[go: up one dir, main page]

CN115496743B - Cerebrovascular lesion segmentation method, device, storage medium, and electronic device - Google Patents

Cerebrovascular lesion segmentation method, device, storage medium, and electronic device

Info

Publication number
CN115496743B
CN115496743B CN202211256557.2A CN202211256557A CN115496743B CN 115496743 B CN115496743 B CN 115496743B CN 202211256557 A CN202211256557 A CN 202211256557A CN 115496743 B CN115496743 B CN 115496743B
Authority
CN
China
Prior art keywords
lesions
segmentation
white matter
model
distribution
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.)
Active
Application number
CN202211256557.2A
Other languages
Chinese (zh)
Other versions
CN115496743A (en
Inventor
刘伟奇
马学升
陈磊
陈金钢
徐鹏
赵友源
赵晓彤
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Kunming Tongxin Yilian Technology Co ltd
Original Assignee
Kunming Tongxin Yilian Technology Co ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Kunming Tongxin Yilian Technology Co ltd filed Critical Kunming Tongxin Yilian Technology Co ltd
Priority to CN202211256557.2A priority Critical patent/CN115496743B/en
Publication of CN115496743A publication Critical patent/CN115496743A/en
Application granted granted Critical
Publication of CN115496743B publication Critical patent/CN115496743B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; 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 OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/20—Image enhancement or restoration using local operators
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; 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
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
    • G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; 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
    • G06T2207/10088—Magnetic resonance imaging [MRI]
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; 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
    • G06T2207/30096—Tumor; Lesion

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Medical Informatics (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Databases & Information Systems (AREA)
  • Software Systems (AREA)
  • Evolutionary Computation (AREA)
  • Computing Systems (AREA)
  • Artificial Intelligence (AREA)
  • Multimedia (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Radiology & Medical Imaging (AREA)
  • Quality & Reliability (AREA)
  • Magnetic Resonance Imaging Apparatus (AREA)
  • Image Processing (AREA)
  • Image Analysis (AREA)

Abstract

The application discloses a method, a device, a storage medium and an electronic device for segmenting cerebrovascular diseases, wherein the method comprises the steps of obtaining MRI image information, obtaining segmentation results in the MRI image information through a pre-trained image segmentation model, carrying out MAP estimation on a posterior distribution model for describing tissue types and space prior distribution of white matter loose tissues by the image segmentation model through a maximum posterior probability model, and obtaining segmentation results for separating white matter loose lesions from acute and chronic apoplexy lesions in the cerebrovascular diseases according to the segmentation results. The application realizes the automatic segmentation of the cerebrovascular disease and has strong consistency with the expert segmentation amount. The method can be used for clinical image processing and can improve the registration accuracy of clinical images.

