CN103606142B - A kind of elastic strain appraisal procedure based on ultrasonic elastic image and system - Google Patents

Abstract

This application discloses a kind of elastic strain appraisal procedure based on ultrasonic elastic image, including: rebuilding Ultrasonic elasticity frame according to ultrasonic elastic image and B-mode image, described Ultrasonic elasticity frame comprises elastic strain information；Described B-mode image is carried out the automatic segmentation of focal area；Described Ultrasonic elasticity frame is carried out soft or hard region automatically define；To the focal area after segmentation automatically and focus neighboring area, carry out stiffness characteristics extraction according to the described soft or hard region automatically defined；Based on the stiffness characteristics extracted, according to default standards of grading, focal area and focus neighboring area are carried out elastic strain assessment.Disclosed herein as well is a kind of elastic strain based on ultrasonic elastic image and assess system.The detailed description of the invention of the application carries out the Automatic image segmentation of focal area for B-mode image, it is to avoid the subjectivity of artificial segmentation, improves the accuracy that elastic characteristic extracts.

Description

A kind of elastic strain appraisal procedure based on ultrasonic elastic image and system

Technical field

The application relates to elastic strain assessment and analytical technology, particularly relates to the elastic strain appraisal procedure based on ultrasonic elastic image and system.

Background technology

Medicine clinical research shows, in human body, the coefficient of elasticity of different tissues or same tissue zones of different is different, elastography can be superimposed upon on two-dimentional audiovideo picture by coloud coding tissue elasticity difference and display, and the diagnosis (discriminating of such as Benign and malignant mammary tumor) for some disease provides new approach.

For the diagnosis of breast tumor, due to the difference of growth mechanism aspect, the difference of the molecular composition of Benign and malignant mammary tumor, tissue density and vascularity around causes the difference of the two elastic characteristic (or hardness).Ultrasonic Elasticity Imaging is to detect for the purpose of biological tissue's mechanical characteristic, it compensate for the conventional Ultrasound deficiency for elasticity measurement, the Flexible change of tissue is showed with the form of gray-scale map or pcolor, the measurement making elastic characteristic is extracted and is possibly realized, and is widely used in Benign and malignant mammary tumor discriminating.

Accurately estimate and on analysing elastic image, Breast Tumors is differentiated most important by tissue elasticity strain.Five point-scores and semiquantitative method are the method for the most frequently used assessment breast tumor elastic strain, but this method of both there is also subjectivity is big, accuracy of identification is low shortcoming.Main reason is that: 1) focal area profile need to by manually determining, it does not have realize segmentation automatically, thus subjectivity is big；2) in focal area soft or hard region define use fixed threshold, due to elastogram, different images defines the actual correspondence adaptability to changes of threshold value different, causes elasticity number quantization error.

Summary of the invention

The application to solve the technical problem that being for the deficiencies in the prior art, it is provided that a kind of elastic information that can improve extracts the elastic strain appraisal procedure based on ultrasonic elastic image of accuracy.

Another technical problem that the application to solve is to provide a kind of elastic strain based on ultrasonic elastic image based on said method and assesses system.

The application to solve the technical problem that and to solve by the following technical programs:

A kind of elastic strain appraisal procedure based on ultrasonic elastic image, including:

Rebuilding Ultrasonic elasticity frame according to ultrasonic elastic image and B-mode image, described Ultrasonic elasticity frame comprises elastic strain information；

Described B-mode image is carried out the automatic segmentation of focal area；

Described Ultrasonic elasticity frame is carried out soft or hard region automatically define；

To the focal area after segmentation automatically and focus neighboring area, carry out stiffness characteristics extraction according to the described soft or hard region automatically defined；

Based on the stiffness characteristics of described extraction, according to default standards of grading, focal area and focus neighboring area are carried out elastic strain assessment.

The described segmentation automatically that described B-mode image carries out focal area includes: described B-mode image adopts the automatic segmentation that the image partition method based on Mumford-Shah segmentation function and level set techniques realizes focal area.

Described soft or hard region that described Ultrasonic elasticity frame is carried out automatically is defined and is included: according to fuzzy clustering criterion, calculate in described elastic information image pixel space the fuzzy similarity matrix of each element and the square distance of cluster centre and, determine element soft or hard interval degree of membership, it is achieved automatically defining of soft or hard region.

Described to the focal area after segmentation automatically and focus neighboring area, carry out stiffness characteristics extraction according to the described soft or hard region automatically defined and include:

For the focal area after segmentation automatically, stiffness characteristics is represented by:Wherein T_hFor the number of pixels sum in territory, hard area, inside, focal area, T_aFor focal area interior pixels number sum；

For the focus neighboring area after segmentation automatically, stiffness characteristics is represented by:Wherein P_hFor the number of pixels sum in internal territory, hard area, focus neighboring area, P_aFor focus neighboring area interior pixels number sum.

