Methods, devices, electronic equipment and storage media for blood vessel segmentation

Abstract

This invention discloses a method, apparatus, electronic device, and storage medium for blood vessel segmentation. The method includes: inputting an original medical image containing a target blood vessel into an initial blood vessel segmentation model to obtain an initial blood vessel segmentation probability map; binarizing the initial blood vessel segmentation probability map to obtain a corresponding binary blood vessel segmentation map; filtering the intersection region of the original medical image and the binary blood vessel segmentation map to obtain a set of sampling points for the target blood vessel location; performing point-by-point segmentation prediction based on the set of sampling points to obtain a segmentation prediction probability value corresponding to each sampling point; updating the initial blood vessel segmentation probability map based on the segmentation prediction probability values ​​corresponding to each sampling point to obtain a target blood vessel segmentation probability map; and determining the target blood vessel segmentation prediction result based on the target blood vessel segmentation probability map. The above technical solution can effectively improve the completeness of the blood vessel segmentation results.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and in particular to a method, apparatus, electronic device, and storage medium for blood vessel segmentation. Background Technology

[0002] With the development of artificial intelligence technology, deep learning technology has been widely used in the medical field.

[0003] In medical image segmentation, the segmentation results obtained by current blood vessel segmentation algorithms are prone to blood vessel breakage (e.g., at blood vessel bifurcation and tortuosity).

[0004] In the process of realizing this invention, the inventors discovered that the prior art has at least the following technical problems: the blood vessels obtained by segmentation in the prior art are incomplete. Summary of the Invention

[0005] This invention provides a method, apparatus, electronic device, and storage medium for segmenting blood vessels to solve the problem of incomplete blood vessels.

[0006] According to one aspect of the present invention, a method for segmenting blood vessels is provided, comprising:

[0007] Obtain the original medical image containing the target blood vessel, and input the original medical image containing the target blood vessel into the initial blood vessel segmentation model to obtain the initial blood vessel segmentation probability map;

[0008] The initial blood vessel segmentation probability map is binarized to obtain the corresponding blood vessel segmentation binary map.

[0009] The intersection region of the original medical image and the binary image of blood vessel segmentation is filtered to obtain a set of sampling points for the target location of the blood vessel;

[0010] Based on the set of sampling points at the target blood vessel location, point-by-point segmentation prediction is performed to obtain the segmentation prediction probability value corresponding to each sampling point at the target blood vessel location. The initial blood vessel segmentation probability map is updated based on the segmentation prediction probability value corresponding to each sampling point at the target blood vessel location to obtain the target blood vessel segmentation probability map. The target blood vessel segmentation prediction result is determined based on the target blood vessel segmentation probability map.

[0011] According to another aspect of the present invention, a blood vessel segmentation device is provided, comprising:

[0012] The initial blood vessel segmentation probability map prediction module is used to acquire the original medical image containing the target blood vessel, and input the original medical image containing the target blood vessel into the initial blood vessel segmentation model to obtain the initial blood vessel segmentation probability map.

[0013] The binarization processing module is used to perform binarization processing on the initial blood vessel segmentation probability map to obtain the blood vessel segmentation binary map corresponding to the initial blood vessel segmentation probability map.

[0014] The filtering sampling module is used to filter the intersection region of the original medical image and the blood vessel segmentation binary image to obtain a set of sampling points for the target location of the blood vessel.

[0015] The target blood vessel segmentation prediction result determination module is used to perform point-by-point segmentation prediction based on the set of sampling points at the target blood vessel location, obtain the segmentation prediction probability value corresponding to each sampling point at the target blood vessel location, update the initial blood vessel segmentation probability map based on the segmentation prediction probability value corresponding to each sampling point at the target blood vessel location, obtain the target blood vessel segmentation probability map, and determine the target blood vessel segmentation prediction result based on the target blood vessel segmentation probability map.

[0016] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0017] At least one processor;

[0018] and a memory communicatively connected to the at least one processor;

[0019] The memory stores a computer program that can be executed by the at least one processor, which is then executed by the at least one processor to enable the at least one processor to perform the blood vessel segmentation method according to any embodiment of the present invention.

[0020] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions for causing a processor to execute and implement the blood vessel segmentation method according to any embodiment of the present invention.

[0021] The technical solution of this invention involves acquiring an original medical image containing the target blood vessel, inputting the original medical image containing the target blood vessel into an initial blood vessel segmentation model to obtain an initial blood vessel segmentation probability map; binarizing the initial blood vessel segmentation probability map to obtain a corresponding binary blood vessel segmentation map; filtering the intersection region of the original medical image and the binary blood vessel segmentation map to obtain a set of sampling points for the target blood vessel location; performing point-by-point segmentation prediction based on the set of sampling points for the target blood vessel location to obtain a segmentation prediction probability value corresponding to each sampling point for the target blood vessel location; updating the initial blood vessel segmentation probability map based on the segmentation prediction probability values ​​corresponding to each sampling point for the target blood vessel location to obtain a target blood vessel segmentation probability map; and determining the target blood vessel segmentation prediction result based on the target blood vessel segmentation probability map. This technical solution, by filtering the image, can determine the set of sampling points for the target blood vessel location, i.e., determine the tortuous and intersection regions of the blood vessel, and then sample and predict these regions, updating the blood vessel segmentation probability map to make it more accurate, thereby ensuring the integrity of the blood vessel segmentation prediction result.

