CN107154047A - Multi-mode brain tumor image blend dividing method and device - Google Patents
Multi-mode brain tumor image blend dividing method and device Download PDFInfo
- Publication number
- CN107154047A CN107154047A CN201710270990.4A CN201710270990A CN107154047A CN 107154047 A CN107154047 A CN 107154047A CN 201710270990 A CN201710270990 A CN 201710270990A CN 107154047 A CN107154047 A CN 107154047A
- Authority
- CN
- China
- Prior art keywords
- mrow
- msub
- image
- segmentation
- msup
- 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
- 208000003174 Brain Neoplasms Diseases 0.000 title claims abstract description 47
- 238000000034 method Methods 0.000 title claims abstract description 20
- 239000000203 mixture Substances 0.000 title abstract 3
- 230000011218 segmentation Effects 0.000 claims abstract description 54
- 206010028980 Neoplasm Diseases 0.000 claims abstract description 13
- 238000001914 filtration Methods 0.000 claims abstract description 11
- 238000012545 processing Methods 0.000 claims abstract description 9
- JXSJBGJIGXNWCI-UHFFFAOYSA-N diethyl 2-[(dimethoxyphosphorothioyl)thio]succinate Chemical compound CCOC(=O)CC(SP(=S)(OC)OC)C(=O)OCC JXSJBGJIGXNWCI-UHFFFAOYSA-N 0.000 claims abstract description 7
- 230000004927 fusion Effects 0.000 claims abstract description 7
- 230000033001 locomotion Effects 0.000 claims description 9
- 230000008602 contraction Effects 0.000 claims description 3
- 230000003247 decreasing effect Effects 0.000 claims description 3
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 3
- 238000002156 mixing Methods 0.000 claims description 2
- 238000007781 pre-processing Methods 0.000 claims description 2
- 206010048962 Brain oedema Diseases 0.000 claims 1
- 208000006752 brain edema Diseases 0.000 claims 1
- 239000003814 drug Substances 0.000 abstract 1
- 238000000638 solvent extraction Methods 0.000 abstract 1
- 238000002595 magnetic resonance imaging Methods 0.000 description 11
- 230000007547 defect Effects 0.000 description 4
- 238000003709 image segmentation Methods 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000012360 testing method Methods 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000009825 accumulation Methods 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000006931 brain damage Effects 0.000 description 1
- 231100000874 brain damage Toxicity 0.000 description 1
- 208000029028 brain injury Diseases 0.000 description 1
- 230000010261 cell growth Effects 0.000 description 1
- 230000000295 complement effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000002059 diagnostic imaging Methods 0.000 description 1
- 238000013399 early diagnosis Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 239000002245 particle Substances 0.000 description 1
Classifications
-
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- 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
- G06T2207/10088—Magnetic resonance imaging [MRI]
-
- 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
- G06T2207/30016—Brain
-
- 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
- G06T2207/30096—Tumor; Lesion
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Magnetic Resonance Imaging Apparatus (AREA)
Abstract
The present invention relates to medicine equipment, medical image, to propose a kind of improved multi-mode brain tumor image blend partitioning algorithm, brain tumor region is extracted using FFCM, the border issue that tumor region is present is modified using mixed-level set algorithm.So that FFCM algorithms and level set algorithm can be more efficiently applied in MRI brain tumor images.The technical solution adopted by the present invention is, multi-mode brain tumor image blend dividing method, input Three models first include T1C, T2 and FLAIR MRI image, and processing is filtered to image using medium filtering and initial segmentation obtains pretreatment image, afterwards using linear fusion；FFCM cluster segmentations are carried out to fused images again, the larger region of wherein gray value is automatically extracted, obtained tumour less divided region carries out mixed-level collection segmentation.Present invention is mainly applied to the acquisition of medical image and processing.
Description
Technical Field
The invention relates to a medical instrument, which is an important aspect in the field of medical imaging. It has important effect in the fields of brain tumor cutting, brain tumor classification, brain tumor identification and the like. And more particularly, to an improved multi-mode brain tumor image blending segmentation method and apparatus.
Background
In recent years, the incidence of brain tumors has been increasing, accounting for about 5% of all tumors and 70% of children's tumors. In 2015, about 23000 new cases of brain tumors were diagnosed in the united states alone. Uncontrolled and unlimited growth of cells leads to the development of brain tumors. Without early diagnosis and treatment of brain tumors, permanent brain damage, or even death, can result. Magnetic Resonance Imaging (MRI) can be used for detecting abnormal changes of body tissues, and is a necessary means for determining brain tumor treatment schemes, and in all treatment methods, any information about the position and size of a tumor is very important, but because brain tumors are complex in shape, random in size and position, large in type difference and the like, no segmentation algorithm can meet clinical requirements at present, real-time performance cannot meet requirements, results of different experts manually segmenting brain tumor images are greatly different, and labor cost is high. Therefore, it is very important to study an accurate full-automatic brain tumor segmentation algorithm.
