CN113344873A - Blood vessel segmentation method, device and computer readable medium - Google Patents
Blood vessel segmentation method, device and computer readable medium Download PDFInfo
- Publication number
- CN113344873A CN113344873A CN202110625471.1A CN202110625471A CN113344873A CN 113344873 A CN113344873 A CN 113344873A CN 202110625471 A CN202110625471 A CN 202110625471A CN 113344873 A CN113344873 A CN 113344873A
- Authority
- CN
- China
- Prior art keywords
- point
- blood vessel
- diagram
- points
- structure point
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 210000004204 blood vessel Anatomy 0.000 title claims abstract description 191
- 230000011218 segmentation Effects 0.000 title claims abstract description 138
- 238000000034 method Methods 0.000 title claims abstract description 59
- 238000010586 diagram Methods 0.000 claims abstract description 131
- 238000000605 extraction Methods 0.000 claims abstract description 43
- 238000002591 computed tomography Methods 0.000 claims description 59
- 210000002989 hepatic vein Anatomy 0.000 claims description 46
- 238000012549 training Methods 0.000 claims description 36
- 238000012545 processing Methods 0.000 claims description 31
- 210000003240 portal vein Anatomy 0.000 claims description 24
- 230000006870 function Effects 0.000 claims description 20
- 230000009466 transformation Effects 0.000 claims description 10
- PXFBZOLANLWPMH-UHFFFAOYSA-N 16-Epiaffinine Natural products C1C(C2=CC=CC=C2N2)=C2C(=O)CC2C(=CC)CN(C)C1C2CO PXFBZOLANLWPMH-UHFFFAOYSA-N 0.000 claims description 9
- 238000004590 computer program Methods 0.000 claims description 2
- 210000004185 liver Anatomy 0.000 description 20
- 230000008569 process Effects 0.000 description 12
- 230000008878 coupling Effects 0.000 description 3
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 238000003745 diagnosis Methods 0.000 description 3
- 210000000056 organ Anatomy 0.000 description 3
- 238000004891 communication Methods 0.000 description 2
- 238000013527 convolutional neural network Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 238000012163 sequencing technique Methods 0.000 description 2
- 230000002792 vascular Effects 0.000 description 2
- 206010028980 Neoplasm Diseases 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 210000004556 brain Anatomy 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 210000002784 stomach Anatomy 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 210000001835 viscera Anatomy 0.000 description 1
- 208000037911 visceral disease Diseases 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0012—Biomedical image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/14—Transformations for image registration, e.g. adjusting or mapping for alignment of images
- G06T3/147—Transformations for image registration, e.g. adjusting or mapping for alignment of images using affine transformations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
- G06T7/33—Determination of transform parameters for the alignment of images, i.e. image registration using feature-based methods
- G06T7/337—Determination of transform parameters for the alignment of images, i.e. image registration using feature-based methods involving reference images or patches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10072—Tomographic images
- G06T2207/10081—Computed x-ray tomography [CT]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30004—Biomedical image processing
- G06T2207/30101—Blood vessel; Artery; Vein; Vascular
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30204—Marker
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Biophysics (AREA)
- Software Systems (AREA)
- Evolutionary Computation (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Medical Informatics (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Radiology & Medical Imaging (AREA)
- Quality & Reliability (AREA)
- Apparatus For Radiation Diagnosis (AREA)
Abstract
The embodiment of the invention discloses a blood vessel segmentation method, a device and a computer readable medium, wherein the method comprises the following steps: carrying out image feature extraction on the CT image to be detected to obtain a multi-scale feature image; marking structural points of the blood vessel in the CT image to be detected to obtain a structural point diagram, wherein the structural points are used for indicating the end points of the blood vessel and the intersection points of the branches of the blood vessel; then, extracting structural point features of the structural point diagram to obtain a structural point feature diagram, wherein the structural point feature diagram is used for indicating the position features of the structural points; and finally, segmenting the blood vessel based on the multi-scale feature map and the structural point feature map to obtain a blood vessel segmentation result. Therefore, the image feature extraction and the structure point feature extraction are respectively carried out on the CT image to be detected, so that the slender and criss-cross blood vessels can be accurately identified, the segmentation of the blood vessels is facilitated, and the integrity of the blood vessel segmentation is improved.
Description
Technical Field
The invention relates to the technical field of artificial intelligence, in particular to a blood vessel segmentation method, a blood vessel segmentation device and a computer readable medium.
Background
A living organism is a very complex system. The distribution of blood vessels in each organ is also intricate and complicated for each organ in the body. Blood vessel detection is the basis of diagnosis of visceral diseases, and therefore, it is important to accurately segment blood vessels in the viscera.
The traditional blood vessel segmentation method is mainly characterized in that a trained segmentation model is used for predicting a CT image to be measured, so that a blood vessel segmentation result is obtained; however, some blood vessels in the zang-fu organs not only belong to long and thin blood vessels but also are criss-cross with other blood vessels in distribution, so the current blood vessel segmentation method often has the problem of incomplete segmentation caused by blood vessel misrecognition. Therefore, for the slender blood vessels with more complicated distribution, it is urgently needed to provide an effective blood vessel segmentation method to improve the completeness and accuracy of blood vessel segmentation.
Disclosure of Invention
In order to solve the problems in the prior art, embodiments of the present invention provide a blood vessel segmentation method, a blood vessel segmentation device, and a computer readable medium, which can accurately identify a blood vessel, thereby improving the integrity of blood vessel segmentation.