[0022] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 This is a flowchart of a blood vessel segmentation method provided according to an embodiment of the present invention;

[0025] Figure 2 This is a flowchart of a blood vessel segmentation method provided according to an embodiment of the present invention;

[0026] Figure 3 This is a filtered image of a blood vessel region provided according to an embodiment of the present invention;

[0027] Figure 4 This is a flowchart of a blood vessel segmentation method provided according to an embodiment of the present invention;

[0028] Figure 5 This is a comparison image of blood vessel segmentation results provided by an embodiment of the present invention;

[0029] Figure 6This is a schematic diagram of a blood vessel segmentation device according to an embodiment of the present invention;

[0030] Figure 7 This is a schematic diagram of the structure of an electronic device that implements the blood vessel segmentation method of this invention. Detailed Implementation

[0031] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0032] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0033] Figure 1 This is a flowchart illustrating a blood vessel segmentation method provided in an embodiment of the present invention. This embodiment is applicable to the automatic segmentation of blood vessels in original medical images. The method can be executed by a blood vessel segmentation device, which can be implemented in hardware and / or software. This device can be configured in a computer device, such as a personal computer, server, or mobile terminal device. Figure 1 As shown, the method includes:

[0034] S110. Obtain the original medical image containing the target blood vessel, and input the original medical image containing the target blood vessel into the initial blood vessel segmentation model to obtain the initial blood vessel segmentation probability map.

[0035] In this embodiment, the original medical image refers to a three-dimensional image containing the target blood vessel. For example, it can be a CT (Computed Tomography) image containing the target blood vessel, or a Digital Imaging and Communications in Medicine (DICOM) image containing the target blood vessel. This embodiment does not limit the specific data type of the aforementioned original medical image. The target blood vessel can be an intracranial blood vessel, a pulmonary blood vessel, etc., and is not limited thereto.

[0036] The initial blood vessel segmentation model refers to a network prediction model capable of segmenting blood vessels. For example, the initial blood vessel segmentation model can be a blood vessel segmentation model built based on the MSNet / Unet segmentation network, or it can be a blood vessel segmentation model built based on other types of segmentation networks such as the Unet network. This embodiment does not impose any limitations on the specific model structure of the initial blood vessel segmentation model. Optionally, the initial blood vessel segmentation probability map predicted by the initial blood vessel segmentation model can be a heatmap.

[0037] Based on the above embodiments, optionally, before inputting the original medical image containing the target blood vessel into the initial model for blood vessel segmentation, the method further includes: selecting pixels in the original medical image whose CT value (Henness unit) is within a preset range; and / or scaling the original medical image to a uniform size.

[0038] For example, pixels with CT values ​​between 100 and 500 in the original medical image can be selected and retained to eliminate information unrelated to blood vessels in the image and enhance the detectability of blood vessels; and / or, the original medical image can be scaled to a size of 256*256*256 to standardize the image processing workflow so that the blood vessel segmentation model can process it.

[0039] S120. The initial blood vessel segmentation probability map is binarized to obtain the blood vessel segmentation binary map corresponding to the initial blood vessel segmentation probability map.

[0040] In some embodiments, if the pixel value of a pixel in the initial blood vessel segmentation probability map is greater than a preset judgment threshold, then the pixel value of the pixel in the initial blood vessel segmentation probability map is set to a first pixel value; if the pixel value of a pixel in the initial blood vessel segmentation probability map is less than the preset judgment threshold, then the pixel value of the pixel in the initial blood vessel segmentation probability map is set to a second pixel value. The preset judgment threshold can be obtained from image experiments and is not limited here; the first pixel value can be 0, and the second pixel value can be 1.

[0041] In some embodiments, if the pixel value of a pixel in the initial blood vessel segmentation probability map is non-zero, the pixel value of the pixel in the initial blood vessel segmentation probability map is set to 1; if the pixel value of a pixel in the initial blood vessel segmentation probability map is zero, the pixel value of the pixel in the initial blood vessel segmentation probability map remains unchanged.

[0042] S130. Filter the intersection region of the original medical image and the binary image of blood vessel segmentation to obtain a set of sampling points for the target location of blood vessels.

[0043] It should be noted that the intersection of the original medical image and the binary image of blood vessel segmentation is the region of interest, which makes the filtering process more targeted. By filtering the intersection of the images, the set of sampling points for the target location of the blood vessel can be determined, that is, the tortuous and intersection areas of the blood vessel can be determined. In other words, the above filtering process realizes the determination of the easily broken areas of the blood vessel, and the set of sampling points for the target location of the blood vessel is the point where the blood vessel is prone to breakage.