Description

Cerebrovascular disease segmentation method, device, storage medium and electronic device
Technical Field
The application relates to the fields of medical technology and machine learning, in particular to a method and a device for segmenting cerebrovascular diseases, a storage medium and an electronic device.
Background
Identification of cerebrovascular abnormalities from brain magnetic resonance imaging (Megnetic Resonace Imaging, MRI) images is critical to understanding cerebral ischemia (insufficient cerebral blood flow). However, different lesion types, such as white matter osteoporosis (small vascular lesions) and stroke, cannot be distinguished purely by the shape or position of the image. Clinicians use anatomical and other medical knowledge to classify and describe lesions.
To understand the susceptibility to cerebral ischemia and the associated risk factors, clinicians manually delineate and analyze the vascular lesions, focus on white matter osteoporosis and separate it from stroke lesions. This approach shows that the white matter loosening burden is lower in patients with transient ischemic attacks than in patients with traumatic cerebral infarction. Each delineated patient required 30 minutes for white matter loosening and stroke lesions, however large studies contained hundreds of thousands of patients. Therefore, automatic segmentation is necessary.
Variability in the shape and location of lesions is one of the major challenges of stroke scan automatic segmentation. The T2-fluid attenuation reversal sequence (Fluid Attenuated Inversion Recovery, FLAIR) appears as a high signal at the site of white matter loose lesions, located around the ventricle, widely varying, approximately bilaterally symmetrical. Although stroke lesions are also characterized by high intensity, they can occur almost anywhere in the brain and vary widely in size and shape. In addition, acute strokes (strokes within the past 48 hours) are visible on dispersion weighted MR (DWI), but chronic strokes (strokes that occur long before imaging) are not.
Furthermore, due to the extremely limited scan time, the image quality in a practical clinical environment is very low, typically 5-7mm thick in image slices, even with bright artifacts. These factors prevent the registration accuracy of clinical images and affect intensity equalization.
Aiming at the problems that the method for segmenting the cerebrovascular lesions cannot realize automatic segmentation and has poor segmentation precision in the related technology, no effective solution is proposed at present.
Disclosure of Invention
The application mainly aims to provide a method, a device, a storage medium and an electronic device for dividing cerebrovascular diseases, which are used for solving the problems that the method for dividing the cerebrovascular diseases cannot realize automatic division and has poor division precision.
In order to achieve the above object, according to one aspect of the present application, there is provided a method of segmenting cerebrovascular lesions.
The method for segmenting the cerebrovascular disease comprises the following steps:
Acquiring MRI image information;
Obtaining a segmentation result in the MRI image information through a pre-trained image segmentation model, wherein the image segmentation model adopts a maximum posterior probability model to carry out MAP estimation on a posterior distribution model for describing tissue types and the spatial prior distribution of white matter loose tissues;
and according to the segmentation result, obtaining a segmentation result for separating white matter loose lesions from acute and chronic stroke lesions in the cerebrovascular lesions.
In some embodiments, the image segmentation model comprises intensity features including intensity distribution of white matter loose lesions and intensity distribution of strokes, shape features, and spatial environment features including spatial distribution of white matter loose lesions, the segmentation results further comprising:
Based on the T2 FLAIR sequence, white matter loose lesions were separated from acute and chronic stroke lesions in the cerebrovascular lesions.
In some embodiments, the segmentation result in the MRI image information includes:
Using the generative model to describe the spatial distribution, shape and appearance of healthy tissue and cerebrovascular lesions, building a posterior distribution model describing tissue classes, wherein the prior of tissue classes captures knowledge of spatial distribution and lesion shape;
And carrying out MAP estimation on the posterior distribution model for describing the tissue type by adopting a maximum posterior probability model to obtain a segmentation result in the MRI image information.
In some embodiments, the segmentation result in the MRI image information includes:
and constructing a probability map by adopting PCA to a training set of binary segmentation mapping of artificially segmented white matter loose tissue lesions, and modeling the spatial range of white matter loose tissue lesions to obtain the spatial prior distribution of white matter loose tissue.
In some embodiments, the MAP estimate is based on an EM algorithm and models the intensity average estimate as a spatial variation while filtering with a low pass filter.
In some embodiments, the image segmentation model employs artificially labeled white matter loose tissue lesion results as the training image.
In order to achieve the above object, according to still another aspect of the present application, there is provided a clinical image processing method.
The clinical image processing method according to the present application includes:
and carrying out image registration by adopting the cerebral vascular lesion segmentation method.
In order to achieve the above object, according to another aspect of the present application, there is provided a cerebrovascular disease dividing apparatus.
The cerebrovascular disease segmentation device according to the present application comprises:
the acquisition module is used for acquiring MRI image information;
The processing module is used for obtaining a segmentation result in the MRI image information through a pre-trained image segmentation model, wherein the image segmentation model adopts a maximum posterior probability model to carry out MAP estimation on a posterior distribution model for describing tissue types and the spatial prior distribution of white matter loose tissues;
And the segmentation module is used for obtaining a segmentation result for separating the white matter loose lesions from the acute and chronic stroke lesions in the cerebrovascular lesions according to the segmentation result.
According to the method, the device, the storage medium and the electronic device for segmenting the cerebrovascular diseases, MRI image information is acquired, segmentation results in the MRI image information are obtained through a pre-trained image segmentation model, the image segmentation model adopts a maximum posterior probability model to carry out MAP estimation on a posterior distribution model for describing tissue types and space prior distribution of white matter loose tissues, and then segmentation results for separating white matter loose diseases from acute and chronic stroke diseases in the cerebrovascular diseases are obtained according to the segmentation results. And (3) separating white matter loose lesions from acute and chronic apoplexy lesions in the cerebrovascular lesions through an image segmentation model.
Drawings
The accompanying drawings, which are included to provide a further understanding of the application, are incorporated in and constitute a part of this specification. The drawings and their description are illustrative of the application and are not to be construed as unduly limiting the application. In the drawings:
FIG. 1 is a flow chart of a method for segmentation of cerebrovascular lesions, according to an embodiment of the present application;
fig. 2 is a schematic structural view of a cerebrovascular disease segmentation apparatus according to an embodiment of the present application;
Fig. 3 is a schematic diagram of the implementation principle of a method for segmenting cerebrovascular lesions according to an embodiment of the present application.
Detailed Description
In order that those skilled in the art will better understand the present application, a technical solution in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in which it is apparent that the described embodiments are only some embodiments of the present application, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present application without making any inventive effort, shall fall within the scope of the present application.
It should be noted that the terms "first," "second," and the like in the description and the claims of the present application and the above figures are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate in order to describe the embodiments of the application herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
In the present application, the terms "upper", "lower", "left", "right", "front", "rear", "top", "bottom", "inner", "outer", "middle", "vertical", "horizontal", "lateral", "longitudinal" and the like indicate an azimuth or a positional relationship based on that shown in the drawings. These terms are only used to better describe the present application and its embodiments and are not intended to limit the scope of the indicated devices, elements or components to the particular orientations or to configure and operate in the particular orientations.
Also, some of the terms described above may be used to indicate other meanings in addition to orientation or positional relationships, for example, the term "upper" may also be used to indicate some sort of attachment or connection in some cases. The specific meaning of these terms in the present application will be understood by those of ordinary skill in the art according to the specific circumstances.
Furthermore, the terms "mounted," "configured," "provided," "connected," "coupled," and "sleeved" are to be construed broadly. For example, they may be fixedly connected, detachably connected, or of unitary construction, they may be mechanically or electrically connected, they may be directly connected, or they may be indirectly connected through intermediaries, or they may be in internal communication between two devices, elements or components. The specific meaning of the above terms in the present application can be understood by those of ordinary skill in the art according to the specific circumstances.
It should be noted that, without conflict, the embodiments of the present application and features of the embodiments may be combined with each other. The application will be described in detail below with reference to the drawings in connection with embodiments.
In an embodiment of the application, the intensity, shape and spatial distribution of lesions are modeled to obtain anatomical knowledge of different kinds of lesions in order to annotate MRI images of stroke patients. The present application is directed to segmenting white matter osteoporosis and separating it from stroke lesions.
Further, a probabilistic model of the effect of cerebrovascular disease on the brain was introduced. The model integrates the important characteristics of each lesion, thereby generating an effective reasoning algorithm to divide different tissues of a cerebral apoplexy patient. The spatial and intensity distribution of white matter osteoporosis and the intensity distribution of stroke are obtained. Modeling options of the present application capture concepts used by clinicians, such as symmetry and co-variation of intensity patterns, by training the model on an expert-labeled dataset. For example, the segmentation model of the present application combines several previously proposed methods of anatomical segmentation to accurately model lesions.