To achieve the above object, according to a first aspect of embodiments of the present invention, there is provided a blood vessel segmentation method, including: carrying out image feature extraction on an electronic Computed Tomography (CT) image to be detected to obtain a multi-scale feature map; marking structural points of a blood vessel in a CT image to be detected to obtain a structural point diagram, wherein the structural points are used for indicating end points of the blood vessel and intersection points of branches of the blood vessel; extracting structural point features from the structural point diagram to obtain a structural point feature diagram, wherein the structural point feature diagram is used for indicating the position features of the structural points; and carrying out segmentation processing on the blood vessel based on the multi-scale feature map and the structural point feature map to obtain a blood vessel segmentation result.
Optionally, the extracting the feature of the structure point from the structure point diagram to obtain the feature diagram of the structure point includes: for any structure point in the structure point diagram: carrying out position coding on the structure points to obtain a structure point position diagram; performing relevance coding on the structure points to obtain a structure point relevance diagram, wherein the structure point relevance diagram is used for indicating the position relation between the adjacent structure points; and determining the structural point position graph and the structural point association graph as a structural point feature graph.
Optionally, the segmenting the blood vessel based on the multi-scale feature map and the structure point feature map to obtain a blood vessel segmentation result, including: extracting the structural point feature map to obtain a blood vessel region, wherein the blood vessel region comprises structural points; registering the structure points in the blood vessel region by using the blood vessel structure point template map to obtain a registration function; carrying out affine transformation processing on the multi-scale feature map and the structure point feature map by using the registration function; and carrying out segmentation processing on the blood vessel based on the transformed multi-scale feature map and the structural point feature map to obtain a blood vessel segmentation result.
Optionally, the extracting the feature of the structure point from the structure point diagram to obtain the feature diagram of the structure point includes: grading all structure points in the structure point diagram to obtain a plurality of grades of structure points; and carrying out feature extraction on the structure points at the same level to obtain a structure point feature map.
Optionally, the vascular hepatic vein and/or portal vein.
Optionally, the segmenting the blood vessel based on the multi-scale feature map and the structure point feature map to obtain a blood vessel segmentation result, including: acquiring an original CT image; respectively carrying out image feature extraction and blood vessel structure point feature extraction on the original CT image to obtain an original multi-scale feature map and an original structure point feature map; taking the original multi-scale feature map and the original structure point feature map as training samples together, and performing model training by using a plurality of training samples to obtain a segmentation model; and carrying out segmentation processing on the multi-scale feature map and the structure point feature map by using a segmentation model to obtain a blood vessel segmentation result.
To achieve the above object, according to a second aspect of the embodiments of the present invention, there is also provided a blood vessel segmentation apparatus, including: the image extraction module is used for extracting image characteristics of the CT image to be detected to obtain a multi-scale characteristic image; the structure point marking module is used for marking the structure points of the blood vessels in the CT image to be detected to obtain a structure point diagram, and the structure points are used for indicating the end points of the blood vessels and the intersection points of the branches of the blood vessels; the structure point extraction module is used for extracting structure point features of the structure point diagram to obtain a structure point feature diagram, and the structure point feature diagram is used for indicating the position features of the structure points; and the segmentation processing module is used for carrying out segmentation processing on the blood vessel based on the multi-scale feature map and the structure point feature map to obtain a blood vessel segmentation result.
Optionally, the structure point extracting module includes: an encoding unit configured to, for any structure point in the structure point map: carrying out position coding on the structure points to obtain a structure point position diagram; performing relevance coding on the structure points to obtain a structure point relevance diagram, wherein the structure point relevance diagram is used for indicating the position relation between the adjacent structure points; and the determining unit is used for determining the structural point position graph and the structural point association graph as structural point feature graphs.
To achieve the above object, according to a third aspect of the embodiments of the present invention, there is also provided an electronic device, including one or more processors; memory for storing one or more programs which, when executed by the one or more processors, cause the one or more processors to implement the vessel segmentation method as described in the first aspect.
To achieve the above object, according to a fourth aspect of the embodiments of the present invention, there is further provided a computer readable medium having a computer program stored thereon, the program, when executed by a processor, implementing the vessel segmentation method according to the first aspect.
Compared with the prior art, the blood vessel segmentation method, the device and the computer readable medium provided by the embodiment of the invention firstly extract the image characteristics of the CT image to be detected to obtain a multi-scale characteristic image; marking structural points of the blood vessels in the CT image to be detected to obtain a structural point diagram, wherein the structural points are used for indicating end points of the blood vessels and intersection points of branches of the blood vessels; then, extracting structural point features of the structural point diagram to obtain a structural point feature diagram, wherein the structural point feature diagram is used for indicating the position features of the structural points; and finally, segmenting the blood vessel based on the multi-scale feature map and the structural point feature map to obtain a blood vessel segmentation result. Therefore, image feature extraction and structure point feature extraction are respectively carried out on the CT image to be detected, so that the slender and criss-cross blood vessels can be accurately identified, the segmentation of the blood vessels is facilitated, the integrity of the blood vessel segmentation is improved, and the problem that the segmentation is incomplete due to the fact that the slender blood vessels are complex in distribution in the prior art is effectively solved.
It is to be understood that the teachings of the present invention need not achieve all of the above-described benefits, but rather that specific embodiments may achieve specific technical results, and that other embodiments of the present invention may achieve benefits not mentioned above.