[0044] Specifically, multi-scale filtering and filter value calculation can be performed on the intersection region of the original medical image and the binary image of blood vessel segmentation to obtain a set of sampling points for the target location of blood vessels of different thicknesses. Multi-scale filtering can be Frangi filtering, etc., and is not limited here.

[0045] S140. Perform point-by-point segmentation prediction based on the set of sampling points at the target location of the blood vessel to obtain the segmentation prediction probability value corresponding to each sampling point at the target location of the blood vessel. Update the initial blood vessel segmentation probability map based on the segmentation prediction probability value corresponding to each sampling point at the target location of the blood vessel to obtain the target blood vessel segmentation probability map. Determine the target blood vessel segmentation prediction result based on the target blood vessel segmentation probability map.

[0046] Specifically, the system can perform point-by-point segmentation prediction on the vulnerable areas of blood vessels corresponding to the set of sampling points at the target locations of blood vessels. More intensive sampling and prediction can be performed on these vulnerable areas to obtain more accurate segmentation prediction probability values ​​for each sampling point at the target location of blood vessels. The initial blood vessel segmentation probability map can then be updated using the segmentation prediction probability values ​​for each sampling point at the target location of blood vessels to obtain the final target blood vessel segmentation probability map. Finally, a threshold can be applied to the target blood vessel segmentation probability map to obtain the target blood vessel segmentation prediction result, thus ensuring the integrity of the segmentation result.

[0047] Optionally, the blood vessel segmentation model includes an initial blood vessel segmentation model and a PointRend module. The training steps of the blood vessel segmentation model include: acquiring multiple medical images containing target blood vessels and blood vessel segmentation annotations for each medical image containing target blood vessels; inputting the multiple medical images containing target blood vessels into the initial blood vessel segmentation model, and outputting a blood vessel segmentation probability map for each medical image; binarizing the blood vessel segmentation probability map of each medical image to obtain a blood vessel segmentation binary map corresponding to the blood vessel segmentation probability map of each medical image; and finding the intersection of each medical image and the blood vessel segmentation binary map corresponding to the blood vessel segmentation probability map of each medical image. The region is filtered to obtain multiple vascular defect location sampling points corresponding to each medical image. The PointRend module is used to perform point-by-point segmentation prediction on the multiple vascular defect location sampling points corresponding to each medical image to obtain the segmentation prediction probability value corresponding to each vascular defect location sampling point. The vascular segmentation probability map is updated based on the segmentation prediction probability value corresponding to each vascular defect location sampling point to obtain the vascular segmentation probability map corresponding to each medical image. The vascular segmentation model is trained based on the segmentation loss between the vascular segmentation probability map corresponding to each medical image and the vascular segmentation annotation corresponding to each medical image until the vascular segmentation model converges.

[0048] The PointRend module is a neural network module that can be embedded into the blood vessel segmentation model. In other words, the PointRend module can be used as part of the blood vessel segmentation model to optimize the training and prediction of the blood vessel segmentation model, thereby improving the segmentation accuracy of the blood vessel segmentation model.

[0049] The technical solution of this invention involves acquiring an original medical image containing the target blood vessel, inputting the original medical image containing the target blood vessel into an initial blood vessel segmentation model to obtain an initial blood vessel segmentation probability map; binarizing the initial blood vessel segmentation probability map to obtain a corresponding binary blood vessel segmentation map; filtering the intersection region of the original medical image and the binary blood vessel segmentation map to obtain a set of sampling points for the target blood vessel location; performing point-by-point segmentation prediction based on the set of sampling points for the target blood vessel location to obtain a segmentation prediction probability value corresponding to each sampling point for the target blood vessel location; updating the initial blood vessel segmentation probability map based on the segmentation prediction probability values ​​corresponding to each sampling point for the target blood vessel location to obtain a target blood vessel segmentation probability map; and determining the target blood vessel segmentation prediction result based on the target blood vessel segmentation probability map. This technical solution, by filtering the image, can determine the set of sampling points for the target blood vessel location, i.e., determine the tortuous and intersection regions of the blood vessel, and then sample and predict these regions, updating the blood vessel segmentation probability map to make it more accurate, thereby ensuring the integrity of the blood vessel segmentation prediction result.

[0050] Figure 2This is a flowchart of a blood vessel segmentation method provided in an embodiment of the present invention. The method of this embodiment can be combined with various optional schemes in the blood vessel segmentation methods provided in the above embodiments. The blood vessel segmentation method provided in this embodiment has been further optimized. Optionally, the step of filtering the intersection region of the original medical image and the blood vessel segmentation binary image to obtain a set of blood vessel target location sampling points includes: determining a filtered image of the blood vessel region based on the original medical image and the blood vessel segmentation binary image; determining the Frangi filter value of each point in the filtered image of the blood vessel region, sorting the Frangi filter values ​​of each point, and determining a first blood vessel target location sampling point based on the sorting result; obtaining a second blood vessel target location sampling point outside the first blood vessel target location sampling point, and forming a set of blood vessel target location sampling points by combining the first blood vessel target location sampling point and the second blood vessel target location sampling point.