In addition, the inventors found that the intensity-based lesion segmentation algorithm uses the difference in tissue intensities to segment lesions. Spatial priors are sometimes added in the form of markov random fields or spatial distributions. These methods successfully describe structures that are high or low signal compared to the surrounding environment, such as MS lesions or tumors. However, these methods cannot be used to distinguish between multiple high intensity structures, such as white matter osteoporosis, stroke, and certain artifacts, because these lesions have the same intensity and can be produced spatially simultaneously. Clinicians distinguish between spatial features such as bilateral symmetry of white matter osteoporosis.
Shape-based methods typically model the shape of a structure by explicit or implicit characterization. In the present application, by capturing the variability of the spatial distribution of white matter osteoporosis using a shape model, it develops in a uniform pattern around the ventricle. In contrast, stroke can occur almost at any random location in the brain and there is no apparent shape or positional profile.
Fig. 1 is a schematic flow chart of a method for segmenting cerebrovascular lesions, which comprises the following steps:
step S110, MRI image information is acquired.
The lesion region is typically included in the MRI image information. I.e. the diseased area of the cerebral blood vessel.
If it is a training phase, spatial distribution, shape and appearance including healthy tissue and cerebrovascular lesions are required.
If the test phase is adopted, only MRI images of the lesion area are input.
Step S120, obtaining a segmentation result in the MRI image information through a pre-trained image segmentation model, wherein the image segmentation model adopts a maximum posterior probability model to carry out MAP estimation on a posterior distribution model for describing tissue types and space prior distribution of white matter loose tissues.
Segmentation results in the MRI image information may be obtained by the pre-trained image segmentation model. The pre-trained image segmentation model is based on a probability estimation model, and a corresponding segmentation result can be obtained according to the estimation result.
The image segmentation model uses a maximum posterior probability model to estimate the maximum posterior probability of a posterior distribution model describing tissue class, the spatial prior distribution of the white matter porous tissue, for example.
And step S130, obtaining a segmentation result for separating the white matter loose lesions from the acute and chronic stroke lesions in the cerebrovascular lesions according to the segmentation result.
According to the segmentation result, a segmentation result for separating white matter loose lesions from acute and chronic stroke lesions in the cerebrovascular lesions can be obtained. By segmenting individual cerebrovascular lesions in the MRI image information, expert knowledge of the disease is captured based on a probabilistic model.
Further, the above method automatically segments tissues indistinguishable only by intensity by modeling the spatial distribution of the white matter loosening lesions and the intensity of the white matter loosening and stroke lesions. The white matter loose lesions and the acute and chronic apoplexy lesions can be accurately segmented by adopting a apoplexy model with combination strength and space environment.
As a preference in this embodiment, the image segmentation model comprises intensity features, shape features and spatial environment features, wherein the intensity features comprise intensity distribution of white matter loose lesions and intensity distribution of strokes, the spatial environment features comprise spatial distribution of white matter loose lesions, and the segmentation result further comprises separating white matter loose lesions from acute and chronic stroke lesions in the cerebrovascular lesions based on a T2 FLAIR sequence.
In particular, the T2-fluid attenuation reversal sequence (Fluid Attenuated Inversion Recovery, FLAIR) can automatically separate/segment white matter loose lesions from acute and chronic stroke lesions in the cerebrovascular lesions.
As a preferable mode in the embodiment, the segmentation result in the MRI image information comprises the steps of describing the spatial distribution, the shape and the appearance of healthy tissues and cerebrovascular lesions by using a generation model, establishing a posterior distribution model for describing the tissue types, wherein the prior of the tissue types captures the knowledge of the spatial distribution and the lesion shape, and carrying out MAP estimation on the posterior distribution model for describing the tissue types by adopting a maximum posterior probability model to obtain the segmentation result in the MRI image information.
In specific implementation, establishing a posterior distribution model for describing the tissue type, and carrying out MAP estimation on the posterior distribution model for describing the tissue type by adopting a maximum posterior probability model to obtain a segmentation result in the MRI image information.
A posterior distribution model for describing tissue classes uses the generative model to describe the spatial distribution, shape and appearance of healthy tissue and cerebrovascular lesions.
Illustratively, Ω is set as a set of all spatial locations (voxels) in one image, and i= { Ix } x e Ω, let image I be generated from one spatially varying marker map c= { Cx } x e Ω, C representing the tissue class. For each voxel x, cx is a binary index vector that encodes three tissue tags—white matter loose tissue (L), stroke tissue (S) and healthy tissue (H).
Further, using the notation Cx (C) =1 to denote that the tissue class at voxel x is C, so C e { L, S, H }, whereas Cx (C) =0, given the label map C, the intensity observations, ix, are generated independently of the gaussian distribution:
Wherein the method comprises the steps of Representing normal distribution, μ is the mean, σ 2 is the variance, c= { L, S, H }, μ= { μ L,μS,μH},σ＝{σL,σS,σH } the prior of tissue class captures knowledge of spatial distribution and lesion shape.
It is assumed that the spatial extent of leukoporosis depends on the spatial distribution m= { M x}x∈Ω, where Mx is a priori the leukoporosis voxel x, M will be parameterized by the parameter α as described in the present application. If voxel x is not assigned as white matter spongiform tissue, it is assigned as stroke tissue with a spatially varying probability βx, and healthy tissue with a probability of (1- βx).
For spatial continuity, a Markov random field (Markov Random Field, MRF) is used as a spatial prior:
wherein pi x＝[Mx(α),(1-Mx(α))βx,(1-Mx(α))(1-βx)]T (3)
Pi x is a vector of prior probabilities for the three tissue classes described above,Is a set of voxel locations adjacent to x and 3*3 matrix a is selected to encourage adjacent voxels to share the same tissue label.
In practice, MRF has a greater interaction with stroke and other tissues than white matter loosening bordering healthy tissues, since stroke tissues are generally found to be more spatially continuous, whereas white matter loosening tissues are more diffuse. Using equations (1), (2) and (3), a posterior distribution of tissue classes is formed.
As a preferable mode in the embodiment, the segmentation result in the MRI image information comprises constructing a probability map by adopting a training set of binary segmentation mapping of artificially segmented white matter loose tissue lesions by PCA, and modeling the spatial range of white matter loose tissue lesions to obtain the spatial prior distribution of white matter loose tissue.
In the specific implementation, PCA is adopted to construct a probability map for the training set of binary segmentation mapping of white matter loose tissue lesions after manual segmentation, so that the modeling of the spatial range of white matter loose tissue lesions is realized, and the spatial priori distribution of white matter loose tissue is obtained.
For the spatial prior distribution of white matter loose tissue, firstly, modeling the spatial range of white matter loose lesions:
Illustratively, a probability map is constructed from a training set of binary segmentation maps of artificial white matter loose lesions by using principal component analysis (PRINCIPAL COMPONENT ANALYSIS, PCA).
Is provided withFor the mean value of the map,Is an important component corresponding to the K largest feature root, and α k should be the weight (or load):
Where Σ is the diagonal covariance matrix containing the K largest feature root. Given α, the spatial prior m= { M x}x∈Ω confirmation is defined as:
As a preference in this embodiment, the MAP estimation is based on the EM algorithm and models the intensity average estimation as a spatial variation while filtering with a low pass filter.
To obtain the segmentation MAP, MAP inference and finding is performed:
Since accurate computation becomes infeasible when the MRF weight matrix A is non-zero, an EM (estimation-mapping) algorithm is used to estimate the solution of MAP.
Illustratively, the posterior distribution P (C|I; μ, σ, α, β) is modeled using a complete factor distribution
W x is the probability vector for three tissue classes at voxel x. Since the a priori of the load P (α) of PCA is unconjugated with the likelihood P (c|α), the corresponding E-step computation is approximated using regularized projection:
U= [ M 1, ], mk ], the resulting value in using shear forcing M (α) is between 0 and 1. In step M, the parameters of the model are updated.
The update is intuitive. The mean and variance estimates for the classes are calculated as weighted averages:
Due to the presence of pathological changes of greater intensity and serious artefacts, the non-uniformity of the image cannot be corrected by the preprocessing step. To account for image non-uniformities of healthy tissue, the intensity average estimate is modeled as a spatial variation and a low pass filter G H is introduced to enhance spatial smoothness, similar to the original EM segmentation formula. The method specifically comprises the following steps:
μH←GH*(wx(H)·I) (10)
Where, represents a spatial convolution. The previous healthy tissue β x is a small fraction of the current stroke and healthy tissue probability estimates:
finally, after weighting adjustment with adjacent voxels, the variational posterior parameter w is obtained x
Wherein pi x (c) is defined in (3). And iteratively updating until the parameter estimation converges.
As a preferred embodiment, the image segmentation model uses a manually labeled white matter loose tissue lesion result as a training image.
As shown in fig. 2, a cerebrovascular disease segmentation apparatus 200 according to an embodiment of the present application includes:
an acquisition module 210 for acquiring MRI image information;
The processing module 220 is configured to obtain a segmentation result in the MRI image information through a pre-trained image segmentation model, where the image segmentation model uses a maximum posterior probability model to perform MAP estimation on a posterior distribution model for describing a tissue class and a spatial prior distribution of white matter loose tissue;
The segmentation module 230 is configured to obtain a segmentation result for separating the white matter loose lesion from the acute and chronic stroke lesions in the cerebrovascular lesions according to the segmentation result.
The acquisition module 210 of the embodiment of the present application generally includes a lesion region in the MRI image information. I.e. the diseased area of the cerebral blood vessel.
If it is a training phase, spatial distribution, shape and appearance including healthy tissue and cerebrovascular lesions are required.
If the test phase is adopted, only MRI images of the lesion area are input.