The described stiffness characteristics based on described extraction, carries out elastic strain assessment according to default standards of grading to focal area and focus neighboring area and includes: if E_t≤ 20%, then it is evaluated as 1 point；If 20% < E_t≤ 50%, it is evaluated as 2 points；If 50% < E_t≤ 80%, it is evaluated as 3 points；If 80% < E_tAnd E_p≤ 50%, then it is evaluated as 4 points；If 80% < E_tAnd E_p> 50%, then be evaluated as 5 points.

A kind of elastic strain based on ultrasonic elastic image assesses system, including rebuilding module, segmentation module, soft or hard region deviding module, stiffness characteristics extraction module and evaluation module；

Described reconstruction module is for rebuilding Ultrasonic elasticity frame according to ultrasonic elastic image and B-mode image, and described Ultrasonic elasticity frame comprises elastic strain information；

Described segmentation module for carrying out the automatic segmentation of focal area to described B-mode image；

Described soft or hard region deviding module defines automatically for described Ultrasonic elasticity frame is carried out soft or hard region；

Described hardness extraction module is for the focal area after segmentation automatically and focus neighboring area, carrying out stiffness characteristics extraction according to the described soft or hard region automatically defined.

Focal area and focus neighboring area, for the stiffness characteristics based on described extraction, are carried out elastic strain assessment according to default standards of grading by described evaluation module.

Described segmentation module is additionally operable to the automatic segmentation that described B-mode image adopts the image partition method based on Mumford-Shah segmentation function and level set techniques realize focal area.

Described soft or hard region deviding module is additionally operable to according to fuzzy clustering criterion, calculate in described elastic information image pixel space the fuzzy similarity matrix of each element and the square distance of cluster centre and, determine element soft or hard interval degree of membership, it is achieved automatically defining of soft or hard region.

Described hardness extraction module is additionally operable to:

For the focal area after segmentation automatically, stiffness characteristics is expressed as:Wherein T_hFor the number of pixels sum in territory, hard area, inside, focal area, T_aFor focal area interior pixels number sum；

For the focus neighboring area after segmentation automatically, stiffness characteristics is represented by:Wherein P_hFor the number of pixels sum in internal territory, hard area, focus neighboring area, P_aFor focus neighboring area interior pixels number sum.

Described evaluation module is additionally operable to, if E_t≤ 20%, then it is evaluated as 1 point；If 20% < E_t≤ 50%, it is evaluated as 2 points；If 50% < E_t≤ 80%, it is evaluated as 3 points；If 80% < E_tAnd E_p≤ 50%, then it is evaluated as 4 points；If 80% < E_tAnd E_p> 50%, then be evaluated as 5 points.

Owing to have employed above technical scheme, make what the application possessed to have the beneficial effects that:

(1) in the detailed description of the invention of the application, carry out the Automatic image segmentation of focal area for B-mode image, it is to avoid the subjectivity of artificial segmentation, improve the accuracy that elastic characteristic extracts；

(2) in the detailed description of the invention of the application, rebuilding elastic information image, the extraction of all elastic strain features is all for this elastic information image, it is to avoid subsequent calculations is produced interference by elastic image, improves the accuracy of elastic strain feature extraction further；

(3) in the detailed description of the invention of the application, fuzzy classification technology is adopted to define soft or hard region, can for the feature of different elastic images, while improving elastic characteristic extraction accuracy, strengthen the suitability and the reliability of firmness zone pixel definition, it is to avoid the elasticity number quantization error that single threshold value is brought；

(4) in the detailed description of the invention of the application, considering the change of focal area and focus neighboring area elastic strain, elastic characteristic analysis is more fully simultaneously；

(5), in the detailed description of the invention of the application, by stiffness characteristics and predetermined standards of grading, elastic strain is estimated, can further improve the accuracy of assessment.

Accompanying drawing explanation

Fig. 1 is the flow chart according to one embodiment of the application method；

Fig. 2 is the structural representation according to one embodiment of the application system.

Detailed description of the invention

The present invention is described in further detail in conjunction with accompanying drawing below by detailed description of the invention.

Fig. 1 illustrates according to the application flow chart based on one embodiment of elastic strain appraisal procedure of ultrasonic elastic image, including:

Step 102: rebuilding Ultrasonic elasticity frame according to ultrasonic elastic image and B-mode image, this Ultrasonic elasticity frame comprises elastic strain information；

Step 104: B-mode image is carried out the automatic segmentation of focal area；

Step 106: Ultrasonic elasticity frame is carried out soft or hard region and automatically defines；

Step 108: to the focal area after segmentation automatically and focus neighboring area, the soft or hard region according to automatically defining carries out stiffness characteristics extraction.

Step 110: based on the stiffness characteristics extracted, according to default standards of grading, focal area and focus neighboring area are carried out elastic strain assessment.

