentation method described above, and will not be repeated here.

[0136] In one embodiment, such as Figure 9 As shown, a blood vessel segmentation device is provided, comprising: a segmentation module 901, an identification module 902, and a splicing module 903, wherein:

[0137] The segmentation module 901 is used to segment the medical image of the target region to obtain multiple sub-images, each of which corresponds to a different local region in the target region.

[0138] The recognition module 902 is used to perform vascular feature recognition on each of the sub-images to obtain the vascular map corresponding to each sub-image.

[0139] The stitching module 903 is used to stitch together the various vascular images to obtain the target vascular image corresponding to the target region.

[0140] In one embodiment, such as Figure 10 As shown, the segmentation module 901 includes an identification unit 9011 and a segmentation unit 9012; wherein, the identification unit 9011 is used to perform tissue identification processing on the medical image to obtain the location of at least one target tissue in the medical image; the segmentation unit 9012 is used to segment the medical image according to the location of the target tissue to obtain multiple sub-images.

[0141] In one embodiment, the segmentation unit 9012 is specifically used to determine the segmentation line based on the location of the target tissue, and to segment the medical image based on the segmentation line to obtain multiple sub-images.

[0142] In one embodiment, the recognition module 902 is specifically used to input each sub-image into the target blood vessel segmentation model to obtain the blood vessel map corresponding to the sub-image output by the target blood vessel segmentation model.

[0143] In one embodiment, the training process of the target blood vessel segmentation network includes: acquiring a set of medical image samples of the target local region corresponding to the sub-image and blood vessel label images corresponding to each medical image sample in the set of medical image samples; and training the target blood vessel segmentation network corresponding to the sub-image based on the set of medical image samples and the blood vessel label images corresponding to each medical image sample.

[0144] In one embodiment, such as Figure 11 As shown, the above-mentioned stitching module 903 includes a determining unit 9031 and a stitching unit 9032; wherein, the determining unit 9031 is used to determine the stitching order of each vascular image based on the position of the sub-image corresponding to each vascular image in the medical image; the stitching unit 9032 is used to stitch each vascular image sequentially based on the stitching order to obtain the target vascular image corresponding to the target region.

[0145] In one embodiment, the splicing unit 9032 includes an identification subunit and a splicing subunit; wherein, the identification subunit is used to identify target vessels in the first vascular image and the second vascular image respectively for any adjacent first vascular image and second vascular image in the splicing sequence, and the splicing subunit is used to perform splicing processing on the first vascular image and the second vascular image based on the identified target vessels, wherein the target vessels include at least one of the aorta and the main vein.

[0146] In one embodiment, the above-mentioned splicing subunit is specifically used to detect whether the positions of the target blood vessels in the first vascular image and the target blood vessels in the second vascular image match; detect whether the sizes of the target blood vessels in the first vascular image and the target blood vessels in the second vascular image match; and perform splicing processing on the first vascular image and the second vascular image based on the detection results.

[0147] In one embodiment, the above-mentioned stitching subunit is specifically used to map the first vascular image and the second vascular image onto a medical image, and to determine the first vascular centerline of the target blood vessel in the first vascular image and the second vascular centerline of the target blood vessel in the second vascular image in the medical image; to detect whether the distance between the first endpoint of the first vascular centerline and the second endpoint of the second vascular centerline is less than or equal to a preset distance threshold; if the distance between the first endpoint of the first vascular centerline and the second endpoint of the second vascular centerline is less than or equal to the preset distance threshold, to determine that the target blood vessel in the first vascular image and the target blood vessel in the second vascular image are matched in position; wherein, the first endpoint of the first vascular centerline is the endpoint of the first vascular centerline that is closer to the second vascular centerline, and the second endpoint of the second vascular centerline is the endpoint of the second vascular centerline that is closer to the first vascular centerline.

[0148] In one embodiment, the splicing subunit is specifically used to detect whether the difference between the size parameter of the cross-section of the target blood vessel at the first endpoint and the size parameter of the cross-section of the target blood vessel at the second endpoint is less than or equal to a preset size threshold; if the difference between the size parameter of the cross-section of the target blood vessel at the first endpoint and the size parameter of the cross-section of the target blood vessel at the second endpoint is less than or equal to the preset size threshold, it is determined that the size of the target blood vessel in the first blood vessel image matches that of the target blood vessel in the second blood vessel image.