[0051] like Figure 2 As shown, the method includes:

[0052] S210. Obtain the original medical image containing the target blood vessel, and input the original medical image containing the target blood vessel into the initial blood vessel segmentation model to obtain the initial blood vessel segmentation probability map.

[0053] S220. The initial blood vessel segmentation probability map is binarized to obtain the blood vessel segmentation binary map corresponding to the initial blood vessel segmentation probability map.

[0054] S230. Determine a filtered image of the blood vessel region based on the original medical image and the blood vessel segmentation binary image.

[0055] Among them, the filtered image of the blood vessel region refers to the intersection region image of the original medical image after filtering and the binary image of blood vessel segmentation.

[0056] Specifically, multi-scale Frangi filtering can be applied to the intersection region of the original medical image and the binary image of blood vessel segmentation to obtain a filtered image of the blood vessel region.

[0057] Based on the above embodiments, optionally, determining the filtered image of the blood vessel region based on the original medical image and the blood vessel segmentation binary image includes: performing multi-scale Frangi filtering on the original medical image to obtain the filtered image corresponding to the original medical image; and multiplying the filtered image corresponding to the original medical image with the blood vessel segmentation binary image to obtain the filtered image of the blood vessel region.

[0058] Based on the above embodiments, optionally, the filtered image of the blood vessel region is determined based on the original medical image and the binary image of blood vessel segmentation, including: multiplying the original medical image and the binary image of blood vessel segmentation to obtain the blood vessel region image; and performing multi-scale Frangi filtering on the blood vessel region image to obtain the filtered image of the blood vessel region.

[0059] It should be noted that this embodiment can determine the filtered image of the blood vessel region in a variety of ways. In the application of blood vessel segmentation methods, the corresponding filtered image method of the blood vessel region can be selected according to the image processing requirements. This embodiment does not limit this method.

[0060] S240. Determine the Frangi filter value of each point in the filtered image of the blood vessel region, sort the Frangi filter values ​​of each point, and determine the sampling point of the first blood vessel target location based on the sorting result.

[0061] S250. In addition to the first blood vessel target location sampling point, obtain a second blood vessel target location sampling point, and combine the first blood vessel target location sampling point and the second blood vessel target location sampling point to form a blood vessel target location sampling point set.

[0062] S260. Based on the set of sampling points at the target location of the blood vessel, perform point-by-point segmentation prediction to obtain the segmentation prediction probability value corresponding to each sampling point at the target location of the blood vessel. Update the initial blood vessel segmentation probability map based on the segmentation prediction probability value corresponding to each sampling point at the target location of the blood vessel to obtain the target blood vessel segmentation probability map. Determine the target blood vessel segmentation prediction result based on the target blood vessel segmentation probability map.

[0063] For example, Figure 3 This is a filtered image of a blood vessel region provided in an embodiment of the present invention, by... Figure 3It is known that the Frangi filter values ​​at bifurcation and tortuosity of blood vessels are lower than those at other locations. Specifically, the top K points with the lowest Frangi filter values ​​in the filtered image of the blood vessel region can be selected and used as the first blood vessel target location sampling points. In other words, the first blood vessel target location sampling points are the points where blood vessels are prone to breakage, namely the tortuosity and bifurcation points of blood vessels. The second blood vessel target location sampling points are randomly selected points other than the first blood vessel target location sampling points. The number of second blood vessel target location sampling points can be N-TopK. Then, the first blood vessel target location sampling points and the second blood vessel target location sampling points form a blood vessel target location sampling point set, that is, the number of sampling points in the blood vessel target location sampling point set is N. This sampling strategy can effectively focus on the regions where blood vessels are prone to breakage during the blood vessel segmentation model training process, and then correct the segmentation prediction probability values ​​corresponding to each blood vessel target location sampling point in the regions where blood vessels are prone to breakage, thereby achieving more refined and richer blood vessel segmentation and improving the completeness of the segmentation results. Where N is the number of sampling points received by the PointRend module, which is a fixed value; TopK is the first K sampling points after sorting in descending or ascending order.

[0064] The technical solution of this invention determines a filtered image of the blood vessel region based on the original medical image and the binary image of blood vessel segmentation, then determines the Frangi filter value of each point in the filtered image of the blood vessel region, sorts the Frangi filter values ​​of each point, and determines a first blood vessel target location sampling point based on the sorting result; and obtains a second blood vessel target location sampling point in addition to the first blood vessel target location sampling point, and combines the first blood vessel target location sampling point and the second blood vessel target location sampling point to form a blood vessel target location sampling point set. This sampling strategy can effectively focus on the regions of blood vessels that are prone to breakage during the blood vessel segmentation model training process, and then correct the segmentation prediction probability value corresponding to each blood vessel target location sampling point corresponding to the regions of blood vessels prone to breakage, realizing strong attention sampling and probability fine-tuning, thereby achieving more refined and richer blood vessel segmentation, improving the completeness of the segmentation result. In addition, strong attention sampling and probability fine-tuning do not increase the computation time and computational complexity too much, thus improving the image processing efficiency.