Drawings
The drawings are included to provide a better understanding of the invention and are not to be construed as unduly limiting the invention. Wherein like or corresponding reference numerals designate like or corresponding parts throughout the several views.
FIG. 1 is a schematic flow chart of a vessel segmentation method according to an embodiment of the present invention;
FIG. 2 is a schematic flow chart of a blood vessel segmentation method according to another embodiment of the present invention;
FIG. 3 is a schematic flow chart of a vessel segmentation method according to still another embodiment of the present invention;
FIG. 4 is a schematic flow chart of a vessel segmentation method according to an embodiment of the present invention;
FIG. 5 is a structure diagram of an embodiment of the present invention;
fig. 6 is a schematic block diagram of a blood vessel segmentation apparatus according to an embodiment of the present invention.
Detailed Description
Exemplary embodiments of the present invention are described below with reference to the accompanying drawings, in which various details of embodiments of the invention are included to assist understanding, and which are to be considered as merely exemplary. Accordingly, those of ordinary skill in the art will recognize that various changes and modifications of the embodiments described herein can be made without departing from the scope and spirit of the invention. Also, descriptions of well-known functions and constructions are omitted in the following description for clarity and conciseness.
Fig. 1 is a schematic flow chart of a blood vessel segmentation method according to an embodiment of the present invention. A blood vessel segmentation method comprises the following operation flows: s101, extracting image characteristics of a CT image to be detected to obtain a multi-scale characteristic image; s102, marking structure points of the blood vessels in the CT image to be detected to obtain a structure point diagram, wherein the structure points are used for indicating end points of the blood vessels and intersection points of branches of the blood vessels; s103, extracting structural point features of the structural point diagram to obtain a structural point feature diagram, wherein the structural point feature diagram is used for indicating the position features of the structural points; and S104, segmenting the blood vessel based on the multi-scale feature map and the structural point feature map to obtain a blood vessel segmentation result.
In S101, specifically, the trained image model is used to perform image feature extraction on the CT image to be detected, so as to obtain a multi-scale feature image. The CT image to be detected refers to a CT image of a specific part, for example, the CT image to be detected may be a liver CT image, a stomach CT image, a brain CT image, or the like, and the CT image to be detected is not particularly limited herein.
For example, a plurality of liver CT images and liver feature images are acquired; and taking the liver CT image as a training sample, inputting a plurality of training samples and the liver characteristic image into a convolutional neural network, performing model training to obtain a training result, and adjusting model parameters based on the training result to obtain an image model. And then, carrying out image feature extraction on the liver CT image to be detected by using the trained image model to obtain a multi-scale liver feature image. Here, multi-scale in the multi-scale liver feature image is used to indicate a plurality of liver feature images different in resolution. The liver feature image includes not only the vascular features of the liver but also other tissue features of the liver.
In S102, specifically, a specified blood vessel in the CT image to be measured is determined, and the end point of the specified blood vessel and the intersection point of the branch are respectively marked with a structure point to obtain a structure point diagram. Therefore, by marking the end point of the blood vessel and the intersection point of the branch, the trend of the blood vessel can be determined, and the integrity of the blood vessel segmentation is improved.
In S103, structure point feature extraction may be performed on each structure point in the structure point diagram to obtain a structure point feature diagram, where one structure point corresponds to one structure point feature diagram. Or extracting the structural point features of a plurality of structural points in the structural point diagram at the same time to obtain a structural point feature diagram; for example, all structure points in the structure chart are graded to obtain a plurality of grades of structure points; and performing feature extraction on a plurality of structure points at the same level to obtain a structure point feature map, wherein the structure points at the same level correspond to one structure point feature map.
It should be noted that there are a plurality of structure point feature maps obtained by extracting the structure point features of all the structure points in the structure point map.
In S104, an original CT image is acquired; carrying out image feature extraction on an original CT image by using an image model to obtain an original multi-scale feature map; respectively marking structure points of the end point of a specified blood vessel and the intersection point of the branch in the original CT image to obtain an original structure point diagram; extracting the structural point feature of the original structural point diagram to obtain an original structural point feature diagram; and the original multi-scale feature map and the original structure point feature map are jointly used as training samples, the training samples are input into a segmentation network for model training to obtain training results, and parameters of the model are adjusted based on the training results to obtain a segmentation model. Therefore, on the basis of the feature learning of the blood vessel image, the feature learning of the position of the blood vessel structure point is combined, so that the segmentation model can be accurately trained, the segmentation of the blood vessel image is facilitated, and the accuracy of the blood vessel segmentation is improved.
Then, inputting the multi-scale feature map and the structure point feature map into a trained segmentation model, and predicting the blood vessel by using the trained segmentation model to obtain a blood vessel segmentation image; therefore, the structural point feature map and the multi-scale feature map are matched for vessel segmentation, and the accuracy of vessel segmentation is improved.
Here, the segmentation network may be a full convolutional neural network, which is composed of 10 cascaded residual units.
The blood vessel segmentation method according to the present embodiment is applicable to any blood vessel segmentation, and is also applicable to a blood vessel which is long and narrow and has a cross with other blood vessels, for example, the segmentation of the portal vein and the hepatic vein in the liver.
According to the embodiment of the invention, the structural point of the CT image to be detected is marked, and the structural point feature of the marked structural point diagram is extracted to obtain a structural point feature diagram; and then, segmenting the blood vessel based on the structural point feature map and the multi-scale feature map to obtain a blood vessel segmentation image. Therefore, the blood vessel can be accurately segmented, and the problem of incomplete segmentation caused by complicated distribution of slender blood vessels in the prior art is solved.