[0149] In one embodiment, the cross-section is elliptical, and the size parameters include the major axis and minor axis of the ellipse. Specifically, the splicing subunit is used to detect whether the difference between the major axis of the cross-section of the target blood vessel at the first endpoint and the major axis of the cross-section of the target blood vessel at the second endpoint is less than or equal to a preset major axis threshold; to detect whether the difference between the minor axis of the cross-section of the target blood vessel at the first endpoint and the minor axis of the cross-section of the target blood vessel at the second endpoint is less than or equal to a preset minor axis threshold; and if the difference between the major axis of the cross-section of the target blood vessel at the first endpoint and the minor axis of the cross-section of the target blood vessel at the second endpoint is less than or equal to the preset major axis threshold, and the difference between the minor axis of the cross-section of the target blood vessel at the first endpoint and the minor axis of the cross-section of the target blood vessel at the second endpoint is less than or equal to the preset minor axis threshold, then it is determined that the difference between the size parameters of the cross-section of the target blood vessel at the first endpoint and the size parameters of the cross-section of the target blood vessel at the second endpoint is less than or equal to a preset size threshold.

[0150] In one embodiment, the above-mentioned stitching subunit is specifically used to stitch together the first vascular image and the second vascular image when the detection result is that the position matches and the size matches.

[0151] In one embodiment, the above-mentioned stitching subunit is specifically used to correct the target blood vessel in the second blood vessel image based on the target blood vessel in the first blood vessel image when the detection result is that the position matches but the size does not match, so as to obtain a corrected second blood vessel image; and to stitch the first blood vessel image and the corrected second blood vessel image together.

[0152] In one embodiment, the above-mentioned splicing subunit is specifically used to determine a first blood vessel fitting curve; based on the first blood vessel fitting curve, the portion of the target blood vessel near the first endpoint in the second blood vessel map is corrected to obtain a corrected second blood vessel map; wherein, one end of the first blood vessel fitting curve intersects with the cross section of the target blood vessel in the first blood vessel map at the first endpoint, and the other end intersects with the outer edge of the target blood vessel in the second blood vessel map.

[0153] In one embodiment, the above-mentioned splicing subunit is specifically used to determine a large-section vascular image and a small-section vascular image from the first vascular image and the second vascular image when the detection result is a mismatch in position and size. Based on the size of the cross-section of the target blood vessel in the large-section vascular image that is close to the target blood vessel in the small-section vascular image, the target blood vessel in the small-section vascular image is corrected to obtain a corrected small-section vascular image. The large-section vascular image and the corrected small-section vascular image are then spliced ​​together. The vascular image with the larger cross-section size at the splicing position between the first and second vascular images is the large-section vascular image, and the vascular image with the smaller cross-section size at the splicing position is the small-section vascular image.

[0154] In one embodiment, the above-mentioned splicing subunit is specifically used to determine the second blood vessel fitting curve; based on the second blood vessel fitting curve, the portion of the target blood vessel in the small cross-section blood vessel image that is close to the target blood vessel in the large cross-section blood vessel image is corrected to obtain the corrected small cross-section blood vessel image; wherein, one end of the second blood vessel fitting curve intersects with the cross-section of the target blood vessel in the large cross-section blood vessel image that is close to the target blood vessel in the small cross-section blood vessel image, and the other end intersects with the outer edge of the target blood vessel in the small cross-section blood vessel image.

[0155] Each module in the aforementioned blood vessel segmentation device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0156] In one embodiment, a computer device is provided, the internal structure of which can be shown as follows: Figure 12 As shown, the computer device includes a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, NFC (Near Field Communication), or other technologies. When executed by the processor, the computer program implements a blood vessel segmentation method. The display screen can be an LCD screen or an e-ink display screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad mounted on the computer device casing, or an external keyboard, touchpad, or mouse.

[0157] Those skilled in the art will understand that Figure 12 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0158] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the image segmentation methods of the above embodiments.

[0159] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the image segmentation methods described in the above embodiments.

[0160] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the image segmentation methods described in the above embodiments.