The brain tumor image segmentation method comprises manual segmentation, semi-automatic segmentation and full-automatic segmentation, and the specific segmentation algorithm comprises a threshold algorithm, a clustering algorithm, a deformation model algorithm and the like. The threshold algorithm is firstly used for image segmentation, and aiming at the problem of brain tumor images, the OTSU algorithm is an automatic adaptive threshold algorithm and can effectively avoid errors caused by fixed thresholds; a local Fuzzy Threshold (FTH) algorithm for multi-region image segmentation also has a certain effect on a complex image such as a brain tumor, and the problem of brain tumor segmentation cannot be effectively solved by threshold class algorithm segmentation due to the complexity of the brain tumor image and insufficient consideration of the threshold algorithm on pixel spatial information.
Fuzzy clustering is a class of algorithms suitable for brain tumor image segmentation, particularly Fuzzy C-mean (FCM) algorithms, and has the advantage of simple method implementation, but because medical image information is complex and edges are not clear, seed point selection has a large influence on clustering results, and the FCM algorithm method is difficult to utilize spatial information of images and is complex in calculation. Therefore, the problem that the Fast FCM (Fast FCM, FFCM) algorithm improves the calculation speed is provided; aiming at the problem of insufficient Spatial information, a Spatial FCM (SFCM) algorithm is used for segmenting images, the correlation among the Spatial information is effectively utilized, but the calculation speed of the SFCM algorithm cannot meet the real-time performance required by medical images; the Level Set algorithm can effectively process various contour problems, and the Fuzzy Clustering with Level Set Methods (FCLSM) algorithm is combined to effectively solve the problem of Level Set edges, but the FCLSM algorithm has the problems of instantaneity and easiness in falling into local optimization.
The level set algorithm belongs to a class of deformation model algorithms, and is also widely applied to brain tumor segmentation based on the level set segmentation algorithm, but because the gray scale of brain tumor tissues is uneven, and no obvious boundary exists between the brain tumor tissues, the problem of edge leakage is easy to occur by adopting the algorithm. A Distance Regularized Level Set Evolution (DRLSE) is an effective algorithm, and the Distance regularization effect in the algorithm eliminates the need for re-initialization, thereby avoiding local errors caused by the Distance regularized level Set Evolution; other methods are also mixed level set algorithms. The method uses object boundary and region information to achieve robust and accurate segmentation. The boundary information may help to detect the precise position of the target object, and the region information may prevent the boundary leakage, but the level set algorithm may not solve the problems of being easily trapped in local optima and strongly depending on the initial value.
Disclosure of Invention
In order to overcome the defects of the prior art, the invention aims to provide an improved multi-mode brain tumor image mixed segmentation algorithm, which adopts FFCM to extract a brain tumor region and uses a mixed level set algorithm to correct the boundary problem of the tumor region. Thereby enabling the FFCM algorithm and the level set algorithm to be more effectively applied to the MRI brain tumor image. The invention adopts the technical scheme that a multi-mode brain tumor image mixed segmentation method comprises the steps of firstly inputting three modes of MRI images including T1C, T2 and FLAIR, carrying out filtering processing and initial segmentation on the images by adopting median filtering to obtain preprocessed images, and then adopting linear fusion; and performing FFCM clustering segmentation on the fused image, automatically extracting a region with a larger gray value, and performing mixed level set segmentation on the obtained under-segmented region of the tumor.
The FFCM clustering segmentation method specifically comprises the steps of dividing data into C classes by a fuzzy C-means theory, and setting { h } for an M × N imagei,i＝1,2,…,n}，n＝M×N，hiIs a set of pixel intensity values in an image histogram, where M and N are the length and width of the image, { vjJ is 1,2, …, c is a set of cluster centers,and muj(hi) Is hiMembership functions belonging to class j, | | | · | |, represents a 2 norm, b is a constant greater than 1, then:
if the iteration formulas (3) and (4) meet the iteration termination condition, T is more than T orStopping, wherein T represents the iteration number, is a stopping condition, T represents the maximum iteration number, after the algorithm is finished, classifying the pixels according to the maximum membership degree, and if mu is in the maximum iteration numberj(hi)＞μj(hk) Then h will beiA j-th region, k being 1, 2.., c; i ≠ k.