The processing module 220 in the embodiment of the present application may obtain the segmentation result in the MRI image information through the pre-trained image segmentation model. The pre-trained image segmentation model is based on a probability estimation model, and a corresponding segmentation result can be obtained according to the estimation result.
The image segmentation model uses a maximum posterior probability model to estimate the maximum posterior probability of a posterior distribution model describing tissue class, the spatial prior distribution of the white matter porous tissue, for example.
According to the segmentation result, the segmentation module 230 of the embodiment of the present application may obtain a segmentation result for separating the white matter loose lesion from the acute and chronic stroke lesions in the cerebrovascular lesion. By segmenting individual cerebrovascular lesions in the MRI image information, expert knowledge of the disease is captured based on a probabilistic model.
Further, the above method automatically segments tissues indistinguishable only by intensity by modeling the spatial distribution of the white matter loosening lesions and the intensity of the white matter loosening and stroke lesions. The white matter loose lesions and the acute and chronic apoplexy lesions can be accurately segmented by adopting a apoplexy model with combination strength and space environment.
As shown in fig. 3, the implementation principle of the method for segmenting cerebrovascular diseases according to the embodiment of the application is shown in the schematic diagram, which specifically includes:
S1, inputting an MRI image.
S2, describing a posterior distribution model of the tissue type.
S3, spatial priori distribution of white matter loose tissues.
S4, MAP inference.
S5, testing.
S6, outputting.
In practice, results of at least 100 manually described test images of white matter loosening and 6 additional test papers were used, each including a white matter loosening lesion manually delineated by a plurality of experts. The segmentation algorithm described above is only run in white matter in embodiments of the application, and it is expected that most white matter osteoporosis and stroke lesions will be seen there.
In embodiments of the application, the scan comprises a T2-FLAIR scan (1 x 1mm in plane, 5-7mm slice thickness, sometimes using PROPELLER sequences as the patient moves). TR and TE factor images were acquired, T1 images obtained for each subject, and parameters registered to a atlas template using ANTs.
Parameters PCA shape model ({ M k }, Σ) was trained on the binary map of artificial white matter loose lesion segmentation in 42 training scans. The fixed parameters λ and a are manually selected to optimize the results in a single test example. In particular, λ=250, a (c, c) =100 is used for c e { L, S, H }, a (L, H) =97, a (S, L) =1 and a (S, H) =20. And a simple threshold classifier learned from the training subjects is used to initialize the posterior estimation.
An embodiment of the application also provides a storage medium having a computer program stored therein, wherein the computer program is arranged to perform the steps of any of the method embodiments described above when run.
Alternatively, in the present embodiment, the above-described storage medium may be configured to store a computer program for performing the steps of:
S1, acquiring MRI image information;
S2, obtaining a segmentation result in the MRI image information through a pre-trained image segmentation model, wherein the image segmentation model adopts a maximum posterior probability model to carry out MAP estimation on a posterior distribution model for describing tissue types and space prior distribution of white matter loose tissues;
S3, according to the segmentation result, obtaining a segmentation result for separating white matter loose lesions from acute and chronic stroke lesions in the cerebrovascular lesions.
Alternatively, in the present embodiment, the storage medium may include, but is not limited to, a USB flash disk, a Read-Only Memory (ROM), a random access Memory (Random Access Memory, RAM), a removable hard disk, a magnetic disk, or an optical disk, etc. various media in which a computer program may be stored.
An embodiment of the application also provides an electronic device comprising a memory having stored therein a computer program and a processor arranged to run the computer program to perform the steps of any of the method embodiments described above.
Optionally, the electronic apparatus may further include a transmission device and an input/output device, where the transmission device is connected to the processor, and the input/output device is connected to the processor.
Alternatively, in the present embodiment, the above-described processor may be configured to execute the following steps by a computer program:
S1, acquiring MRI image information;
S2, obtaining a segmentation result in the MRI image information through a pre-trained image segmentation model, wherein the image segmentation model adopts a maximum posterior probability model to carry out MAP estimation on a posterior distribution model for describing tissue types and space prior distribution of white matter loose tissues;
S3, according to the segmentation result, obtaining a segmentation result for separating white matter loose lesions from acute and chronic stroke lesions in the cerebrovascular lesions.
Alternatively, specific examples in this embodiment may refer to examples described in the foregoing embodiments and optional implementations, and this embodiment is not described herein.
The above description is only of the preferred embodiments of the present application and is not intended to limit the present application, but various modifications and variations can be made to the present application by those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the protection scope of the present application.