A kind of embodiment, organization ultrasonic imaging technique applies a small strain to tissue by probe apparatus, collect measured body echo-signal fragment and it is analyzed, estimating deformation extent and the tissue elasticity coefficient magnitude of tissue, then with GTG or coloud coding imaging.In elastic image, characterize tissue elasticity coefficient magnitude colouring information be superimposed upon on its B-mode image show, to tissue elasticity information retrieval analyze before, first have to reconstruct elastic information image.If elastic mixed image is I_c, B-mode image is I_b, after reconstruction, image is I_r, according to elastogram principle, elastic information image can be set up by following formula:

I_r=I_c-I_b(1)

A kind of embodiment, step 104 farther includes: described B-mode image adopts the image partition method based on Mumford-Shah segmentation function and level set techniques realize the automatic segmentation of focal area.

Its cutting procedure is as follows:

For B-mode image, definition Ω is the coordinate set of its image I, if image I is divided into Cin (curvilinear inner) and two regions of Cout (curved exterior), c by curve C_inAnd c_outThe respectively average gray in the two region, definition energy function is as follows:

$F(c_{\mathrm{in}},c_{\mathrm{out}},\mathrm{\&phi;})=\mathrm{\&mu;}{\&Integral;}_{\mathrm{\&Omega;}}|\&dtri;H\left(\mathrm{\&phi;}\right)|\mathrm{dxdy}+v{\&Integral;}_{\mathrm{\&Omega;}}H\left(\mathrm{\&phi;}\right)\mathrm{dxdy}+{\mathrm{\&lambda;}}_1{\&Integral;}_{\mathrm{\&Omega;}}{|{\mathrm{\&mu;}}_0-c_{\mathrm{in}}|}^{2}H\left(\mathrm{\&phi;}\right)\mathrm{dxdy}+{\mathrm{\&lambda;}}_2{\&Integral;}_{\mathrm{\&Omega;}}{|{\mathrm{\&mu;}}_0-c_{\mathrm{out}}|}^2(1-H\left(\mathrm{\&phi;}\right))\mathrm{dxdy}$

This is (2) formula, and wherein, φ is level set function, represents Cin (curvilinear inner) and Cout (curved exterior), μ, v, λ₁, λ₂Being all constant coefficient, H (φ) is Heaviside function.When above formula energy function is minimum, the zero level collection of level set function φ is desired object boundary curve C.φ, c_inAnd c_outCan be obtained by following formula:

$c_{\mathrm{in}}\left(\mathrm{\&phi;}\right)=\frac{{\&Integral;}_{\mathrm{\&Omega;}}{u}_0(x,y)H\left(\mathrm{\&phi;}\right)\mathrm{dxdy}}{{\&Integral;}_{\mathrm{\&Omega;}}H\left(\mathrm{\&phi;}\right)\mathrm{dxdy}}---\left(3\right)$

$c_{\mathrm{out}}\left(\mathrm{\&phi;}\right)=\frac{{\&Integral;}_{\mathrm{\&Omega;}}{u}_0(x,y)(1-H\left(\mathrm{\&phi;}\right))\mathrm{dxdy}}{{\&Integral;}_{\mathrm{\&Omega;}}(1-H\left(\mathrm{\&phi;}\right))\mathrm{dxdy}}---\left(4\right)$

Function phi t in time develop, can derive about φ (t, x, Euler-Lagrange equation y) is as follows:

$\frac{\&PartialD;\mathrm{\&phi;}}{\&PartialD;t}={\mathrm{\&delta;}}_{\mathrm{\&epsiv;}}\left(\mathrm{\&phi;}\right)[\mathrm{\&mu;}\mathrm{div}\left(\frac{\&dtri;\mathrm{\&phi;}}{|\&dtri;\mathrm{\&phi;}|}\right)-v-{\mathrm{\&lambda;}}_1{(u_0-c_{\mathrm{in}})}^2+{\mathrm{\&lambda;}}_2{(u_0-c_{\mathrm{out}})}^2]=0,\mathrm{in}(0,\&infin;)\&times;\mathrm{\&Omega;}$

φ (0, x, y)=φ₀(x,y),inΩ

$\frac{{\mathrm{\&delta;}}_{\mathrm{\&epsiv;}}\left(\mathrm{\&phi;}\right)}{|\&dtri;\mathrm{\&phi;}|}\frac{\&PartialD;\mathrm{\&phi;}}{\&PartialD;\stackrel{\&OverBar;}{n}}=0,\mathrm{on}\&PartialD;\mathrm{\&Omega;}---\left(5\right)$

δ_ε(φ) it is the approximate of Dirac functionRepresent the outer normal vector on border, δ₀(x y) is the symbolic measurement obtained by initial boundary.Solving equation (3), the zero level collection of its steady state solution is desired object boundary curve C.

Due to only coefficient of elasticity relation between display tissue, focal area segmentation first realization on B-mode image on elastic information image, then it is mapped on elastic information image, in extracting for ensuing elastic characteristic.It should be appreciated by those skilled in the art that the edge extracting of focal area (tumor) may be used without the image partition method of other routine.

A kind of embodiment, step 106 farther includes: according to fuzzy clustering criterion, calculate in described elastic information image pixel space the fuzzy similarity matrix of each element and the square distance of cluster centre and, it is determined that element soft or hard interval degree of membership, it is achieved automatically defining of soft or hard region.