The concrete step of the mixed level set segmentation is that, a zero set of an embedding function phi is used for representing an active contour C ═ { X | phi (X) ═ 0}, points inside/outside the contour have positive/negative phi values, and the proposed function definition needing minimization is as follows:
in the formula (5), I is an image to be segmented,is a boundary feature map associated with the image gradient,is a gradient operator, H (φ) is a Hervesseld function, Ω is an image domain, α and β are predefined weights to balance the two terms, μ is a predefined parameter indicating a lower bound of the gray level of the target objectCounting;
wherein,is a normal vector pointing outside the curve, so that the dominant curve of the active contour evolves a partial differential equation of
In the formulaAnd curvature<·,·>Is an inner product; since only the geometric changes of the curve are of interest in the segmentation, it can be noted from equation (6) that all points on the curve move in the normal direction, the first term of equation (6) is a propagation term describing the expansion motion of the curved part inside the target object and the contraction motion of the outer part, the second term is an advection term describing the curve movement in the vector field caused by the gradient of g to attract the curve to the boundary of the target object, the third term describes the curvature flow weighted by the gradient feature map g, functioning to smooth the curve of the part whose boundary supports weakness;
in the horizontal concentration, the concentration of the water is in the horizontal direction,anddescribing the same curve variation if phi is a signed distance function, i.e.The derivative of the level set embedding function with time is
g is a decreasing function of the number of bits,wherein c is the control slope.
A multi-mode brain tumor image hybrid segmentation device is provided with a computer and is used for processing MRI images of T1C, T2 and FLAIR, and the computer comprises the following modules: performing filtering processing and an initial segmentation module on the image by adopting median filtering to obtain a preprocessed image; then preprocessing the image input linear fusion module; inputting the fused image into an FFCM clustering segmentation module, and automatically extracting a region with a larger gray value to obtain a tumor under-segmentation region; and then inputting the data to be processed by a mixed level set segmentation module to obtain a final result.
The invention has the characteristics and beneficial effects that:
compared with some classical methods, the invention mainly has the following advantages that the MRI image with the brain tumor is segmented by the improved multi-mode brain tumor image mixed segmentation algorithm:
1) the novelty is as follows: the FFCM algorithm and the mixed level set algorithm are used for segmenting the MRI image with the brain tumor for the first time, and the aim of rapidly segmenting the brain tumor image is fulfilled according to the characteristics of the MRI brain tumor image and the advantages of the FFCM algorithm and the mixed level set.
2) Effectiveness: the FFCM can be used for quickly and effectively obtaining the under-segmented area, and the convergence boundary can be accelerated when the under-segmented area is input into the mixed level set, so that the defects of the algorithm are effectively overcome, and meanwhile, the accuracy is improved.
3) The practicability is as follows: because the existing segmentation algorithm is difficult to meet the requirements of practicability and real-time performance, the invention combines a reasonable part between a mixed level set algorithm and an FFCM algorithm, thereby overcoming the defects of some algorithms and increasing the practicability of the algorithms to a certain extent. And further discussion is made for the technology of automatically segmenting the brain tumor.
Description of the drawings:
FIG. 1 is a flow chart of the improved multi-mode brain tumor image hybrid segmentation algorithm of the present invention for segmenting MRI brain tumors.
FIG. 2 shows the similarity coefficient (Dice) of the algorithm of the present invention in 10 brain tumor images.
Detailed Description
Fast FCM theory based on histogram
The core idea of FFCM is that pixel intensity value seeks proper membership and cluster center to minimize variance and iterative error of cost function in cluster, the value of cost function is weighted accumulation sum of 2 norm measure from pixel to cluster center, FFCM cluster segmentation algorithm is to divide data into C class by fuzzy C mean value theory, for an M × N image, supposing { h } hi,i＝1,2,…,n}，n＝M×N，hiIs a collection of pixel intensity values in an image histogram. { vjJ is 1,2, …, c is a set of cluster centers, and μj(hi) Is hiMembership function to class j, so the objective function of FFCM is
And is
In the formula, | | | |, represents a 2 norm, b is a constant greater than 1, and the ambiguity of the clustering result is controlled. To calculate JfOf such that
From the formulae (1) and (2) it can be deduced
If the iteration formulas (3) and (4) meet the iteration termination condition, T is more than T orThen stop, where T represents the number of iterations, is a stop condition and T represents the maximum number of iterations. After the algorithm is finished, classifying the pixels according to the maximum membership degree, and if mu is not greater than the maximum membership degreej(hi)＞μj(hk) Then h will beiA j-th region, k being 1, 2.., c; i ≠ k.