Fig. 2 shows a schematic flow chart of a blood vessel segmentation method according to another embodiment of the present invention. The embodiment is further optimized on the basis of the previous embodiment. A blood vessel segmentation method at least comprises the following operation flows: s201, carrying out image feature extraction on a CT image to be detected to obtain a multi-scale feature map; s202, marking structure points of the blood vessels in the CT image to be detected to obtain a structure point diagram, wherein the structure points are used for indicating end points of the blood vessels and intersection points of branches of the blood vessels; s203, for any structure point in the structure point diagram: carrying out position coding on the structure points to obtain a structure point position diagram; performing relevance coding on the structure points to obtain a structure point relevance diagram, wherein the structure point relevance diagram is used for indicating the position relation between the adjacent structure points; s204, determining the structural point position diagram and the structural point association diagram as a structural point feature diagram; and S205, segmenting the blood vessel based on the multi-scale feature map and the structure point feature map to obtain a blood vessel segmentation result.
The specific implementation processes of S201, S202, and S205 are similar to the specific implementation processes of S101, S102, and S104 in the embodiment shown in fig. 1, and are not described here again.
In S203 and S204, the structure point position diagramPosition information for representing the structure points. The structure location map may include location information of one structure point, or may include location information of a plurality of structure points. The structure point position map may be a gaussian distribution thermodynamic diagram centered on the structure point or may be a position coordinate map. Taking a Gaussian distribution thermodynamic diagram with a structure point as a center as an example, the pixel value of the corresponding position of the structure point is 1, and the pixel values around the structure point are represented by
Is decreased, thereby representing the position information of the structure point by means of probability distribution.
The structure point association map is used to represent the positional relationship between adjacent structure points. The structure point correlation map may include a positional relationship with respect to one structure point, or may include a positional relationship with respect to a plurality of structure points. Specifically, the structure point correlation map may be a unit vector map of adjacent structure points, or may be an oblique square difference matrix map. The unit vector diagram of the adjacent structure points comprises an X-axis coordinate diagram, a Y-axis coordinate diagram and a Z-axis coordinate diagram.
In the embodiment, a structural point position diagram and a structural point association diagram are obtained by extracting structural point features; therefore, the position and the trend of the structure point can be accurately positioned, so that the segmentation of the blood vessel image is facilitated, and the accuracy of the segmentation of the blood vessel image is improved.
Fig. 3 is a schematic flow chart of a blood vessel segmentation method according to still another embodiment of the present invention. The embodiment is further optimized on the basis of the previous embodiment. A blood vessel segmentation method comprises the following operation flows: s301, extracting image characteristics of the CT image to be detected to obtain a multi-scale characteristic image; s302, marking structure points of the blood vessel in the CT image to be detected to obtain a structure point diagram, wherein the structure points are used for indicating the end points of the blood vessel and the intersection points of the branches of the blood vessel; s303, carrying out grade division on all structure points in the structure point diagram to obtain a plurality of grades of structure points; s304, aiming at the same level structure points in the structure point diagram: carrying out position coding on all structure points of the same grade to obtain a structure point position diagram; performing relevance coding on all structure points at the same level to obtain a structure point relevance diagram, wherein the structure point relevance diagram is used for indicating the position relation between adjacent structure points; s305, determining the structural point position diagram and the structural point association diagram as a structural point feature diagram; s306, extracting the structural point feature map to obtain a blood vessel region, wherein the blood vessel region comprises structural points; s307, registering the structure points in the blood vessel region by using the blood vessel structure point template picture to obtain a registration function; s308, performing affine transformation processing on the multi-scale feature map and the structure point feature map by using a registration function; s309, based on the transformed multi-scale feature map and the structural point feature map, segmenting the blood vessel to obtain a blood vessel segmentation result.
The specific implementation processes of S301 and S302 are similar to those of S101 and S102 in the embodiment shown in fig. 1, and are not described here again.
Determining a structure point at the root of the blood vessel as a root node in S303 to S304; for structural points on either vessel branch: determining the distance between the structure point and the root node, sequencing the distances from small to large, determining the grade of the structure point according to the sequencing result, for example, the structure point corresponding to the minimum distance is a grade 1 structure point, and so on, thereby obtaining the structure points with different grades. On the same branch, the closer the structure point is to the root node, the lower the grade of the structure point is, and the farther the structure point is from the root node, the higher the grade of the structure point is. Then, carrying out position coding on all structure points belonging to the same grade on different branches of the blood vessel to obtain a structure point position diagram; at this time, the structure point position map includes position information of a plurality of structure points at the same level. And performing relevance coding on all structure points belonging to the same level on different branches of the blood vessel to obtain a structure point relevance map, wherein the structure point relevance map comprises position relation information corresponding to each structure point in the same level. Therefore, all the structure points in the structure point diagram are graded, and the structure point feature extraction is simultaneously carried out on the structure points at the same grade, so that the feature extraction can be orderly carried out on all the structure points in the structure point diagram, repeated extraction and feature omission are avoided, the speed of feature extraction of the structure points is improved, and the diagnosis of later doctors on blood vessels is facilitated.