In elastic image, the tissue that coefficient of elasticity is little is softer, shown in red, and the medium tissue of coefficient of elasticity is shown in green, and tissue that coefficient of elasticity is big is harder is shown as blue.Elastic strain quantized value represents the extent of elasticity of tissue, i.e. soft or hard degree, therefore can by estimating that the hardness number of tissue quantifies its elastic strain value.In order to ensure the accuracy that elastic strain quantized value extracts, it is necessary to accurately define between the hard area in focus and neighboring area thereof.

During conventional elastic strain quantized value extracts, soft or hard region is defined by a fixed threshold.Under elastic image, the adaptability to changes that in different patient image, same color reality is corresponding is not necessarily identical, and the fixed threshold cutting techniques suitability is not strong, causes elasticity number quantization error.On mammary gland elastic information image, the change of organization internal color does not have obvious boundary, inside tumor soft or hard region is the fuzzy set of boundary fuzzy.For this feature, the present embodiment adopts fuzzy cluster analysis technique to realize automatically defining of soft or hard region:

For image after flexible reconstruction, if its gray level image is I, u_ijRepresenting that in image, jth pixel belongs to the degree of membership of the i-th class, n is the sum of all pixels of image I, then cluster object function is as follows:

$E(U,V)=\underset{i=1}{\overset{2}{\mathrm{\&Sigma;}}}\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}{\left(u_{\mathrm{ij}}\right)}^m{\left(d_{\mathrm{ij}}\right)}^2,(u_{\mathrm{ij}}\&Element;[\mathrm{0,1}],\underset{i=1}{\overset{2}{\mathrm{\&Sigma;}}}{u}_{\mathrm{ij}}=1)---\left(6\right)$

Wherein, U is initial subordinated-degree matrix；M is weighted index, and m ∈ [1 ,+∞)；d_ijFor each pixel to center vector distance；V is cluster centre, V=(v₁,v₂)^T, namely classification number is 2, i=1,2.After degree of membership and cluster centre are determined, utilize Lagrange multiplier, order:

$\left\{\begin{array}{c}F=\underset{i=1}{\overset{2}{\mathrm{\&Sigma;}}}\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}{\left(u_{\mathrm{ij}}\right)}^m{\left(d_{\mathrm{ij}}\right)}^2+\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}(\underset{i=1}{\overset{2}{\mathrm{\&Sigma;}}}{u}_{\mathrm{ij}}-1)\\ \frac{\&PartialD;F}{{\&PartialD;x}_i^{\&prime;}}=0,\frac{\&PartialD;F}{{\&PartialD;u}_{\mathrm{ij}}}=0\end{array}\right.---\left(7\right)$

Solving equation group (5), just can obtain:

$u_{\mathrm{ij}}=\frac{1}{\underset{l=1}{\overset{2}{\mathrm{\&Sigma;}}}{\left(\frac{d_{\mathrm{ij}}}{d_{\mathrm{lj}}}\right)}^{\frac{2}{(m-1)}}},(i=\mathrm{1,2};j=\mathrm{1,2},...,n)---\left(8\right)$

$v_i=\frac{\underset{n}{\overset{j=1}{\mathrm{\&Sigma;}}}{\left(u_{\mathrm{ij}}\right)}^m{x}_j}{\underset{j=1}{\overset{2}{\mathrm{\&Sigma;}}}{\left(u_{\mathrm{ij}}\right)}^m},(i=\mathrm{1,2})---\left(9\right)$

Subordinated-degree matrix U and cluster centre V obtains as follows:

Step S01: under the premise meeting degree of membership constraints, initializes subordinated-degree matrix U between 0-1；

Step S02: application (9) formula solves 2 cluster centres (territory, hard area cluster centre, soft zone territory cluster centre)；

Step S03: according to (6) given price value function.

Repeating said process, until the change of cluster centre V stops calculating less than certain threshold value or when being basically unchanged, obtaining best fuzzy classified matrix and cluster centre, thus realizing the interval definition automatically of soft or hard.It should be appreciated by those skilled in the art that the definition of soft or hard region can also apply other two sorting algorithm.

Clinical experience according to oncological pathology and doctor, on elastic image, optimum focus shows as red in yellow, and inner elastomeric distribution uniform is less with the elastic contrast of surrounding tissue；Pernicious focus shows as green to blueness, and the elasticity distribution of intralesional is uneven (such as blood vessel, calcification etc.), bigger with the elastic contrast of surrounding tissue.Good pernicious focus also has larger difference in its neighboring area, and result of study shows, pernicious focus neighboring area still there will be more blue region.Stiffness characteristics characterizes organization internal soft or hard degree by computation organization's internal territory, hard area proportion, is the feature desirably describing elastic strain.In view of innocent and malignant tumour therein and the difference of neighboring area, the present embodiment mainly extracts the stiffness characteristics of the two.

A kind of embodiment, focal area and tumor region tumorarea are its border inner region, its hardness E_tIt is represented by:

$E_t=\frac{T_h}{T_a}\&times;100\%---\left(7\right)$

In formula, T_hFor the hard area pixel in inside, focal area and, T_aFor pixel summation in focal area.E_tMore big, then focal area hardness is more big.