2 hybrid model level set principle
Osher and Sethian propose a level set method, and represent a low-dimensional curve as a zero level set of a high-dimensional curved surface. At any time, if phi is known, the zero level set curve can be obtained, wherein phi (X) is a level set function. To deal with changes in the surface, this can be done by simply representing the surface as a zero level set in a higher one-dimensional space. Compared with particle models and parametric models, level set models have significant advantages, with concept and numerical implementation adapted to solve any size problem; the regions inside and outside the active contour can be easily determined.
Since MRI images of brain tumors are extremely complex, the present invention uses a level set based segmentation method to integrate boundary problems and regional information while remedying the boundary problems left behind by the FFCM algorithm. Before describing the model, several parameters need to be specified, and the zero set of the embedding function phi is used to represent the active contour C ═ { X | phi (X) ═ 0}, and points inside/outside the contour have positive/negative values of phi. Proposed function definition requiring minimization
In the formula (5), I is an image to be segmented,is a boundary feature map associated with the image gradient, where g is a decreasing function,where a is the control slope, H (φ) is the Herveseided function, Ω is the image domain, α and β are predefined weights to balance the two terms μ is a predefined parameter indicating the lower limit of the gray level of the target object.Is a gradient operator.
Wherein,is a normal vector pointing outside the curve. The dominant curve of the active contour evolves a partial differential equation of
In the formulaAnd curvature<·,·>Is the inner product. Since only the geometric changes of the curve are of interest in the segmentation, it can be noted from equation (6) that all points on the curve move in the normal direction. The first term of equation (6) is a propagation term describing the expansion motion of a curved portion inside the target object and the contraction motion of an outer portion, the second term is an advection term describing the curve movement in the vector field caused by the gradient of g to attract the curve to the boundary of the target object, and the third term describes the curvature flow weighted by the gradient feature map g, functioning to smooth the curve of a portion whose boundary supports weakness.
In the horizontal concentration, the concentration of the water is in the horizontal direction,anddescribing the same curve variation if phi is a signed distance function, i.e.The derivative of the level set embedding function with time is
TABLE 1
Table 1 shows the results of the algorithm of the present invention on 47 brain tumor images, where Jaccard coefficient, similarity coefficient (Dice) and recall are the most commonly used evaluation criteria.
Since the MRI brain tumor image is not of high quality and cannot be used for direct segmentation, the invention firstly adopts the mixed segmentation algorithm framework shown in fig. 1, and since the three modality images can provide partially irrelevant and complementary information for tumor segmentation, the invention firstly inputs MRI images of three modes including T1C, T2 and FLAIR. Because the image has certain noise, the median filtering is adopted to carry out filtering processing and initial segmentation on the image to obtain a preprocessed image, and then linear fusion is adopted; and (3) carrying out FFCM clustering algorithm segmentation on the fused image, automatically extracting a region with a larger gray value, carrying out mixed level set segmentation on the obtained under-segmented region of the tumor, and finally evaluating a segmentation result. The FFCM clustering algorithm is combined with the mixed model level set algorithm, so that on one hand, the speed of the level set algorithm is increased, and meanwhile, the defect that the mixed model level set algorithm depends on the initial value is overcome. In order to test a proper fusion ratio, the invention passes a plurality of ratio tests and compares, and finally obtains a more proper ratio of FLAIR: T2: T1C to 5:4: 1.
Claims (4)
1. A multi-mode brain tumor image mixed segmentation method is characterized in that MRI images with three modes including T1C, T2 and FLAIR are input, median filtering is adopted to carry out filtering processing and initial segmentation on the images to obtain preprocessed images, and then linear fusion is adopted; and performing FFCM clustering segmentation on the fused image, automatically extracting a region with a larger gray value, and performing mixed level set segmentation on the obtained under-segmented region of the tumor.