[0065] Figure 4This is a flowchart of a blood vessel segmentation method provided in an embodiment of the present invention. The method in this embodiment can be combined with various optional schemes in the blood vessel segmentation methods provided in the above embodiments. The blood vessel segmentation method provided in this embodiment refines the steps of "performing point-by-point segmentation prediction based on the set of sampling points at the target blood vessel location to obtain the segmentation prediction probability value corresponding to each sampling point at the target blood vessel location, updating the initial blood vessel segmentation probability map based on the segmentation prediction probability value corresponding to each sampling point at the target blood vessel location to obtain the target blood vessel segmentation probability map, and determining the target blood vessel segmentation prediction result based on the target blood vessel segmentation probability map".

[0066] like Figure 4 As shown, the method includes:

[0067] S310. Obtain the original medical image containing the target blood vessel, and input the original medical image containing the target blood vessel into the initial blood vessel segmentation model to obtain the initial blood vessel segmentation probability map.

[0068] S320. The initial blood vessel segmentation probability map is binarized to obtain the blood vessel segmentation binary map corresponding to the initial blood vessel segmentation probability map.

[0069] S330. Filter the intersection region of the original medical image and the binary image of blood vessel segmentation to obtain a set of sampling points for the target location of the blood vessel.

[0070] S340. The set of sampling points at the target location of the blood vessel is represented point by point to obtain the feature information corresponding to each sampling point at the target location of the blood vessel.

[0071] S350. Input the feature information corresponding to the sampling points of each blood vessel target location into the multilayer perceptron to obtain the segmentation prediction probability value corresponding to the sampling points of each blood vessel target location.

[0072] S360. Based on the location information of the sampling points at the target location of the blood vessel, fill the segmentation prediction probability value corresponding to each sampling point at the target location of the blood vessel into the initial blood vessel segmentation probability map to obtain the target blood vessel segmentation probability map.

[0073] S370. The target blood vessel segmentation probability map is binarized to obtain the target blood vessel segmentation prediction result.

[0074] For example, fine-grained features and coarse predicted features are represented for each sampling point in the set of blood vessel target location sampling points to obtain the feature information corresponding to each blood vessel target location sampling point. Then, a multilayer perceptron (MLP) is used to predict the feature information corresponding to each blood vessel target location sampling point to obtain the segmentation prediction probability value corresponding to each blood vessel target location sampling point. In this embodiment, the location information of the blood vessel target location sampling points can be the coordinate information of the blood vessel target location sampling points. Specifically, based on the location information of the blood vessel target location sampling points, the segmentation prediction probability value corresponding to each blood vessel target location sampling point can be put back into the initial blood vessel segmentation probability map to realize the probability correction of the blood vessel target location sampling points. Then, the corrected target blood vessel segmentation probability map is binarized to obtain a more refined and richer blood vessel segmentation prediction result.

[0075] Figure 5 This is a comparison image of blood vessel segmentation results provided in an embodiment of the present invention. Figure 5 The left side shows the blood vessel segmentation prediction results obtained from the initial blood vessel segmentation model. Figure 5 The right side shows the target vessel segmentation prediction results obtained by the vessel segmentation method of this embodiment. As can be seen from the figure, the vessel segmentation method of this embodiment solves the problem of easy vessel segmentation breakage or incomplete segmentation, improves the integrity of vessel segmentation, and thus reduces the false negative rate of lesions.

[0076] The technical solution of this invention represents the set of sampling points at the target blood vessel location point by point to obtain the feature information corresponding to each sampling point, thus realizing the construction of the point-by-point feature representation of each sampling point. Then, the feature information corresponding to each sampling point at the target blood vessel location is input into a multilayer perceptron to achieve point-by-point segmentation prediction, resulting in a more accurate segmentation prediction probability value. Then, based on the location information of the sampling points at the target blood vessel location, the segmentation prediction probability value corresponding to each sampling point at the target blood vessel location is filled into the initial blood vessel segmentation probability map to obtain the corrected target blood vessel segmentation probability map. The target blood vessel segmentation probability map is binarized to obtain the target blood vessel segmentation prediction result, ensuring the integrity of the segmentation result.

[0077] Figure 6 This is a schematic diagram of a blood vessel segmentation device provided in an embodiment of the present invention. Figure 6 As shown, the device includes:

[0078] The initial blood vessel segmentation probability map prediction module 410 is used to acquire the original medical image containing the target blood vessel, and input the original medical image containing the target blood vessel into the initial blood vessel segmentation model to obtain the initial blood vessel segmentation probability map.

[0079] The binarization processing module 420 is used to perform binarization processing on the initial blood vessel segmentation probability map to obtain the blood vessel segmentation binary map corresponding to the initial blood vessel segmentation probability map.