Focal area periphery and tumor neighboring area hardness: adopt morphological images processing method, set radius as the collar plate shape structural element of 20, lesion region tumorarea is done dilation operation, obtains tumor perienchyma region, it is expressed as tumorsudarea, its hardness E_pFor:

$E_p=\frac{P_h}{P_a}\&times;100\%---\left(8\right)$

P_hFor the internal hard area pixel in focus neighboring area and, P_aFor pathological changes neighboring pixel and.With E_tSimilar, E_pMore big, then tumor neighboring area hardness is more big.

A kind of embodiment, the predetermined standards of grading in step 110 are as follows:

1 point: focus is wholly or largely green, E_t≤20%；2 points: focus small portion is blue, green areas more than blue colored area, 20% < E_t≤50%；3 points: focus is indicated generally at the suitable green of ratio and blueness, or blue colored area is slightly more than green, 50% < E_t≤80%；4 points: within the scope of focus, be shown as blue, internal with or without a little green, 80% < E_tAnd E_p≤ 50%)；5 points: focus and perienchyma are blueness, can there be green inside, 80% < E_tAnd E_p> 50%.

According to clinical experience, being divided into separation with 3, less than 3 points (containing 3 points) are categorized as optimum, are categorized as pernicious more than 4 points (containing 4 points).

Fig. 2 illustrates the structural representation assessing one embodiment of system according to the application based on the elastic strain of ultrasonic elastic image, including rebuilding module, segmentation module, soft or hard region deviding module, stiffness characteristics extraction module and evaluation module.Rebuilding module for rebuilding Ultrasonic elasticity frame according to ultrasonic elastic image and B-mode image, Ultrasonic elasticity frame comprises elastic strain information.Segmentation module for carrying out the automatic segmentation of focal area to B-mode image.Soft or hard region deviding module defines automatically for described Ultrasonic elasticity frame is carried out soft or hard region.Hardness extraction module is for the focal area after segmentation automatically and focus neighboring area, and the soft or hard region according to automatically defining carries out stiffness characteristics extraction.Evaluation module is for based on the stiffness characteristics extracted, carrying out elastic strain assessment according to default standards of grading to focal area and focus neighboring area.

A kind of embodiment, organization ultrasonic imaging technique applies a small strain to tissue by probe apparatus, collect measured body echo-signal fragment and it is analyzed, estimating deformation extent and the tissue elasticity coefficient magnitude of tissue, then with GTG or coloud coding imaging.In elastic image, characterize tissue elasticity coefficient magnitude colouring information be superimposed upon on its B-mode image show, to tissue elasticity information retrieval analyze before, first have to reconstruct elastic information image.Rebuild in module, if elastic mixed image is I_c, B-mode image is I_b, after reconstruction, image is I_r, according to elastogram principle, elastic information image can be set up by following formula:

I_r=I_c-I_b(1)

A kind of embodiment, segmentation module is additionally operable to the automatic segmentation that B-mode image adopts the image partition method based on Mumford-Shah segmentation function and level set techniques realize focal area.

The cutting procedure of segmentation module is as follows:

For B-mode image, definition Ω is the coordinate set of its image I, if image I is divided into Cin (curvilinear inner) and two regions of Cout (curved exterior), c by curve C_inAnd c_outThe respectively average gray in the two region, definition energy function is as follows:

$F(c_{\mathrm{in}},c_{\mathrm{out}},\mathrm{\&phi;})=\mathrm{\&mu;}{\&Integral;}_{\mathrm{\&Omega;}}|\&dtri;H\left(\mathrm{\&phi;}\right)|\mathrm{dxdy}+v{\&Integral;}_{\mathrm{\&Omega;}}H\left(\mathrm{\&phi;}\right)\mathrm{dxdy}+{\mathrm{\&lambda;}}_1{\&Integral;}_{\mathrm{\&Omega;}}{|{\mathrm{\&mu;}}_0-c_{\mathrm{in}}|}^{2}H\left(\mathrm{\&phi;}\right)\mathrm{dxdy}+{\mathrm{\&lambda;}}_2{\&Integral;}_{\mathrm{\&Omega;}}{|{\mathrm{\&mu;}}_0-c_{\mathrm{out}}|}^2(1-H\left(\mathrm{\&phi;}\right))\mathrm{dxdy}$

This is (2) formula, and wherein, φ is level set function, represents Cin (curvilinear inner) and Cout (curved exterior), μ, v, λ₁, λ₂Being all constant coefficient, H (φ) is Heaviside function.When above formula energy function is minimum, the zero level collection of level set function φ is desired object boundary curve C.φ, c_inAnd c_outCan be obtained by following formula:

$c_{\mathrm{in}}\left(\mathrm{\&phi;}\right)=\frac{{\&Integral;}_{\mathrm{\&Omega;}}{u}_0(x,y)H\left(\mathrm{\&phi;}\right)\mathrm{dxdy}}{{\&Integral;}_{\mathrm{\&Omega;}}H\left(\mathrm{\&phi;}\right)\mathrm{dxdy}}---\left(3\right)$