2. The multimodal brain swelling of claim 1A tumor image mixed segmentation method is characterized in that FFCM clustering segmentation specifically comprises the steps of dividing data into C classes by a fuzzy C-means theory, and setting { h } for an M × N imagei,i＝1,2,…,n}，n＝M×N，hiIs a set of pixel intensity values in an image histogram, where M and N are the length and width of the image, { vjJ is 1,2, …, c is a set of cluster centers, and μj(hi) Is hiMembership functions belonging to class j, | | | · | |, represents a 2 norm, b is a constant greater than 1, then:
<mrow> <msub> <mi>v</mi> <mi>j</mi> </msub> <mo>=</mo> <mfrac> <mrow> <munderover> <mo>&Sigma;</mo> <mrow> <mi>i</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>n</mi> </munderover> <msup> <mrow> <mo>&lsqb;</mo> <msub> <mi>&mu;</mi> <mi>j</mi> </msub> <mrow> <mo>(</mo> <msub> <mi>h</mi> <mi>i</mi> </msub> <mo>)</mo> </mrow> <mo>&rsqb;</mo> </mrow> <mi>b</mi> </msup> <msub> <mi>h</mi> <mi>i</mi> </msub> </mrow> <mrow> <munderover> <mo>&Sigma;</mo> <mrow> <mi>i</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>n</mi> </munderover> <msup> <mrow> <mo>&lsqb;</mo> <msub> <mi>&mu;</mi> <mi>j</mi> </msub> <mrow> <mo>(</mo> <msub> <mi>h</mi> <mi>i</mi> </msub> <mo>)</mo> </mrow> <mo>&rsqb;</mo> </mrow> <mi>b</mi> </msup> </mrow> </mfrac> <mo>,</mo> <mi>j</mi> <mo>=</mo> <mn>1</mn> <mo>,</mo> <mn>2</mn> <mo>,</mo> <mo>...</mo> <mo>,</mo> <mi>c</mi> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>3</mn> <mo>)</mo> </mrow> </mrow>
<mrow> <msub> <mi>&mu;</mi> <mi>j</mi> </msub> <mrow> <mo>(</mo> <msub> <mi>h</mi> <mi>i</mi> </msub> <mo>)</mo> </mrow> <mo>=</mo> <mfrac> <msup> <mrow> <mo>(</mo> <mn>1</mn> <mo>/</mo> <mo>|</mo> <mo>|</mo> <msub> <mi>h</mi> <mi>i</mi> </msub> <mo>-</mo> <msub> <mi>v</mi> <mi>j</mi> </msub> <mo>|</mo> <msup> <mo>|</mo> <mn>2</mn> </msup> <mo>)</mo> </mrow> <mrow> <mn>1</mn> <mo>/</mo> <mrow> <mo>(</mo> <mi>b</mi> <mo>-</mo> <mn>1</mn> <mo>)</mo> </mrow> </mrow> </msup> <mrow> <munderover> <mo>&Sigma;</mo> <mrow> <mi>k</mi> <mo>=</mo> <mn>1</mn> </mrow> <mi>c</mi> </munderover> <msup> <mrow> <mo>(</mo> <mn>1</mn> <mo>/</mo> <mo>|</mo> <mo>|</mo> <msub> <mi>h</mi> <mi>i</mi> </msub> <mo>-</mo> <msub> <mi>v</mi> <mi>k</mi> </msub> <mo>|</mo> <msup> <mo>|</mo> <mn>2</mn> </msup> <mo>)</mo> </mrow> <mrow> <mn>1</mn> <mo>/</mo> <mrow> <mo>(</mo> <mi>b</mi> <mo>-</mo> <mn>1</mn> <mo>)</mo> </mrow> </mrow> </msup> </mrow> </mfrac> <mo>,</mo> <mi>i</mi> <mo>=</mo> <mn>1</mn> <mo>,</mo> <mn>2</mn> <mo>,</mo> <mo>...</mo> <mo>,</mo> <mi>n</mi> <mo>;</mo> <mi>j</mi> <mo>=</mo> <mn>1</mn> <mo>,</mo> <mn>2</mn> <mo>,</mo> <mo>...</mo> <mo>,</mo> <mi>c</mi> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>4</mn> <mo>)</mo> </mrow> </mrow>
if the iteration formulas (3) and (4) meet the iteration termination condition, t>T orStopping, wherein T represents the iteration number, is a stopping condition, T represents the maximum iteration number, after the algorithm is finished, classifying the pixels according to the maximum membership degree, and if mu is in the maximum iteration numberj(hi)>μj(hk) Then h will beiA j-th region, k being 1, 2.., c; i ≠ k.