[0080] The filtering sampling module 430 is used to filter the intersection region of the original medical image and the blood vessel segmentation binary image to obtain a set of sampling points for the target location of the blood vessel.

[0081] The target blood vessel segmentation prediction result determination module 440 is used to perform point-by-point segmentation prediction based on the set of sampling points at the target blood vessel location, obtain the segmentation prediction probability value corresponding to each sampling point at the target blood vessel location, update the initial blood vessel segmentation probability map based on the segmentation prediction probability value corresponding to each sampling point at the target blood vessel location, obtain the target blood vessel segmentation probability map, and determine the target blood vessel segmentation prediction result based on the target blood vessel segmentation probability map.

[0082] The technical solution of this invention involves acquiring an original medical image containing the target blood vessel, inputting the original medical image containing the target blood vessel into an initial blood vessel segmentation model to obtain an initial blood vessel segmentation probability map; binarizing the initial blood vessel segmentation probability map to obtain a corresponding binary blood vessel segmentation map; filtering the intersection region of the original medical image and the binary blood vessel segmentation map to obtain a set of sampling points for the target blood vessel location; performing point-by-point segmentation prediction based on the set of sampling points for the target blood vessel location to obtain a segmentation prediction probability value corresponding to each sampling point for the target blood vessel location; updating the initial blood vessel segmentation probability map based on the segmentation prediction probability values ​​corresponding to each sampling point for the target blood vessel location to obtain a target blood vessel segmentation probability map; and determining the target blood vessel segmentation prediction result based on the target blood vessel segmentation probability map. This technical solution, by filtering the image, can determine the set of sampling points for the target blood vessel location, i.e., determine the tortuous and intersection regions of the blood vessel, and then sample and predict these regions, updating the blood vessel segmentation probability map to make it more accurate, thereby ensuring the integrity of the segmentation result.

[0083] In some optional implementations, the binarization processing module 420 is specifically used for:

[0084] If the pixel value of a pixel in the initial blood vessel segmentation probability map is greater than a preset judgment threshold, then the pixel value of the pixel in the initial blood vessel segmentation probability map is set to the first pixel value;

[0085] If the pixel value of a pixel in the initial blood vessel segmentation probability map is less than a preset judgment threshold, then the pixel value of the pixel in the initial blood vessel segmentation probability map is set to the second pixel value.

[0086] In some alternative implementations, the filtering sampling module 430 includes:

[0087] An image filtering unit is used to determine a filtered image of the blood vessel region based on the original medical image and the blood vessel segmentation binary image.

[0088] The first blood vessel target location sampling point determination unit is used to determine the Fragi filter value of each point in the filtered image of the blood vessel region, sort the Fragi filter values ​​of each point, and determine the first blood vessel target location sampling point based on the sorting result.

[0089] The blood vessel target location sampling point set determination unit is used to obtain a second blood vessel target location sampling point in addition to the first blood vessel target location sampling point, and to form a blood vessel target location sampling point set by combining the first blood vessel target location sampling point and the second blood vessel target location sampling point.

[0090] In some alternative implementations, the image filtering unit is further configured to:

[0091] The original medical image is subjected to multi-scale Fragi filtering to obtain the filtered image corresponding to the original medical image.

[0092] The filtered image corresponding to the original medical image is multiplied by the binary image of the blood vessel segmentation to obtain the filtered image of the blood vessel region.

[0093] In some alternative implementations, the image filtering unit is further configured to:

[0094] The original medical image is multiplied by the binary image of blood vessel segmentation to obtain the blood vessel region image.

[0095] Multi-scale Fragi filtering is applied to the image of the blood vessel region to obtain a filtered image of the blood vessel region.

[0096] In some optional implementations, the target vessel segmentation prediction result determination module 440 is further configured to:

[0097] The set of sampling points at the target location of the blood vessel is represented point by point to obtain the feature information corresponding to each sampling point at the target location of the blood vessel.

[0098] The feature information corresponding to the sampling points of each blood vessel target location is input into the multilayer perceptron to obtain the segmentation prediction probability value corresponding to the sampling points of each blood vessel target location.

[0099] In some optional implementations, the target vessel segmentation prediction result determination module 440 is further configured to:

[0100] Based on the location information of the sampling points at the target blood vessel locations, the segmentation prediction probability values ​​corresponding to each sampling point at the target blood vessel location are filled into the initial blood vessel segmentation probability map to obtain the target blood vessel segmentation probability map.

[0101] In some optional implementations, the target vessel segmentation prediction result determination module 440 is further configured to:

[0102] The target blood vessel segmentation probability map is binarized to obtain the target blood vessel segmentation prediction result.

[0103] In some optional implementations, the vessel segmentation model includes a vessel segmentation initial model and a PointRend module, and the training steps of the vessel segmentation model include:

[0104] Acquire multiple medical images containing target blood vessels and blood vessel segmentation annotations for each of the medical images containing target blood vessels;

[0105] The multiple medical images containing the target blood vessels are input into the initial blood vessel segmentation model, and a blood vessel segmentation probability map is output for each of the medical images.