$c_{\mathrm{out}}\left(\mathrm{\&phi;}\right)=\frac{{\&Integral;}_{\mathrm{\&Omega;}}{u}_0(x,y)(1-H\left(\mathrm{\&phi;}\right))\mathrm{dxdy}}{{\&Integral;}_{\mathrm{\&Omega;}}(1-H\left(\mathrm{\&phi;}\right))\mathrm{dxdy}}---\left(4\right)$

Function phi t in time develop, can derive about φ (t, x, Euler-Lagrange equation y) is as follows:

$\frac{\&PartialD;\mathrm{\&phi;}}{\&PartialD;t}={\mathrm{\&delta;}}_{\mathrm{\&epsiv;}}\left(\mathrm{\&phi;}\right)[\mathrm{\&mu;}\mathrm{div}\left(\frac{\&dtri;\mathrm{\&phi;}}{|\&dtri;\mathrm{\&phi;}|}\right)-v-{\mathrm{\&lambda;}}_1{(u_0-c_{\mathrm{in}})}^2+{\mathrm{\&lambda;}}_2{(u_0-c_{\mathrm{out}})}^2]=0,\mathrm{in}(0,\&infin;)\&times;\mathrm{\&Omega;}$

φ (0, x, y)=φ₀(x,y),inΩ

$\frac{{\mathrm{\&delta;}}_{\mathrm{\&epsiv;}}\left(\mathrm{\&phi;}\right)}{|\&dtri;\mathrm{\&phi;}|}\frac{\&PartialD;\mathrm{\&phi;}}{\&PartialD;\stackrel{\&OverBar;}{n}}=0,\mathrm{on}\&PartialD;\mathrm{\&Omega;}---\left(5\right)$

δ_ε(φ) being the approximate of Dirac function, n represents the outer normal vector on border, δ₀(x y) is the symbolic measurement obtained by initial boundary.Solving equation (3), the zero level collection of its steady state solution is desired object boundary curve C.

Due to only coefficient of elasticity relation between display tissue, focal area segmentation first realization on B-mode image on elastic information image, then it is mapped on elastic information image, in extracting for ensuing elastic characteristic.It should be appreciated by those skilled in the art that the edge extracting of focal area (tumor) may be used without the image partition method of other routine.

A kind of embodiment, soft or hard region deviding module is additionally operable to according to fuzzy clustering criterion, calculate in elastic information image pixel space the fuzzy similarity matrix of each element and the square distance of cluster centre and, it is determined that element soft or hard interval degree of membership, it is achieved automatically defining of soft or hard region.

In elastic image, the tissue that coefficient of elasticity is little is softer, shown in red, and the medium tissue of coefficient of elasticity is shown in green, and tissue that coefficient of elasticity is big is harder is shown as blue.Elastic strain quantized value represents the extent of elasticity of tissue, i.e. soft or hard degree, therefore can by estimating that the hardness number of tissue quantifies its elastic strain value.In order to ensure the accuracy that elastic strain quantized value extracts, it is necessary to accurately define between the hard area in focus and neighboring area thereof.

During conventional elastic strain quantized value extracts, soft or hard region is defined by a fixed threshold.Under elastic image, the adaptability to changes that in different patient image, same color reality is corresponding is not necessarily identical, and the fixed threshold cutting techniques suitability is not strong, causes elasticity number quantization error.On mammary gland elastic information image, the change of organization internal color does not have obvious boundary, inside tumor soft or hard region is the fuzzy set of boundary fuzzy.For this feature, the present embodiment adopts fuzzy cluster analysis technique to realize automatically defining of soft or hard region:

For image after flexible reconstruction, if its gray level image is I, u_ijRepresenting that in image, jth pixel belongs to the degree of membership of the i-th class, n is the sum of all pixels of image I, then cluster object function is as follows:

$E(U,V)=\underset{i=1}{\overset{2}{\mathrm{\&Sigma;}}}\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}{\left(u_{\mathrm{ij}}\right)}^m{\left(d_{\mathrm{ij}}\right)}^2,(u_{\mathrm{ij}}\&Element;[\mathrm{0,1}],\underset{i=1}{\overset{2}{\mathrm{\&Sigma;}}}{u}_{\mathrm{ij}}=1)---\left(6\right)$

Wherein, U is initial subordinated-degree matrix；M is weighted index, and m ∈ [1 ,+∞)；d_ijFor each pixel to center vector distance；V is cluster centre, V=(v₁,v₂)^T, namely classification number is 2, i=1,2.After degree of membership and cluster centre are determined, utilize Lagrange multiplier, order:

$\left\{\begin{array}{c}F=\underset{i=1}{\overset{2}{\mathrm{\&Sigma;}}}\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}{\left(u_{\mathrm{ij}}\right)}^m{\left(d_{\mathrm{ij}}\right)}^2+\underset{j=1}{\overset{n}{\mathrm{\&Sigma;}}}(\underset{i=1}{\overset{2}{\mathrm{\&Sigma;}}}{u}_{\mathrm{ij}}-1)\\ \frac{\&PartialD;F}{{\&PartialD;x}_i^{\&prime;}}=0,\frac{\&PartialD;F}{{\&PartialD;u}_{\mathrm{ij}}}=0\end{array}\right.---\left(7\right)$

Solving equation group (5), just can obtain:

$u_{\mathrm{ij}}=\frac{1}{\underset{l=1}{\overset{2}{\mathrm{\&Sigma;}}}{\left(\frac{d_{\mathrm{ij}}}{d_{\mathrm{lj}}}\right)}^{\frac{2}{(m-1)}}},(i=\mathrm{1,2};j=\mathrm{1,2},...,n)---\left(8\right)$

$v_i=\frac{\underset{n}{\overset{j=1}{\mathrm{\&Sigma;}}}{\left(u_{\mathrm{ij}}\right)}^m{x}_j}{\underset{j=1}{\overset{2}{\mathrm{\&Sigma;}}}{\left(u_{\mathrm{ij}}\right)}^m},(i=\mathrm{1,2})---\left(9\right)$

Subordinated-degree matrix U and cluster centre V obtains as follows:

Under the premise meeting degree of membership constraints, between 0-1, initialize subordinated-degree matrix U；

Application (9) formula solves 2 cluster centres (territory, hard area cluster centre, soft zone territory cluster centre)；

According to (6) given price value function.

Repeating said process, until the change of cluster centre V stops calculating less than certain threshold value or when being basically unchanged, obtaining best fuzzy classified matrix and cluster centre, thus realizing the interval definition automatically of soft or hard.It should be appreciated by those skilled in the art that the definition of soft or hard region can also apply other two sorting algorithm.

Clinical experience according to oncological pathology and doctor, on elastic image, optimum focus shows as red in yellow, and inner elastomeric distribution uniform is less with the elastic contrast of surrounding tissue；Pernicious focus shows as green to blueness, and the elasticity distribution of intralesional is uneven (such as blood vessel, calcification etc.), bigger with the elastic contrast of surrounding tissue.Good pernicious focus also has larger difference in its neighboring area, and result of study shows, pernicious focus neighboring area still there will be more blue region.Stiffness characteristics characterizes organization internal soft or hard degree by computation organization's internal territory, hard area proportion, is the feature desirably describing elastic strain.In view of innocent and malignant tumour therein and the difference of neighboring area, the present embodiment mainly extracts the stiffness characteristics of the two.

A kind of embodiment, focal area and tumor region tumorarea are its border inner region, and hardness extraction module is by its hardness E_tIt is represented by:

$E_t=\frac{T_h}{T_a}\&times;100\%---\left(7\right)$

In formula, T_hFor the hard area pixel in inside, focal area and, T_aFor pixel summation in focal area.E_tMore big, then focal area hardness is more big.

Focal area periphery and tumor neighboring area hardness: adopt morphological images processing method, set radius as the collar plate shape structural element of 20, lesion region tumorarea is done dilation operation, obtains tumor perienchyma region, being expressed as tumorsudarea, hardness extraction module is by its hardness E_pIt is expressed as:

$E_p=\frac{P_h}{P_a}\&times;100\%---\left(8\right)$

P_hFor the internal hard area pixel in focus neighboring area and, P_aFor pathological changes neighboring pixel and.With E_tSimilar, E_pMore big, then tumor neighboring area hardness is more big.

A kind of embodiment, predetermined standards of grading are:

1 point: focus is wholly or largely green, E_t≤20%；2 points: focus small portion is blue, green areas more than blue colored area, 20% < E_t≤50%；3 points: focus is indicated generally at the suitable green of ratio and blueness, or blue colored area is slightly more than green, 50% < E_t≤80%；4 points: within the scope of focus, be shown as blue, internal with or without a little green, 80% < E_tAnd E_p≤ 50%)；5 points: focus and perienchyma are blueness, can there be green inside, 80% < E_tAnd E_p> 50%.

According to clinical experience, being divided into separation with 3, less than 3 points (containing 3 points) are categorized as optimum, are categorized as pernicious more than 4 points (containing 4 points).

Above-described embodiment gives hardness elastic characteristic parameter and the method and system of elastic strain assessment, describes hardness and the lesion degree thereof of tumor.The intervention of image processing techniques, it is achieved the reconstruction of elastic information image, it is ensured that the accuracy of the processes such as feature extraction；The application of automatic cutting techniques, breaks away from conventional method by doctor's manual segmentation pattern；Fuzzy clustering techniques analytical element spatial fuzzy membership automatically extracts focal area elasticity quantitative information, it is to avoid Problems existing in existing algorithm so that handling process more comprehensively, system, science.

Above-described embodiment is to carry out stiffness characteristics extraction and quantization for tumor region application the method on mammary gland elastic image, and the method for the application can also be applied in process and the feature extraction of lesion region on the position elastic images such as liver, lung, thyroid.