Claims (9)

1. A method of segmenting a cerebrovascular disorder, the method comprising:
Acquiring MRI image information;
Obtaining a segmentation result in the MRI image information through a pre-trained image segmentation model, wherein the image segmentation model adopts a maximum posterior probability model to carry out MAP estimation on a posterior distribution model for describing tissue types and the spatial prior distribution of white matter loose tissues;
Obtaining a segmentation result of separating white matter loose lesions from acute and chronic stroke lesions in the cerebrovascular lesions according to the segmentation result;
The image segmentation model comprises an intensity feature, a shape feature and a spatial environment feature, wherein the intensity feature comprises an intensity distribution of white matter loose lesions and an intensity distribution of strokes, the spatial environment feature comprises a spatial distribution of white matter loose lesions, the segmentation result further comprises:
Based on the T2 FLAIR sequence, white matter loose lesions were separated from acute and chronic stroke lesions in the cerebrovascular lesions.
2. The method of claim 1, wherein the segmentation result in the MRI image information comprises:
Using the generative model to describe the spatial distribution, shape and appearance of healthy tissue and cerebrovascular lesions, building a posterior distribution model describing tissue classes, wherein the prior of tissue classes captures knowledge of spatial distribution and lesion shape;
And carrying out MAP estimation on the posterior distribution model for describing the tissue type by adopting a maximum posterior probability model to obtain a segmentation result in the MRI image information.
3. The method of claim 1, wherein the segmentation result in the MRI image information comprises:
and constructing a probability map by adopting PCA to a training set of binary segmentation mapping of artificially segmented white matter loose tissue lesions, and modeling the spatial range of white matter loose tissue lesions to obtain the spatial prior distribution of white matter loose tissue.
4. The method of claim 1, wherein the MAP estimate is based on an EM algorithm and modeling the intensity average estimate as a spatial variation while filtering with a low pass filter.
5. The method of claim 1, wherein the image segmentation model uses artificially labeled white matter loose tissue lesion results as training images.
6. A clinical image processing method characterized in that image registration is performed using the cerebrovascular lesion segmentation method according to any one of claims 1 to 5.
7. A cerebrovascular lesion segmentation device, the device comprising:
the acquisition module is used for acquiring MRI image information;
The processing module is used for obtaining a segmentation result in the MRI image information through a pre-trained image segmentation model, wherein the image segmentation model adopts a maximum posterior probability model to carry out MAP estimation on a posterior distribution model for describing tissue types and the spatial prior distribution of white matter loose tissues;
The segmentation module is used for obtaining a segmentation result for separating white matter loose lesions from acute and chronic apoplexy lesions in the cerebrovascular lesions according to the segmentation result;
The image segmentation model comprises an intensity feature, a shape feature and a spatial environment feature, wherein the intensity feature comprises an intensity distribution of white matter loose lesions and an intensity distribution of strokes, the spatial environment feature comprises a spatial distribution of white matter loose lesions, the segmentation result further comprises:
Based on the T2 FLAIR sequence, white matter loose lesions were separated from acute and chronic stroke lesions in the cerebrovascular lesions.
8. A computer readable storage medium, characterized in that the computer readable storage medium has stored therein a computer program, wherein the computer program is arranged to execute the method of any of the claims 1 to 5 and/or the method of claim 6 when run.
9. An electronic device comprising a memory and a processor, characterized in that the memory has stored therein a computer program, the processor being arranged to run the computer program to perform the method of any of the claims 1 to 5 and/or the method of claim 6.
CN202211256557.2A 2022-10-13 2022-10-13 Cerebrovascular lesion segmentation method, device, storage medium, and electronic device Active CN115496743B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202211256557.2A CN115496743B (en) 2022-10-13 2022-10-13 Cerebrovascular lesion segmentation method, device, storage medium, and electronic device