[0161] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0162] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0163] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for segmenting blood vessels, characterized in that, The method includes: Tissue recognition processing is performed on the medical image to obtain the location of at least one target tissue in the medical image; Based on the location of the target tissue, the medical image is segmented from top to bottom along the axis of the medical image to obtain multiple sub-images corresponding to different local regions from top to bottom; each sub-image corresponds to a different local region in the target region, and each sub-image is an image that includes a tissue within the local region corresponding to the sub-image, and the imaging effect of blood vessels in the sub-image is consistent; For each of the sub-images, the sub-image is input into the target blood vessel segmentation model corresponding to the sub-image to obtain the blood vessel map corresponding to the sub-image output by the target blood vessel segmentation model; the target blood vessel segmentation model corresponding to the sub-image corresponds to the target local region corresponding to the sub-image, and the target blood vessel segmentation model corresponding to different local regions is different. The vascular images are stitched together to obtain the target vascular image corresponding to the target region.

2. The method according to claim 1, characterized in that, The step of segmenting the medical image from top to bottom along its axis based on the location of the target tissue to obtain multiple sub-images corresponding to different local regions from top to bottom includes: Determine the dividing line based on the location of the target tissue; The medical image is segmented according to the dividing line to obtain multiple sub-images corresponding to different local regions from top to bottom.

3. The method according to claim 1, characterized in that, The step of stitching together the various vascular images to obtain the target vascular image corresponding to the target region includes: The stitching order of each vascular image is determined based on the position of the sub-image corresponding to each vascular image in the medical image; Based on the stitching order, the vascular maps are stitched together sequentially to obtain the target vascular map corresponding to the target region.

4. The method according to claim 3, characterized in that, The process of sequentially stitching together the various vascular images based on the stitching order includes: For any adjacent first and second vascular images in the splicing sequence, target vessels are identified in the first and second vascular images respectively, and the first and second vascular images are spliced ​​based on the identified target vessels. The target vessels include at least one of the aorta and the main vein.

5. The method according to claim 4, characterized in that, The process of stitching together the first and second vascular images based on the identified target blood vessels includes: Detect whether the positions of the target blood vessels in the first vascular image and the target blood vessels in the second vascular image match; Detect whether the size of the target blood vessel in the first vascular image matches that of the target blood vessel in the second vascular image; The first and second vascular images are stitched together based on the test results.

6. The method according to claim 5, characterized in that, The step of stitching together the first vascular image and the second vascular image based on the detection results includes: If the detection result is that the position matches but the size does not match, then based on the target blood vessel in the first blood vessel image, the target blood vessel in the second blood vessel image is corrected to obtain the corrected second blood vessel image. The first vascular image and the corrected second vascular image are stitched together.

7. The method according to claim 5, characterized in that, The step of stitching together the first vascular image and the second vascular image based on the detection results includes: If the detection result is that the position does not match and the size does not match, then a large-section vascular image and a small-section vascular image are determined from the first vascular image and the second vascular image. The vascular image with the larger size at the splicing position of the first vascular image and the second vascular image is the large-section vascular image, and the vascular image with the smaller size at the splicing position is the small-section vascular image. Based on the size of the cross-section of the target blood vessel in the large cross-section vascular image where it is close to the target blood vessel in the small cross-section vascular image, the target blood vessel in the small cross-section vascular image is corrected to obtain the corrected small cross-section vascular image. The large-section vascular image and the corrected small-section vascular image are stitched together.

8. A blood vessel segmentation device, characterized in that, The device includes: A segmentation module is used to perform tissue recognition processing on a medical image to obtain the location of at least one target tissue in the medical image; based on the location of the target tissue, the medical image is segmented from top to bottom along the axis of the medical image to obtain multiple sub-images corresponding to different local regions from top to bottom; each sub-image corresponds to a different local region in the target region, and each sub-image is an image including one tissue in the local region corresponding to the sub-image, and the imaging effect of blood vessels in the sub-image is consistent; The recognition module is used to input each of the sub-images into the target blood vessel segmentation model corresponding to the sub-image to obtain the blood vessel map corresponding to the sub-image output by the target blood vessel segmentation model; the target blood vessel segmentation model corresponding to the sub-image corresponds to the target local region corresponding to the sub-image, and different target blood vessel segmentation models correspond to different local regions; The stitching module is used to stitch together the various vascular images to obtain the target vascular image corresponding to the target region.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

Images