The concrete step of the mixed level set segmentation is that, a zero set of an embedding function phi is used for representing an active contour C ═ { X | phi (X) ═ 0}, points inside/outside the contour have positive/negative phi values, and the proposed function definition needing minimization is as follows:
<mrow> <mi>M</mi> <mi>i</mi> <mi>n</mi> <mi>&epsiv;</mi> <mrow> <mo>(</mo> <mi>&phi;</mi> <mo>)</mo> </mrow> <mo>=</mo> <mo>-</mo> <mi>&alpha;</mi> <munder> <mo>&Integral;</mo> <mi>&Omega;</mi> </munder> <mrow> <mo>(</mo> <mi>I</mi> <mo>-</mo> <mi>&mu;</mi> <mo>)</mo> </mrow> <mi>H</mi> <mrow> <mo>(</mo> <mi>&phi;</mi> <mo>)</mo> </mrow> <mi>d</mi> <mi>&Omega;</mi> <mo>+</mo> <mi>&beta;</mi> <munder> <mo>&Integral;</mo> <mi>&Omega;</mi> </munder> <mi>g</mi> <mo>|</mo> <mo>&dtri;</mo> <mi>H</mi> <mrow> <mo>(</mo> <mi>&phi;</mi> <mo>)</mo> </mrow> <mo>|</mo> <mi>d</mi> <mi>&Omega;</mi> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>5</mn> <mo>)</mo> </mrow> </mrow>
in the formula (5), I is an image to be segmented,is a boundary feature map associated with the image gradient,is a gradient operator, H (Φ) is the hervesseld function, Ω is the image domain, α and β are predefined weights to balance the two terms, μ is a predefined parameter indicating the lower limit of the gray level of the target object;
wherein,is a normal vector pointing outside the curve, so that the dominant curve of the active contour evolves a partial differential equation of
<mrow> <msub> <mi>C</mi> <mi>t</mi> </msub> <mo>=</mo> <mi>&alpha;</mi> <mrow> <mo>(</mo> <mi>I</mi> <mo>-</mo> <mi>&mu;</mi> <mo>)</mo> </mrow> <mover> <mi>N</mi> <mo>&RightArrow;</mo> </mover> <mo>-</mo> <mi>&beta;</mi> <mo><</mo> <mo>&dtri;</mo> <mi>g</mi> <mo>&CenterDot;</mo> <mover> <mi>N</mi> <mo>&RightArrow;</mo> </mover> <mo>></mo> <mo>+</mo> <mi>&beta;</mi> <mi>g</mi> <mi>&kappa;</mi> <mover> <mi>N</mi> <mo>&RightArrow;</mo> </mover> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>6</mn> <mo>)</mo> </mrow> </mrow>
In the formulaAnd curvature<·,·>Is an inner product; since only the geometric changes of the curve are of interest in the segmentation, it can be noted from equation (6) that all points on the curve move in the normal direction, the first term of equation (6) is a propagation term describing the expansion motion of the curved part inside the target object and the contraction motion of the outer part, the second term is an advection term describing the curve movement in the vector field caused by the gradient of g to attract the curve to the boundary of the target object, the third term describes the curvature flow weighted by the gradient feature map g, functioning to smooth the curve of the part whose boundary supports weakness;
in the horizontal concentration, the concentration of the water is in the horizontal direction,anddescribing the same curve variation if phi is a signed distance function, i.e.The derivative of the level set embedding function with time is
<mrow> <msub> <mi>&phi;</mi> <mi>t</mi> </msub> <mo>=</mo> <mi>&alpha;</mi> <mrow> <mo>(</mo> <mi>I</mi> <mo>-</mo> <mi>&mu;</mi> <mo>)</mo> </mrow> <mo>+</mo> <mi>&beta;</mi> <mi>d</mi> <mi>i</mi> <mi>v</mi> <mrow> <mo>(</mo> <mi>g</mi> <mo>&dtri;</mo> <mi>&phi;</mi> <mo>)</mo> </mrow> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mo>(</mo> <mn>7</mn> <mo>)</mo> <mo>.</mo> </mrow>1
3. The method of multi-modal brain tumor image blending segmentation of claim 1, wherein g is a decreasing function,wherein c is a control slopeAnd (4) rate.