In S306 to S308, the structure point template map is obtained in advance, for example, CT images of a plurality of persons are obtained, and structure point calibration is performed on a specified blood vessel in each person CT image; and clustering the calibrated structure points to obtain a plurality of structure point template pictures. Then extracting a blood vessel region based on the structure points in the structure point feature map; selecting a structural point template picture with the highest similarity with the blood vessel region from the plurality of structural point template pictures; and registering the structure points in the blood vessel region by using the selected structure point template graph to obtain a registration function. And finally, respectively carrying out affine transformation processing on the multi-scale feature map, the structure point position map and the structure point association map by utilizing a registration function to obtain the transformed scale feature map, the structure point position map and the structure point association map. Therefore, the structure points of the blood vessel region are registered, and the registration function is used for carrying out affine transformation on the multi-scale feature map, the structure point position map and the structure point association map, so that the blood vessel segmentation is facilitated, and the accuracy of the blood vessel segmentation is improved.
In S309, the trained segmentation model is used to perform a vessel segmentation process on the transformed multi-scale feature map and the structural feature map corresponding to the CT image to be measured, so as to obtain a vessel segmentation image.
The segmentation model is obtained by the following method: acquiring an original CT image; carrying out image feature extraction on an original CT image by using an image model to obtain an original multi-scale feature map; respectively marking structure points of the end points and the intersection points of the branches of the blood vessels in the original CT image to obtain an original structure point diagram; and extracting the structural point feature of the original structural point diagram to obtain an original structural point feature diagram. Carrying out grade division on all structure points in the original structure point diagram to obtain a plurality of grades of structure points; and respectively carrying out position coding and relevance coding on the same-level structure points in the structure point diagram to obtain an original structure point position diagram and an original structure point relevance diagram. Extracting the original structure point feature map to obtain an original blood vessel region with structure points; registering the structure points in the original blood vessel region by using the selected blood vessel structure point template picture to obtain a registration function; carrying out affine transformation processing on the original multi-scale feature map, the original structure point position map and the original structure point association map by utilizing a registration function; inputting the transformed original multi-scale feature map, the original structure point position map and the original structure point association map into a segmentation network, performing model training to obtain segmentation results, and adjusting model parameters based on a plurality of segmentation results to obtain a segmentation model. The greater the degree of difference between blood vessels of different patients, the more difficult the training of the segmentation model; therefore, the difference degree of different blood vessels can be reduced through registration, and the accuracy of the segmentation model training is improved.
In order to enable the segmentation network to effectively learn the image features and the structure point features, the original multi-scale feature map, the original structure point position map, and the original structure point correlation map are all adjusted to the same size as the original CT image size during model training.
In the embodiment, a structural point in a blood vessel region is registered by using a blood vessel structural point template map to obtain a registration function; and then, carrying out affine transformation processing on the multi-scale feature map, the structure point position map and the structure point association map by using a registration function, so that the blood vessel images are registered to a uniform direction according to the blood vessel structure points, the accuracy of blood vessel segmentation is improved, and a foundation is provided for tumor diagnosis and surgical planning.
It should be understood that, in the embodiments of the present invention, the sequence numbers of the above-mentioned processes do not mean the execution sequence, and the execution sequence of each process should be determined by its function and the inherent logic, and should not constitute any limitation to the implementation process of the embodiments of the present invention.
The blood vessel segmentation method of the above embodiment will be described in detail below with reference to specific application scenarios.
Because the hepatic vein and the portal vein belong to the slender blood vessel, and the hepatic vein and the portal vein are distributed in the liver in a criss-cross mode. In order to effectively segment the hepatic vein and the portal vein, the blood vessel segmentation method of the present embodiment is used to segment the hepatic vein and the portal vein, and the specific process is as follows with reference to fig. 4.
As shown in fig. 4, inputting a CT image of a liver to be detected into an image model for image feature extraction, and outputting a multi-scale feature map; and marking structural points of the to-be-detected liver CT image to obtain a hepatic vein structure point diagram and a portal vein structure point diagram. Respectively carrying out position coding and relevance coding on the hepatic vein structure point diagram to obtain a structural point position diagram and a structural point relevance diagram corresponding to the hepatic vein; and respectively carrying out position coding and relevance coding on the portal vein structure point diagram to obtain a structure point position diagram and a structure point relevance diagram corresponding to the portal vein. And then inputting the multi-scale characteristic map, the structural point position map and the structural point association map corresponding to the hepatic vein, and the structural point position map and the structural point association map corresponding to the portal vein into a segmentation model, and performing blood vessel segmentation processing to obtain a blood vessel segmentation result. Further, after the feature of the structural point is extracted, structural point registration processing is required to obtain a multi-scale feature map, a structural point position map and a structural point association map corresponding to the transformed hepatic vein, and a multi-scale feature map, a structural point position map and a structural point association map corresponding to the transformed portal vein; and then inputting the multi-scale feature map, the structure point position map and the structure point association map corresponding to the transformed hepatic vein, and the multi-scale feature map, the structure point position map and the structure point association map corresponding to the transformed portal vein into the trained segmentation model, and performing segmentation processing to obtain a blood vessel segmentation result. Here, the blood vessel segmentation result is used to indicate a hepatic vein segmentation image and a portal vein segmentation image.
As shown in fig. 5, a portal vein structure point diagram and a hepatic vein structure point diagram can be clearly shown by fig. 5, in which gray dots represent hepatic vein structure points and black dots represent portal vein structure points.
The segmentation model training comprises two stages of image preprocessing and model training.