[0106] The blood vessel segmentation probability map of each medical image is binarized to obtain the blood vessel segmentation binary map corresponding to the blood vessel segmentation probability map of each medical image.

[0107] The intersection region of each medical image and the corresponding binary image of blood vessel segmentation probability map of each medical image is filtered to obtain multiple sampling points of blood vessel defect location for each medical image.

[0108] The PointRend module performs point-by-point segmentation prediction on multiple vascular defect location sampling points corresponding to each medical image to obtain the segmentation prediction probability value corresponding to each vascular defect location sampling point. Based on the segmentation prediction probability value corresponding to each vascular defect location sampling point, the vascular segmentation probability map is updated to obtain the vascular segmentation probability map corresponding to each medical image.

[0109] The blood vessel segmentation model is trained based on the segmentation loss between the blood vessel segmentation probability map corresponding to each medical image and the blood vessel segmentation annotation corresponding to each medical image until the blood vessel segmentation model converges.

[0110] The blood vessel segmentation device provided in the embodiments of the present invention can execute the blood vessel segmentation method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the method.

[0111] Figure 7A schematic diagram of an electronic device 10 that can be used to implement embodiments of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0112] like Figure 7 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded into the RAM 13 from storage unit 18. The RAM 13 can also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An I / O interface 15 is also connected to the bus 14.

[0113] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0114] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as a blood vessel segmentation method, which includes:

[0115] Obtain the original medical image containing the target blood vessel, and input the original medical image containing the target blood vessel into the initial blood vessel segmentation model to obtain the initial blood vessel segmentation probability map;

[0116] The initial blood vessel segmentation probability map is binarized to obtain the corresponding blood vessel segmentation binary map.

[0117] The intersection region of the original medical image and the binary image of blood vessel segmentation is filtered to obtain a set of sampling points for the target location of the blood vessel;

[0118] Based on the set of sampling points at the target blood vessel location, point-by-point segmentation prediction is performed to obtain the segmentation prediction probability value corresponding to each sampling point at the target blood vessel location. The initial blood vessel segmentation probability map is updated based on the segmentation prediction probability value corresponding to each sampling point at the target blood vessel location to obtain the target blood vessel segmentation probability map. The target blood vessel segmentation prediction result is determined based on the target blood vessel segmentation probability map.

[0119] In some embodiments, the vessel segmentation method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or mounted on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the vessel segmentation method described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform the vessel segmentation method by any other suitable means (e.g., by means of firmware).

[0120] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0121] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0122] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0123] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0124] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or computing systems that include middleware components (e.g., application servers), or computing systems that include frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0125] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0126] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0127] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for segmenting blood vessels, characterized in that, include: Obtain the original medical image containing the target blood vessel, and input the original medical image containing the target blood vessel into the initial blood vessel segmentation model to obtain the initial blood vessel segmentation probability map; The initial blood vessel segmentation probability map is binarized to obtain the corresponding blood vessel segmentation binary map. A filtered image of the blood vessel region is determined based on the original medical image and the binary image of blood vessel segmentation; wherein, the filtered image of the blood vessel region refers to the intersection region image of the original medical image and the binary image of blood vessel segmentation after filtering. Determine the Frangi filter value of each point in the filtered image of the blood vessel region, sort the Frangi filter values ​​of each point, and determine the sampling point of the first blood vessel target location based on the sorting result; In addition to the first blood vessel target location sampling point, a second blood vessel target location sampling point is obtained, and the first blood vessel target location sampling point and the second blood vessel target location sampling point are combined to form a blood vessel target location sampling point set. Based on the set of sampling points at the target blood vessel location, point-by-point segmentation prediction is performed to obtain the segmentation prediction probability value corresponding to each sampling point at the target blood vessel location. The initial blood vessel segmentation probability map is updated based on the segmentation prediction probability value corresponding to each sampling point at the target blood vessel location to obtain the target blood vessel segmentation probability map. The target blood vessel segmentation prediction result is determined based on the target blood vessel segmentation probability map.

2. The method according to claim 1, characterized in that, The step of binarizing the initial blood vessel segmentation probability map to obtain the corresponding binary blood vessel segmentation map includes: If the pixel value of a pixel in the initial blood vessel segmentation probability map is greater than a preset judgment threshold, then the pixel value of the pixel in the initial blood vessel segmentation probability map is set to the first pixel value; If the pixel value of a pixel in the initial blood vessel segmentation probability map is less than a preset judgment threshold, then the pixel value of the pixel in the initial blood vessel segmentation probability map is set to the second pixel value.

3. The method according to claim 1, characterized in that, The filtered image used to determine the vascular region based on the original medical image and the vascular segmentation binary image includes: The original medical image is subjected to multi-scale Fragi filtering to obtain the filtered image corresponding to the original medical image. The filtered image corresponding to the original medical image is multiplied by the binary image of the blood vessel segmentation to obtain the filtered image of the blood vessel region.