4. A multi-mode brain tumor image hybrid segmentation device is characterized in that a computer is provided for processing MRI images of T1C, T2 and FLAIR, and the computer comprises the following modules: performing filtering processing and an initial segmentation module on the image by adopting median filtering to obtain a preprocessed image; then preprocessing the image input linear fusion module; inputting the fused image into an FFCM clustering segmentation module, and automatically extracting a region with a larger gray value to obtain a tumor under-segmentation region; and then inputting the data to be processed by a mixed level set segmentation module to obtain a final result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710270990.4A CN107154047A (en) | 2017-04-24 | 2017-04-24 | Multi-mode brain tumor image blend dividing method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710270990.4A CN107154047A (en) | 2017-04-24 | 2017-04-24 | Multi-mode brain tumor image blend dividing method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107154047A true CN107154047A (en) | 2017-09-12 |
Family
ID=59793903
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710270990.4A Pending CN107154047A (en) | 2017-04-24 | 2017-04-24 | Multi-mode brain tumor image blend dividing method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107154047A (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107845098A (en) * | 2017-11-14 | 2018-03-27 | 南京理工大学 | Liver cancer image full-automatic partition method based on random forest and fuzzy clustering |
| CN107909577A (en) * | 2017-10-18 | 2018-04-13 | 天津大学 | Fuzzy C-mean algorithm continuous type max-flow min-cut brain tumor image partition method |
| CN108416792A (en) * | 2018-01-16 | 2018-08-17 | 辽宁师范大学 | Medical computer tomoscan image dividing method based on movable contour model |
| CN108961214A (en) * | 2018-05-31 | 2018-12-07 | 天津大学 | Brain tumor MRI three-dimensional dividing method based on improved continuous type maximum-flow algorithm |
| CN109671054A (en) * | 2018-11-26 | 2019-04-23 | 西北工业大学 | The non-formaldehyde finishing method of multi-modal brain tumor MRI |
| CN109685767A (en) * | 2018-11-26 | 2019-04-26 | 西北工业大学 | A kind of bimodal brain tumor MRI dividing method based on Cluster-Fusion algorithm |
| CN110309827A (en) * | 2019-05-06 | 2019-10-08 | 上海海洋大学 | A kind of area of edema parted pattern based on OCT image |
| CN112634280A (en) * | 2020-12-08 | 2021-04-09 | 辽宁师范大学 | MRI image brain tumor segmentation method based on energy functional |
| CN112686916A (en) * | 2020-12-28 | 2021-04-20 | 淮阴工学院 | Curved surface reconstruction system based on heterogeneous multi-region CT scanning data processing |
| CN116363160A (en) * | 2023-05-30 | 2023-06-30 | 杭州脉流科技有限公司 | CT perfusion image brain tissue segmentation method and computer equipment based on level set |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106127794A (en) * | 2016-07-29 | 2016-11-16 | 天津大学 | Based on probability FCM algorithm MRI tumor image dividing method and system |
| CN106296699A (en) * | 2016-08-16 | 2017-01-04 | 电子科技大学 | Cerebral tumor dividing method based on deep neural network and multi-modal MRI image |
-
2017
- 2017-04-24 CN CN201710270990.4A patent/CN107154047A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106127794A (en) * | 2016-07-29 | 2016-11-16 | 天津大学 | Based on probability FCM algorithm MRI tumor image dividing method and system |
| CN106296699A (en) * | 2016-08-16 | 2017-01-04 | 电子科技大学 | Cerebral tumor dividing method based on deep neural network and multi-modal MRI image |
Non-Patent Citations (4)
| Title |
|---|
| SHAHEEN AHMED,ET.AL: "Efficacy of Texture, Shape, and Intensity Feature Fusion for Posterior-Fossa Tumor Segmentation in MRI", 《IEEE TRANSACTIONS ON INFORMATION TECHNOLOGY IN BIOMEDICINE》 * |
| YAN ZHANG，ET.AL: "Medical Image Segmentation Using New Hybrid Level-Set Method", 《FIFTH INTERNATIONAL CONFERENCE BIOMEDICAL VISUALIZATION: INFORMATION VISUALIZATION IN MEDICAL AND BIOMEDICAL INFORMATICS》 * |
| 张腾达: "基于模糊水平集的脑肿瘤MR图像分割方法", 《现代电子技术》 * |
| 王黎明: "基于模糊C均值算法的医学图像分割研究", 《硕士学位论文全文数据库》 * |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107909577A (en) * | 2017-10-18 | 2018-04-13 | 天津大学 | Fuzzy C-mean algorithm continuous type max-flow min-cut brain tumor image partition method |
| CN107845098A (en) * | 2017-11-14 | 2018-03-27 | 南京理工大学 | Liver cancer image full-automatic partition method based on random forest and fuzzy clustering |
| CN108416792B (en) * | 2018-01-16 | 2021-07-06 | 辽宁师范大学 | Medical computed tomography image segmentation method based on active contour model |
| CN108416792A (en) * | 2018-01-16 | 2018-08-17 | 辽宁师范大学 | Medical computer tomoscan image dividing method based on movable contour model |