Image preprocessing includes image feature extraction, structure point labeling, structure point feature extraction, and structure registration. Specifically, the method comprises the following steps: and acquiring a liver CT image, and performing image feature extraction on the liver CT image by using the trained image model to obtain a multi-scale feature map. And marking the end point of the hepatic vein blood vessel in the liver CT image and the intersection point of the branch to obtain a hepatic vein structure point diagram. Acquiring a root node in a hepatic vein structure point diagram; and determining the structure point adjacent to the root node as a level 1 structure point, determining the structure point next adjacent to the root node as a level 2 structure point, and so on to determine the level of the hepatic vein structure point. As can be seen from fig. 5, there are 2 hepatic vein class 1 structural points, 3 hepatic vein class 2 structural points, and 7 hepatic vein class 3 structural points. Then, carrying out position coding and relevance coding on structure points belonging to the same level in hepatic veins; and obtaining a structural point position diagram and a structural point association diagram of the hepatic vein. Finally, extracting a hepatic vein region including hepatic vein structure points from the hepatic vein structure point diagram, selecting a hepatic vein structure point template diagram similar to the hepatic vein region from the hepatic vein structure point template diagrams, and registering the structure points in the hepatic vein region by using the selected hepatic vein structure point template diagram to obtain a registration function; and then carrying out affine transformation processing on the multi-scale feature map, the hepatic vein structure point position map and the hepatic vein structure point association map by utilizing a registration function to obtain the transformed multi-scale feature map, the hepatic vein structure point position map and the hepatic vein structure point association map. Similarly, the portal vein in the liver CT image is preprocessed in the same way to obtain a transformed multi-scale feature map, a portal vein structure point position map and a portal vein structure point association map.
The specific process of model training is as follows: taking the transformed multi-scale feature map, the hepatic vein structure point position map and the hepatic vein structure point association map corresponding to the hepatic vein as a first training sample, and taking the transformed multi-scale feature map, the hepatic vein structure point position map and the hepatic vein structure point association map corresponding to the portal vein as a second training sample; inputting the first training sample into a segmentation network, and performing model training to obtain a first training result; inputting the second catenary sample into the segmentation network, and performing model training to obtain a second training result; and adjusting the model parameters based on the plurality of first training results and the plurality of second training results to obtain the segmentation model.
Therefore, the blood vessel segmentation method can automatically identify the hepatic vein and the portal vein, improves the accuracy of hepatic vein and portal vein identification, can completely segment the hepatic vein and the portal vein, and solves the problems of misidentification and misinterpretation caused by complicated distribution of slender blood vessels in the prior art.
Fig. 6 is a schematic block diagram of a blood vessel segmentation apparatus according to an embodiment of the present invention. A vessel segmentation device, the device 600 comprising: the image extraction module 601 is configured to perform image feature extraction on a CT image to be detected to obtain a multi-scale feature map; a structure point marking module 602, configured to mark a structure point of a blood vessel in a CT image to be detected, so as to obtain a structure point diagram, where the structure point is used to indicate an end point of the blood vessel and an intersection point of branches of the blood vessel; a structure point extracting module 603, configured to perform structure point feature extraction on the structure point diagram to obtain a structure point feature diagram, where the structure point feature diagram is used to indicate a position feature of the structure point; and a segmentation processing module 604, configured to perform segmentation processing on the blood vessel based on the multi-scale feature map and the structure point feature map to obtain a blood vessel segmentation result.
In an alternative embodiment, the structure point extracting module 602 includes: an encoding unit configured to, for any structure point in the structure point map: carrying out position coding on the structure points to obtain a structure point position diagram; performing relevance coding on the structure points to obtain a structure point relevance diagram, wherein the structure point relevance diagram is used for indicating the position relation between the adjacent structure points; and the determining unit is used for determining the structure point position graph and the structure point association graph as the structure point feature graph.
In an alternative embodiment, the segmentation processing module 604 includes: the extraction processing unit is used for extracting the structural point feature map to obtain a blood vessel region, and the blood vessel region comprises structural points; the registration unit is used for registering the structure points in the blood vessel region by using the blood vessel structure point template picture to obtain a registration function; the transformation unit is used for carrying out affine transformation processing on the multi-scale feature map and the structure point feature map by utilizing the registration function; and the segmentation processing unit is used for carrying out segmentation processing on the blood vessel based on the transformed multi-scale feature map and the structural point feature map to obtain a blood vessel segmentation result.
In an alternative embodiment, the structure point extracting module 603 includes: the grade division unit is used for carrying out grade division on all structure points in the structure point diagram to obtain a plurality of grades of structure points; and the feature extraction unit is used for extracting features of the structure points at the same level to obtain a structure point feature map.
In alternative embodiments, the blood vessel is a hepatic vein and/or a portal vein.
In an alternative embodiment, the segmentation processing module 604 includes: an acquisition unit for acquiring an original CT image; the characteristic extraction unit is used for respectively carrying out image characteristic extraction and blood vessel structure point characteristic extraction on the original CT image to obtain an original multi-scale characteristic image and an original structure point characteristic image; the model training unit is used for taking the original multi-scale feature map and the original structure point feature map as training samples together, and performing model training by using a plurality of training samples to obtain a segmentation model; and the segmentation processing unit is used for performing segmentation processing on the multi-scale feature map and the structure point feature map by using the segmentation model to obtain a blood vessel segmentation result.
The device can execute the blood vessel segmentation method provided by the embodiment of the invention, and has corresponding functional modules and beneficial effects for executing the blood vessel segmentation method. For details of the blood vessel segmentation method provided in the embodiment of the present invention, reference may be made to the following description.