Above content is further description the application made in conjunction with specific embodiment, it is impossible to assert the application be embodied as be confined to these explanations.For the application person of an ordinary skill in the technical field, under the premise conceived without departing from the application, it is also possible to make some simple deduction or replace.

Claims (10)

1. the elastic strain appraisal procedure based on ultrasonic elastic image, it is characterised in that including:

Rebuilding Ultrasonic elasticity frame according to ultrasonic elastic image and B-mode image, described Ultrasonic elasticity frame comprises elastic strain information；

Described B-mode image is carried out the automatic segmentation of focal area；

Described Ultrasonic elasticity frame is carried out soft or hard region automatically define；

To the focal area after segmentation automatically and focus neighboring area, carry out stiffness characteristics extraction according to the described soft or hard region automatically defined；

Based on the stiffness characteristics of described extraction, according to default standards of grading, focal area and focus neighboring area are carried out elastic strain assessment.

2. the method for claim 1, it is characterised in that the described segmentation automatically that described B-mode image carries out focal area includes:

Described B-mode image adopt the image partition method based on Mumford-Shah segmentation function and level set techniques realize the automatic segmentation of focal area.

3. the method for claim 1, it is characterised in that described soft or hard region that described Ultrasonic elasticity frame is carried out automatically is defined and included:

According to fuzzy clustering criterion, calculate in described elastic information image pixel space the fuzzy similarity matrix of each element and the square distance of cluster centre and, it is determined that element soft or hard interval degree of membership, it is achieved automatically defining of soft or hard region.

4. the method for claim 1, it is characterised in that described to the focal area after segmentation automatically and focus neighboring area, carries out stiffness characteristics extraction according to the described soft or hard region automatically defined and includes:

For the focal area after segmentation automatically, stiffness characteristics is represented by:Wherein T_hFor the number of pixels sum in territory, hard area, inside, focal area, T_aFor focal area interior pixels number sum；

For the focus neighboring area after segmentation automatically, stiffness characteristics is represented by:Wherein P_hFor the number of pixels sum in internal territory, hard area, focus neighboring area, P_aFor focus neighboring area interior pixels number sum.

5. method as claimed in claim 4, it is characterised in that the described stiffness characteristics based on described extraction, carries out elastic strain assessment according to default standards of grading to focal area and focus neighboring area and includes:

If E_t≤ 20%, then it is evaluated as 1 point；If 20% < E_t≤ 50%, it is evaluated as 2 points；If 50% < E_t≤ 80%, it is evaluated as 3 points；If 80% < E_tAnd E_p≤ 50%, then it is evaluated as 4 points；If 80% < E_tAnd E_p> 50%, then be evaluated as 5 points.

6. the elastic strain based on ultrasonic elastic image assesses system, it is characterised in that include rebuilding module, segmentation module, soft or hard region deviding module, stiffness characteristics extraction module and evaluation module,

Described reconstruction module is for rebuilding Ultrasonic elasticity frame according to ultrasonic elastic image and B-mode image, and described Ultrasonic elasticity frame comprises elastic strain information；

Described segmentation module for carrying out the automatic segmentation of focal area to described B-mode image；

Described soft or hard region deviding module defines automatically for described Ultrasonic elasticity frame is carried out soft or hard region；

Described stiffness characteristics extraction module is for the focal area after segmentation automatically and focus neighboring area, carrying out stiffness characteristics extraction according to the described soft or hard region automatically defined；

Focal area and focus neighboring area, for the stiffness characteristics based on described extraction, are carried out elastic strain assessment according to default standards of grading by described evaluation module.

7. system as claimed in claim 6, it is characterised in that described segmentation module is additionally operable to the automatic segmentation that described B-mode image adopts the image partition method based on Mumford-Shah segmentation function and level set techniques realize focal area.

8. system as claimed in claim 6, it is characterized in that, described soft or hard region deviding module is additionally operable to according to fuzzy clustering criterion, calculate in described elastic information image pixel space the fuzzy similarity matrix of each element and the square distance of cluster centre and, determine element soft or hard interval degree of membership, it is achieved automatically defining of soft or hard region.

9. system as claimed in claim 6, it is characterised in that described stiffness characteristics extraction module is additionally operable to:

For the focal area after segmentation automatically, stiffness characteristics is expressed as:Wherein T_hFor the number of pixels sum in territory, hard area, inside, focal area, T_aFor focal area interior pixels number sum；

For the focus neighboring area after segmentation automatically, stiffness characteristics is represented by:Wherein P_hFor the number of pixels sum in internal territory, hard area, focus neighboring area, P_aFor focus neighboring area interior pixels number sum.

10. the system as according to any one of claim 6 to 9, it is characterised in that described evaluation module is additionally operable to, if E_t≤ 20%, then it is evaluated as 1 point；If 20% < E_t≤ 50%, it is evaluated as 2 points；If 50% < E_t≤ 80%, it is evaluated as 3 points；If 80% < E_tAnd E_p≤ 50%, then it is evaluated as 4 points；If 80% < E_tAnd E_p> 50%, then be evaluated as 5 points.