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202211256557.2A CN115496743B (en) 2022-10-13 2022-10-13 Cerebrovascular lesion segmentation method, device, storage medium, and electronic device

Publications (2)

Publication Number Publication Date
CN115496743A CN115496743A (en) 2022-12-20
CN115496743B true CN115496743B (en) 2025-09-23

Family

ID=84473869

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202211256557.2A Active CN115496743B (en) 2022-10-13 2022-10-13 Cerebrovascular lesion segmentation method, device, storage medium, and electronic device

Country Status (1)

Country Link
CN (1) CN115496743B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116071555B (en) * 2023-03-15 2023-06-23 同心智医科技(北京)有限公司 A method for establishing a WMHs segmentation model, a WMHs segmentation method and device

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1846614A (en) * 2005-04-04 2006-10-18 沈渊瑶 Method for constructing three-dimensional images of tissues
CN112991314A (en) * 2021-03-30 2021-06-18 昆明同心医联科技有限公司 Blood vessel segmentation method, device and storage medium

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR100291861B1 (en) * 1999-06-26 2001-06-01 박종원 A method of separating the white matter and the grey matter from brain image and calculating the volume of thereof
US7995825B2 (en) * 2001-04-05 2011-08-09 Mayo Foundation For Medical Education Histogram segmentation of FLAIR images
WO2011040473A1 (en) * 2009-09-29 2011-04-07 大日本印刷株式会社 Method, device and program for medical image processing
CN113223111A (en) * 2021-06-21 2021-08-06 暨南大学附属第一医院（广州华侨医院） Diffusion-weighted rapid imaging method for rapid liquid attenuation inversion recovery
CN113777546B (en) * 2021-09-13 2023-04-25 西安交通大学医学院第一附属医院 Magnetic resonance multi-parameter white matter high signal quantification method based on three-dimensional mapping

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1846614A (en) * 2005-04-04 2006-10-18 沈渊瑶 Method for constructing three-dimensional images of tissues
CN112991314A (en) * 2021-03-30 2021-06-18 昆明同心医联科技有限公司 Blood vessel segmentation method, device and storage medium

Also Published As

Publication number Publication date
CN115496743A (en) 2022-12-20

Similar Documents

Publication Publication Date Title
US11813047B2 (en) Automatic quantification of cardiac MRI for hypertrophic cardiomyopathy
Saba et al. Image enhancement and segmentation techniques for detection of knee joint diseases: A survey
Despotović et al. MRI segmentation of the human brain: challenges, methods, and applications
US10235750B2 (en) Segmentation of cardiac magnetic resonance (CMR) images using a memory persistence approach
US6985612B2 (en) Computer system and a method for segmentation of a digital image
Al-Tamimi et al. TUMOR BRAIN DETECTION THROUGH MR IMAGES: A REVIEW OF LITERATURE.
Wang et al. Principles and methods for automatic and semi-automatic tissue segmentation in MRI data
US20090129671A1 (en) Method and apparatus for image segmentation
Pohl et al. Anatomical guided segmentation with non-stationary tissue class distributions in an expectation-maximization framework
US20160140751A1 (en) Automated 3D Reconstruction of the Cardiac Chambers from MRI and Ultrasound
Balan et al. Smart histogram analysis applied to the skull-stripping problem in T1-weighted MRI
DE102007046582A1 (en) System and method for segmenting chambers of a heart in a three-dimensional image
Bandhyopadhyay et al. Segmentation of brain MRI image–a review
Meera et al. A Review On Automatic Detection of Brain Tumor Using Computer Aided Diagnosis System Through MRI.
CN115496743B (en) Cerebrovascular lesion segmentation method, device, storage medium, and electronic device
US20050069183A1 (en) Semi-automated measurement of anatomical structures using statistical and morphological priors
de Luis-Garcia et al. Gaussian mixtures on tensor fields for segmentation: Applications to medical imaging
Van Leemput et al. Model-based segmentation of hippocampal subfields in ultra-high resolution in vivo MRI
JP2008513164A (en) Image segmentation using isometric tree
Beejesh et al. Brain MR kurtosis imaging study: contrasting gray and white matter
Xu¹ et al. Segmentation of brain MRI
Chen et al. Segmentation of ventricles in brain CT images using gaussian mixture model method
EP1080449B1 (en) Autosegmentation / autocontouring system
Grande-Barreto et al. Segmentation of MRI brain scans using spatial constraints and 3D features
Aghdam et al. Challenges in brain magnetic resonance image segmentation

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
GR01 Patent grant
GR01 Patent grant