According to still another embodiment of the present invention, there is also provided an electronic apparatus including: one or more processors; the storage device is used for storing one or more programs, and when the one or more programs are executed by the one or more processors, the one or more processors implement the blood vessel segmentation method provided by the above embodiment of the invention.
It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.
In the several embodiments provided in the present application, it should be understood that the disclosed apparatus and method may be implemented in other ways. The above-described device embodiments are merely illustrative, for example, the division of a unit is only one logical function division, and there may be other division ways in actual implementation, such as: multiple units or components may be combined, or may be integrated into another device, or some features may be omitted, or not implemented. In addition, the coupling, direct coupling or communication connection between the components shown or discussed may be through some interfaces, and the indirect coupling or communication connection between the devices or units may be electrical, mechanical or other forms.
The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units; can be located in one place or distributed on a plurality of network units; some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.
In addition, all functional units in the embodiments of the present application may be integrated into one processing unit, or each unit may be separately regarded as one unit, or two or more units may be integrated into one unit; the integrated unit can be realized in a form of hardware, or in a form of hardware plus a software functional unit.
Those of ordinary skill in the art will understand that: all or part of the steps for realizing the method embodiments can be completed by hardware related to program instructions, the program can be stored in a computer readable storage medium, and the program executes the steps comprising the method embodiments when executed; and the aforementioned storage medium includes: various media capable of storing program codes, such as a removable storage medium, a Read Only Memory (ROM), a magnetic disk, and an optical disk.
Alternatively, the integrated units described above in the present application may be stored in a computer-readable storage medium if they are implemented in the form of software functional modules and sold or used as independent products. Based on such understanding, the technical solution of this embodiment of the present application may be essentially or partially implemented in the form of a software product, which is stored in a storage medium and includes several instructions to enable a computer device (which may be a personal computer, a server, or a network device) to perform the following operation steps: s101, extracting image characteristics of a CT image to be detected to obtain a multi-scale characteristic image; s102, marking structure points of the blood vessels in the CT image to be detected to obtain a structure point diagram, wherein the structure points are used for indicating end points of the blood vessels and intersection points of branches of the blood vessels; s103, extracting structural point features of the structural point diagram to obtain a structural point feature diagram, wherein the structural point feature diagram is used for indicating the position features of the structural points; and S104, segmenting the blood vessel based on the multi-scale feature map and the structural point feature map to obtain a blood vessel segmentation result.
And the aforementioned storage medium includes: a removable storage medium, a ROM, a magnetic disk, an optical disk, or the like, which can store the program code.
The above description is only for the specific embodiments of the present application, but the scope of the present application is not limited thereto, and any person skilled in the art can easily conceive of the changes or substitutions within the technical scope of the present application, and shall be covered by the scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.
Claims (10)
1. A method of vessel segmentation, comprising:
carrying out image feature extraction on an electronic Computed Tomography (CT) image to be detected to obtain a multi-scale feature map;
marking structural points of a blood vessel in a CT image to be detected to obtain a structural point diagram, wherein the structural points are used for indicating end points of the blood vessel and intersection points of branches of the blood vessel;
extracting structural point features from the structural point diagram to obtain a structural point feature diagram, wherein the structural point feature diagram is used for indicating the position features of the structural points;
and carrying out segmentation processing on the blood vessel based on the multi-scale feature map and the structural point feature map to obtain a blood vessel segmentation result.
2. The method according to claim 1, wherein the performing structure point feature extraction on the structure point diagram to obtain a structure point feature diagram comprises:
for any structure point in the structure point diagram: carrying out position coding on the structure points to obtain a structure point position diagram; performing relevance coding on the structure points to obtain a structure point relevance diagram, wherein the structure point relevance diagram is used for indicating the position relation between the adjacent structure points;
and determining the structural point position graph and the structural point association graph as a structural point feature graph.
3. The method according to claim 1 or 2, wherein the segmenting the blood vessel based on the multi-scale feature map and the structure point feature map to obtain a blood vessel segmentation result comprises:
extracting the structural point feature map to obtain a blood vessel region, wherein the blood vessel region comprises structural points;
registering the structure points in the blood vessel region by using the blood vessel structure point template map to obtain a registration function;
carrying out affine transformation processing on the multi-scale feature map and the structure point feature map by using the registration function;
and carrying out segmentation processing on the blood vessel based on the transformed multi-scale feature map and the structural point feature map to obtain a blood vessel segmentation result.
4. The method according to claim 1, wherein the performing structure point feature extraction on the structure point diagram to obtain a structure point feature diagram comprises:
grading all structure points in the structure point diagram to obtain a plurality of grades of structure points;
and carrying out feature extraction on the structure points at the same level to obtain a structure point feature map.
5. The method of claim 1, wherein the blood vessel is a hepatic vein and/or a portal vein.
6. The method according to claim 1, wherein the segmenting the blood vessel based on the multi-scale feature map and the structure point feature map to obtain a blood vessel segmentation result comprises:
acquiring an original CT image;
respectively carrying out image feature extraction and blood vessel structure point feature extraction on the original CT image to obtain an original multi-scale feature map and an original structure point feature map;
taking the original multi-scale feature map and the original structure point feature map as training samples together, and performing model training by using a plurality of training samples to obtain a segmentation model;
and carrying out segmentation processing on the multi-scale feature map and the structure point feature map by using a segmentation model to obtain a blood vessel segmentation result.