| CN108961214A (en) * | 2018-05-31 | 2018-12-07 | 天津大学 | Brain tumor MRI three-dimensional dividing method based on improved continuous type maximum-flow algorithm |
| CN109685767A (en) * | 2018-11-26 | 2019-04-26 | 西北工业大学 | A kind of bimodal brain tumor MRI dividing method based on Cluster-Fusion algorithm |
| CN109671054A (en) * | 2018-11-26 | 2019-04-23 | 西北工业大学 | The non-formaldehyde finishing method of multi-modal brain tumor MRI |
| CN110309827A (en) * | 2019-05-06 | 2019-10-08 | 上海海洋大学 | A kind of area of edema parted pattern based on OCT image |
| CN110309827B (en) * | 2019-05-06 | 2022-10-14 | 上海海洋大学 | Edema region segmentation model based on OCT image |
| CN112634280A (en) * | 2020-12-08 | 2021-04-09 | 辽宁师范大学 | MRI image brain tumor segmentation method based on energy functional |
| CN112634280B (en) * | 2020-12-08 | 2023-06-16 | 辽宁师范大学 | MRI image brain tumor segmentation method based on energy functional |
| CN112686916A (en) * | 2020-12-28 | 2021-04-20 | 淮阴工学院 | Curved surface reconstruction system based on heterogeneous multi-region CT scanning data processing |
| CN112686916B (en) * | 2020-12-28 | 2024-04-05 | 淮阴工学院 | Curved surface reconstruction system based on heterogeneous multi-region CT scanning data processing |
| CN116363160A (en) * | 2023-05-30 | 2023-06-30 | 杭州脉流科技有限公司 | CT perfusion image brain tissue segmentation method and computer equipment based on level set |
| CN116363160B (en) * | 2023-05-30 | 2023-08-29 | 杭州脉流科技有限公司 | CT perfusion image brain tissue segmentation method and computer equipment based on level set |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107154047A (en) | Multi-mode brain tumor image blend dividing method and device | |
| US7876938B2 (en) | System and method for whole body landmark detection, segmentation and change quantification in digital images | |
| Lemaître et al. | Computer-aided detection and diagnosis for prostate cancer based on mono and multi-parametric MRI: a review | |
| CN107644420B (en) | Blood vessel image segmentation method based on centerline extraction and nuclear magnetic resonance imaging system | |
| CN107403201A (en) | Tumour radiotherapy target area and jeopardize that organ is intelligent, automation delineation method | |
| Dhara et al. | Computer-aided detection and analysis of pulmonary nodule from CT images: a survey | |
| CN102324109A (en) | Method for three-dimensionally segmenting insubstantial pulmonary nodule based on fuzzy membership model | |
| McClure et al. | A novel NMF guided level-set for DWI prostate segmentation | |
| Campadelli et al. | A segmentation framework for abdominal organs from CT scans | |
| CN104616289A (en) | Removal method and system for bone tissue in 3D CT (Three Dimensional Computed Tomography) image | |
| Mughal et al. | Adaptive hysteresis thresholding segmentation technique for localizing the breast masses in the curve stitching domain | |
| Kaur | An automatic brain tumor extraction system using different segmentation methods | |
| Agarwal et al. | A combination of bias-field corrected fuzzy c-means and level set approach for brain MRI image segmentation | |
| Ramasamy et al. | Machine learning in cyber physical systems for healthcare: brain tumor classification from MRI using transfer learning framework | |
| Garg et al. | Spinal cord MRI segmentation techniques and algorithms: A survey | |
| Li et al. | Automated segmentation of prostate MR images using prior knowledge enhanced random walker | |
| Taghizadeh et al. | The fast and accurate approach to detection and segmentation of melanoma skin cancer using fine-tuned YOLOV3 and SegNet based on deep transfer learning | |
| Dhalia Sweetlin et al. | Patient-Specific Model Based Segmentation of Lung Computed Tomographic Images. | |
| CN108961214A (en) | Brain tumor MRI three-dimensional dividing method based on improved continuous type maximum-flow algorithm | |
| Ghorbanzadeh et al. | An Investigation into the Performance of Adaptive Neuro-Fuzzy Inference System for Brain Tumor Delineation Using ExpectationMaximization Cluster Method; a Feasibility Study | |
| Cao et al. | Segmentation of immunohistochemical image of lung neuroendocrine tumor based on double layer watershed | |
| Shao et al. | A segmentation method of airway from chest CT image based on VGG-Unet neural network | |
| Hussain et al. | Segmentation of brain MRI with statistical and 2D wavelet features by using neural networks | |
| Dai et al. | Lung segmentation with improved graph cuts on chest CT images | |
| Urschler et al. | Automatic intervertebral disc localization and segmentation in 3d mr images based on regression forests and active contours |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20170912 |