4. The method according to claim 1, characterized in that, The filtered image used to determine the vascular region based on the original medical image and the vascular segmentation binary image includes: The original medical image is multiplied by the binary image of blood vessel segmentation to obtain the blood vessel region image. Multi-scale Fragi filtering is applied to the image of the blood vessel region to obtain a filtered image of the blood vessel region.

5. The method according to claim 1, characterized in that, The step of performing point-by-point segmentation prediction based on the set of sampling points at the target vessel location to obtain the segmentation prediction probability value corresponding to each sampling point at the target vessel location includes: The set of sampling points at the target location of the blood vessel is represented point by point to obtain the feature information corresponding to each sampling point at the target location of the blood vessel. The feature information corresponding to the sampling points of each blood vessel target location is input into the multilayer perceptron to obtain the segmentation prediction probability value corresponding to the sampling points of each blood vessel target location.

6. The method according to claim 1, characterized in that, The process of updating the initial blood vessel segmentation probability map based on the segmentation prediction probability value corresponding to the sampling points at each blood vessel target location to obtain the target blood vessel segmentation probability map includes: Based on the location information of the sampling points at the target blood vessel locations, the segmentation prediction probability values ​​corresponding to each sampling point at the target blood vessel location are filled into the initial blood vessel segmentation probability map to obtain the target blood vessel segmentation probability map.

7. The method according to claim 1, characterized in that, The step of determining the target vessel segmentation prediction result based on the target vessel segmentation probability map includes: The target blood vessel segmentation probability map is binarized to obtain the target blood vessel segmentation prediction result.

8. The method according to claim 1, characterized in that, The blood vessel segmentation model includes an initial blood vessel segmentation model and a PointRend module. The training steps of the blood vessel segmentation model include: Acquire multiple medical images containing target blood vessels and blood vessel segmentation annotations for each of the medical images containing target blood vessels; The multiple medical images containing the target blood vessels are input into the initial blood vessel segmentation model, and a blood vessel segmentation probability map is output for each of the medical images. The blood vessel segmentation probability map of each medical image is binarized to obtain the blood vessel segmentation binary map corresponding to the blood vessel segmentation probability map of each medical image. The intersection region of each medical image and the corresponding binary image of blood vessel segmentation probability map of each medical image is filtered to obtain multiple sampling points of blood vessel defect location for each medical image. The PointRend module performs point-by-point segmentation prediction on multiple vascular defect location sampling points corresponding to each medical image to obtain the segmentation prediction probability value corresponding to each vascular defect location sampling point. Based on the segmentation prediction probability value corresponding to each vascular defect location sampling point, the vascular segmentation probability map is updated to obtain the vascular segmentation probability map corresponding to each medical image. The blood vessel segmentation model is trained based on the segmentation loss between the blood vessel segmentation probability map corresponding to each medical image and the blood vessel segmentation annotation corresponding to each medical image until the blood vessel segmentation model converges.

9. A blood vessel segmentation device, characterized in that, include: The initial blood vessel segmentation probability map prediction module is used to acquire the original medical image containing the target blood vessel, and input the original medical image containing the target blood vessel into the initial blood vessel segmentation model to obtain the initial blood vessel segmentation probability map. The binarization processing module is used to perform binarization processing on the initial blood vessel segmentation probability map to obtain the blood vessel segmentation binary map corresponding to the initial blood vessel segmentation probability map. A filtering sampling module is used to determine a filtered image of a blood vessel region based on the original medical image and the blood vessel segmentation binary image; wherein, the filtered image of the blood vessel region refers to the intersection region image of the original medical image and the blood vessel segmentation binary image after filtering; determining the Frangi filter value of each point in the filtered image of the blood vessel region, sorting the Frangi filter values ​​of each point, and determining a first blood vessel target location sampling point based on the sorting result; obtaining a second blood vessel target location sampling point outside the first blood vessel target location sampling point, and forming a blood vessel target location sampling point set by combining the first blood vessel target location sampling point and the second blood vessel target location sampling point; The target blood vessel segmentation prediction result determination module is used to perform point-by-point segmentation prediction based on the set of sampling points at the target blood vessel location, obtain the segmentation prediction probability value corresponding to each sampling point at the target blood vessel location, update the initial blood vessel segmentation probability map based on the segmentation prediction probability value corresponding to each sampling point at the target blood vessel location, obtain the target blood vessel segmentation probability map, and determine the target blood vessel segmentation prediction result based on the target blood vessel segmentation probability map.

10. An electronic device, characterized in that, The electronic device includes: At least one processor; and a memory communicatively connected to the at least one processor; The memory stores a computer program that can be executed by the at least one processor, which enables the at least one processor to perform the blood vessel segmentation method according to any one of claims 1-8.

11. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that cause a processor to execute the blood vessel segmentation method according to any one of claims 1-8.

Images