7. A vessel segmentation device, comprising:
the image extraction module is used for extracting image characteristics of the CT image to be detected to obtain a multi-scale characteristic image;
the structure point marking module is used for marking the structure points of the blood vessels in the CT image to be detected to obtain a structure point diagram, and the structure points are used for indicating the end points of the blood vessels and the intersection points of the branches of the blood vessels;
the structure point extraction module is used for extracting structure point features of the structure point diagram to obtain a structure point feature diagram, and the structure point feature diagram is used for indicating the position features of the structure points;
and the segmentation processing module is used for carrying out segmentation processing on the blood vessel based on the multi-scale feature map and the structure point feature map to obtain a blood vessel segmentation result.
8. The apparatus of claim 8, wherein the structure point extraction module comprises:
an encoding unit configured to, for any structure point in the structure point map: carrying out position coding on the structure points to obtain a structure point position diagram; performing relevance coding on the structure points to obtain a structure point relevance diagram, wherein the structure point relevance diagram is used for indicating the position relation between the adjacent structure points;
and the determining unit is used for determining the structural point position graph and the structural point association graph as structural point feature graphs.
9. An electronic device, comprising:
one or more processors;
a memory for storing one or more programs,
the one or more programs, when executed by the one or more processors, cause the one or more processors to implement the method of any of claims 1-6.
10. A computer-readable medium, on which a computer program is stored which, when being executed by a processor, carries out the method according to any one of claims 1-6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110625471.1A CN113344873A (en) | 2021-06-04 | 2021-06-04 | Blood vessel segmentation method, device and computer readable medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110625471.1A CN113344873A (en) | 2021-06-04 | 2021-06-04 | Blood vessel segmentation method, device and computer readable medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113344873A true CN113344873A (en) | 2021-09-03 |
Family
ID=77475297
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110625471.1A Pending CN113344873A (en) | 2021-06-04 | 2021-06-04 | Blood vessel segmentation method, device and computer readable medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113344873A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114155193A (en) * | 2021-10-27 | 2022-03-08 | 北京医准智能科技有限公司 | Blood vessel segmentation method and device based on feature enhancement |
| CN115187512A (en) * | 2022-06-10 | 2022-10-14 | 珠海市人民医院 | Hepatocellular carcinoma great vessel invasion risk prediction method, system, device and medium |
-
2021
- 2021-06-04 CN CN202110625471.1A patent/CN113344873A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114155193A (en) * | 2021-10-27 | 2022-03-08 | 北京医准智能科技有限公司 | Blood vessel segmentation method and device based on feature enhancement |
| CN114155193B (en) * | 2021-10-27 | 2022-07-26 | 北京医准智能科技有限公司 | Blood vessel segmentation method and device based on feature enhancement |
| CN115187512A (en) * | 2022-06-10 | 2022-10-14 | 珠海市人民医院 | Hepatocellular carcinoma great vessel invasion risk prediction method, system, device and medium |
| CN115187512B (en) * | 2022-06-10 | 2024-01-30 | 珠海市人民医院 | Method, system, device and medium for predicting invasion risk of large blood vessel of hepatocellular carcinoma |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108648172B (en) | CT (computed tomography) map pulmonary nodule detection system based on 3D-Unet | |
| CN109166133B (en) | Soft tissue organ image segmentation method based on key point detection and deep learning | |
| CN108520519B (en) | Image processing method and device and computer readable storage medium | |
| CN110310256B (en) | Coronary stenosis detection method, coronary stenosis detection device, computer equipment and storage medium | |
| CN111462047B (en) | Vascular parameter measurement method, vascular parameter measurement device, vascular parameter measurement computer device and vascular parameter measurement storage medium | |
| CN113344873A (en) | Blood vessel segmentation method, device and computer readable medium | |
| CN113223013B (en) | Method, device, equipment and storage medium for pulmonary vessel segmentation positioning | |
| CN112037146B (en) | Automatic correction method and device for medical image artifacts and computer equipment | |
| CN112102294A (en) | Training method and device for generating countermeasure network, and image registration method and device | |
| CN110910441A (en) | Method and device for extracting center line | |
| CN113808125A (en) | Medical image processing method, focus type identification method and related product | |
| CN115713590A (en) | Three-dimensional reconstruction image processing method and system based on CT | |
| CN114155193B (en) | Blood vessel segmentation method and device based on feature enhancement | |
| EP3858245B1 (en) | Automatic recognition method for measurement point in cephalo image | |
| CN110634554A (en) | Spine image registration method | |
| CN115049660B (en) | Method and device for positioning characteristic points of cardiac anatomical structure | |
| CN111599444A (en) | Intelligent tongue diagnosis detection method and device, intelligent terminal and storage medium | |
| Lee et al. | Enhancement for automatic extraction of RoIs for bone age assessment based on deep neural networks | |
| CN113989269B (en) | Traditional Chinese medicine tongue image tooth trace automatic detection method based on convolutional neural network multi-scale feature fusion | |
| CN115661142A (en) | Tongue diagnosis image processing method, device and medium based on key point detection | |
| Rashid et al. | Segmenting melanoma lesion using single shot detector (SSD) and level set segmentation technique | |
| CN107610105B (en) | Method, device and equipment for positioning ROI and machine-readable storage medium | |
| CN114241198A (en) | Method, device, equipment and storage medium for obtaining local imagery omics characteristics | |
| CN114141337A (en) | Method, system and application for constructing image automatic annotation model | |
| CN113111879A (en) | Cell detection method and system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |