CN103578099B - The extracting method of tumor elastic characteristic based on ultrasonic elastograph imaging - Google Patents

Abstract

The invention provides the extracting method of a kind of tumor elastic characteristic based on ultrasonic elastograph imaging, comprise the steps: to gather medical image and elastic image；Pretreatment to described medical image；Automatically extract the borderline tumor information of medical image after pretreatment；By described borderline tumor information fusion in described elastic image；Extract the elastic characteristic population of parameters of the elastic image merging described borderline tumor information.Above-mentioned tumor elastic characteristic extracting method based on elastogram, medical image is utilized to extract borderline tumor information, simultaneously by this borderline tumor information fusion in elastic image, and create and quantify elastic characteristic population of parameters based on elastic image, this elastic characteristic population of parameters can describe the attribute of tumor from different perspectives, thus eliminate the subjective differences of clinician, make diagnosis become the most accurate and science.

Description

The extracting method of tumor elastic characteristic based on ultrasonic elastograph imaging

Technical field

The present invention relates to image processing techniques, particularly to carrying of a kind of tumor elastic characteristic based on ultrasonic elastograph imaging Access method.

Background technology

Elastogram is that tissue is applied an external drive, and under the physics law effects such as pressure-deformation, tissue will produce A raw response, obtains elastogram figure according to this response.After own elasticity imaging concept proposes, Ultrasonic Elasticity Imaging is The nearly more than ten years have obtained rapid development.

" PLA's medical journal " volume 36 o. 11ths 1131-1133 page in 2011 and " Chinese medicine image technology " 2012 The elastic characteristic parameter area-of-interest based on ultrasonic Transient elastography technology of year volume 28 the 3rd phase 529-533 page proposition Elastic average, but the instantaneous elasticity detection technique used is a kind of one-dimensional image technology, although can be with the tissue of quantitative Average elastic modulus value, but two-dimension elastic imaging cannot be expanded to obtain tissue elasticity distributed intelligence, typically it is only applicable to Detection dispersivity pathological changes；" Chinese medicine image technology " 2010 the 9th phase 1682-1684 page of volume 26, " modern biomedical is entered Exhibition " 2010 years volume 26 the 9th phases 492-494 page and " West China medical science " 2010 volume 25 the 2nd phase 294-297 page propose based on The elastic characteristic elastic image distribution characteristics of ultrasonic real-time elastography, although give clinical diagnostic applications widest Elastic characteristic, but in real time elastography is a kind of quasistatic compression elastography, can only provide relative displacement/should Become figure, it is impossible to provide the concrete numerical value of tissue local hardness, lack objectivity and science；" Chinese medicine image technology " 2009 Year volume 18 the 7th phase 589-591 page and " world Chinese digests magazine " volume 18 the 30th phase 3254-3258 page proposition in 2010 Elastic characteristic strain rate ratio based on ultrasonic real-time elastography, is the new elastic characteristic proposed in the recent period, but from Qualitatively two-dimension displacement/strain figure extracts, although focus can be reflected than simple elasticity distribution feature more objective quantitative Firmness change degree, but this feature cannot be carried out the lateral comparison between different focus equally, such as on liver cirrhosis basis The pernicious occupying lesion of liver exists certain overlapping with the strain rate ratio interval that benign liver tumours sexually transmitted disease (STD) becomes, and only depends on Mistaken diagnosis is easily caused by strain rate ratio identification.

Therefore, conventional ultrasound elastography has many limitation so that the elastic characteristic of acquisition is qualitative or half Quantitatively, objectivity and repeatability are lacked, it is impossible to analyze quantitatively, bigger by diagnosis person's subjective impact, it is impossible to set up public affairs Recognize, the diagnostic criteria of specification, limit Ultrasonic Elasticity Imaging extensive application clinically.

Summary of the invention

Based on this, it is necessary to propose the extracting method of a kind of ultrasonic tumor elastic characteristic based on ultrasonic elastograph imaging, use In instructing the good pernicious differentiation of tumor, provide new foundation for the pathological research of tumor and early diagnosis treatment.

The extracting method of a kind of tumor elastic characteristic based on ultrasonic elastograph imaging, comprises the steps: to gather medical science figure Picture and elastic image；Pretreatment to described medical image；Automatically extract the borderline tumor letter of medical image after pretreatment Breath；By the borderline tumor information fusion of extraction in corresponding elastic image；Extract the elasticity merging described borderline tumor information The elastic characteristic population of parameters of image.

Above-mentioned tumor elastic characteristic extracting method based on elastogram, utilizes medical image to extract borderline tumor information, Simultaneously by this borderline tumor information fusion in elastic image, and create and quantify elastic characteristic parameter based on elastic image Group, this elastic characteristic population of parameters can describe the attribute of tumor from different perspectives, thus eliminate the subjective differences of clinician, make Diagnosis becomes the most accurate and science.

Accompanying drawing explanation

The flow process of the extracting method of the tumor elastic characteristic based on ultrasonic elastograph imaging that Fig. 1 provides for the embodiment of the present invention Figure.

The method flow diagram of the pretreatment of the B ultrasonic image that Fig. 2 provides for the embodiment of the present invention.

The borderline tumor information of the B ultrasonic image of the breast tumor that Fig. 3 provides for the embodiment of the present invention.

Borderline tumor information fusion the showing to elastic image extracted in the B ultrasonic image that Fig. 4 provides for the embodiment of the present invention It is intended to.

The elastic image area-of-interest that Fig. 5 provides for inventive embodiments chooses schematic diagram.

Schematic diagram is chosen in the gray level co-occurrence matrixes zoning that Fig. 6 provides for the embodiment of the present invention.

The quantization method flow chart of the quantization parameter value that Fig. 7 provides for the embodiment of the present invention.

Schematic diagram is chosen in the tumor peripheral organization region that Fig. 8 provides for the embodiment of the present invention.

Schematic diagram is chosen in the tumor center region that Fig. 9 provides for the embodiment of the present invention.

The B ultrasonic image of the mammary gland fibroadenoma that Figure 10 (a) provides for the embodiment of the present invention.

The lesion segmentation result figure of the mammary gland fibroadenoma that Figure 10 (b) provides for the embodiment of the present invention.

The springform spirogram of the mammary gland fibroadenoma that Figure 10 (c) provides for the embodiment of the present invention.

The B ultrasonic image of the infiltration ductal carcinomas of breast that Figure 11 (a) provides for the embodiment of the present invention.

The lesion segmentation result figure of the infiltration ductal carcinomas of breast that Figure 11 (b) provides for the embodiment of the present invention.

The springform spirogram of the infiltration ductal carcinomas of breast that Figure 11 (c) provides for the embodiment of the present invention.

Detailed description of the invention

Refer to Fig. 1, the extracting method of the tumor elastic characteristic of a kind of elastogram, specifically comprise the following steps that

Step S10: input medical image and elastic image.

Elastography is a kind of more ripe imaging technique, can obtain elastic image, medical science according to this technology Image can be B ultrasonic image or CT image or MRI (magnetic resonance) image or X-ray (X-ray) image, in the reality that the present invention provides Executing in example, medical image is preferably B ultrasonic image.

Step S20: the pretreatment to described medical image.

Refer to Fig. 2, for the embodiment of the present invention provide the flow chart to B ultrasonic Image semantic classification, step S20 particularly as follows:

Step S21: medical image is carried out speckle noise Filtering Processing.

The coherence of ultra sonic imaging causes the speckle noise that B ultrasonic image is intrinsic, and speckle noise reduces picture quality, especially It is to mask some detailed information of image, brings difficulty to subsequent treatment such as the rim detection of image, feature extractions.Clinical super Acoustic imaging system is built-in Nonlinear harmonic oscillator (such as logarithmic compression, low-pass filtering etc.), compression and back wave envelope signal dynamic Scope is to adapt to the small dynamic range of display device, and the speckle noise model of explicit logarithmic compression B ultrasonic image is represented by:

$I_0=I+\sqrt{I}n---\left(1\right)$

In formula (1), I is original signal, I₀For observation signal, n is zero-mean, and standard variance is σ_nGaussian noise.

Speckle noise model based on (1) formula, uses a kind of anisotropic diffusion filtering device (Speckle ReducingAnisotropic Diffusion, SRAD), it is possible to while noise reduction, retain and even strengthen the edge in image Information, Anisotropic Diffusion Model is represented by:

$\frac{\∂I}{\∂t}=div\[c\left(q\right)\▿I\]$

I (t=0)=I₀ (2)

In formula (2), q be instantaneous variation coefficient operator (Instantaneous Coefficient ofVariation, ICOV), it is the edge detector in SARD, is expressed as:

$q=\sqrt{\frac{(1/2){\left(\right|\▿I|/I)}^2-(1/16){({\▿}^{2}I/I)}^2}{{\[1+(1/4)({\▿}^{2}I/I)\]}^2}}---\left(3\right)$

In formula (3), q contains gradient operatorAnd Laplace operatorSecond dervative character can use In distinguishing the grey scale change caused by noise and the grey scale change caused by edge, therefore combineWithMake at speckle Rim detection in noise circumstance is more accurate.Diffusion coefficient c (q) is expressed as:

$c\left(q\right)=\frac{1}{1+(q^2-q_0{\left(t\right)}^2)/\[q_0{\left(t\right)}^2(1+q_0{\left(t\right)}^2)\]}---\left(4\right)$

In formula (4), t represents diffusion time, q₀T () is diffusion thresholding, ideally in reflection image, speckle noise is equal The statistical property of even distributed areas:

$q_0\left(t\right)=\frac{\sqrt{\mathrm{var}\[z\left(t\right)\]}}{\stackrel{\‾}{z\left(t\right)}}---\left(5\right)$

In formula (5), z (t) represents that speckle noise is uniformly distributed region, var [z (t)] andRepresent this region respectively Variance and average.It is typically approximately during implementing: q₀(t)≈q₀Exp (-ρ t), parameter q₀Taking 1 with ρ, iterations is 100。

Step S22: the medical image through speckle noise Filtering Processing is smoothed.

Less becoming clear is there is through speckle noise filtered B ultrasonic image relatively low gray value region (such as inside tumor) Details, higher gray value region (such as the soft tissue of surrounding) also exists less dark-coloured details, and this is mainly fine by inside tumor Structure (such as blood vessel, calcification etc.) forms, and can be smoothed by Morphologic filters.Here use alternating sequence filtering, i.e. use The structural elements of a series of continuous increases usually performs opening and closing filtering, and structural element takes collar plate shape, radius 2～5.

Step S30: automatically extract the borderline tumor information of medical image after pretreatment.

Refer to Fig. 3, for the borderline tumor information of B ultrasonic image of the breast tumor that the embodiment of the present invention provides.The present invention Use borderline tumor extracting method based on Chan-Vese model: specifically, define image I₀Coordinate set be Ω, ω be fixed Justice subgraph in Ω, curve C is the border of ω, then curve C divides an image into inside (C) and outside (C) Liang Ge district Territory.If c₁And c₂Represent the average gray in the two region respectively, and with level set function φ represent inside (C) and Outside (C), then, when following energy function minimum, it is bent that the zero level collection of level set function φ is desired object boundary Line C:

$\begin{array}{c}F(c_1,c_2,\mathrm{\φ})=\mathrm{\μ}{\∫}_{\mathrm{\Ω}}|\▿H\left(\mathrm{\φ}\right)|dxdy+v{\∫}_{\mathrm{\Ω}}H\left(\mathrm{\φ}\right)dxdy+\\ {\mathrm{\λ}}_1{\∫}_{\mathrm{\Ω}}|{\mathrm{\μ}}_0-c_1{|}^{2}H\left(\mathrm{\φ}\right)dxdy+{\mathrm{\λ}}_2{\∫}_{\mathrm{\Ω}}|{\mathrm{\μ}}_0-c_2{|}^2(1-H(\mathrm{\φ}\left)\right)dxdy\end{array}---\left(6\right)$

In formula (6), μ, v, λ₁, λ₂Being all constant coefficient, H (φ) is Heaviside function.The φ that each step is developed, c₁ And c₂Can be calculated by following two formulas:

$c_1\left(\mathrm{\φ}\right)=\frac{{\∫}_{\mathrm{\Ω}}{u}_0(x,y)H\left(\mathrm{\φ}\right)dxdy}{{\∫}_{\mathrm{\Ω}}H\left(\mathrm{\φ}\right)dxdy}---\left(7\right)$

$c_2\left(\mathrm{\φ}\right)=\frac{{\∫}_{\mathrm{\Ω}}{u}_0(x,y)(1-H(\mathrm{\φ}\left)\right)dxdy}{{\∫}_{\mathrm{\Ω}}(1-H(\mathrm{\φ}\left)\right)dxdy}---\left(8\right)$

Defined function φ t in time develops, formula (6) can derive about φ (t, x, Euler-Lagrange side y) Journey:

$\frac{\∂\mathrm{\φ}}{\∂t}={\mathrm{\δ}}_{\mathrm{\ϵ}}\left(\mathrm{\φ}\right)\[\mathrm{\μ}div\left(\frac{\▿\mathrm{\φ}}{|\▿\mathrm{\φ}|}\right)-v-{\mathrm{\λ}}_1{(u_0-c_1)}^2+{\mathrm{\λ}}_2{(u_0-c_2)}^2\]=0,in(0,\mathrm{\∞})\×\mathrm{\Ω}$

φ (0, x, y)=φ₀(x,y),inΩ

$\frac{{\mathrm{\δ}}_{\mathrm{\ϵ}}\left(\mathrm{\φ}\right)}{|\▿\mathrm{\φ}|}\frac{\∂\mathrm{\φ}}{\∂\stackrel{\‾}{n}}=0,on\∂\mathrm{\Ω}---\left(9\right)$

In formula (9), δ_ε(φ) it is the approximation of Dirac function,Represent the outer normal vector on border, δ₀(x is y) by initially The symbolic measurement that border obtains.Solving equation (9), the zero level collection of its steady state solution is desired object boundary curve C.

It is appreciated that the edge extracting of tumor may be used without the image partition method of other routine.

Step S40: by described borderline tumor information fusion in described elastic image.

Refer to Fig. 4, for the borderline tumor information fusion of extraction in the B ultrasonic image that the embodiment of the present invention provides to elastic graph The schematic diagram of picture.The clinical sonography system with elastogram function shows the B ultrasonic figure of human body same area the most simultaneously Picture and elastic image, therefore, it can, directly by the borderline tumor information fusion of extraction in B ultrasonic image to elastic image, split Go out lesion region, be designated as TUMORarea.It is appreciated that other medical images such as CT image or MRI image or X-ray image The borderline tumor information of middle extraction can also be fused in elastic image.

Step S50: the elastic characteristic population of parameters of the elastic image of borderline tumor information described in Quantitative fusion.

Elastic characteristic population of parameters includes elasticity number parameter.Elasticity number parameter includes that the elastic modelling quantity of area-of-interest is average Value, the elastic modelling quantity standard deviation of area-of-interest, the meansigma methods of lesion region, the standard deviation of lesion region, surrounding tissue area Meansigma methods and elastic ratio.

Referring again to Fig. 4, binding of pathological is gained knowledge and the clinical experience of doctor, it is considered that: optimum focus is softer, focus Inner elastomeric distribution uniform, the contrast elastic with surrounding tissue is less；Pernicious focus shows as harder, the bullet of intralesional Property skewness (such as blood vessel, calcification etc.), the contrast elastic with surrounding tissue is bigger.Pseudocolour picture is recovered to elasticity After modulus value gray-scale map, in conjunction with clinical experience and the image procossing knowledge of doctor.

Referring to Fig. 5, the elastic image area-of-interest provided for inventive embodiments chooses schematic diagram.Wherein, white is real Line rectangular area is area-of-interest.

In the embodiment that the present invention provides, area-of-interest includes lesion region and surrounding tissue area.Region of interest Territory is designated as ROIarea, and lesion region is designated as TUMORarea, and surrounding tissue area is designated as SUDarea.Wherein, area-of-interest Choosing method is:

Step one: calculate the external square of level according to the tumor's profiles curve of the elastic image merging borderline tumor information Shape.

Step 2: by boundary rectangle to four direction continuation, constitutes one and comprises tumor and size is tumor 2～3 Rectangular area again is as area-of-interest.Area-of-interest is designated as ROIarea.

Be appreciated that surrounding tissue area extracts lesion region TUMORarea remainder for region of interest ROI area Region.

In the embodiment that the present invention provides, the quantization method of the elastic modelling quantity meansigma methods of ROIarea is:

$ROIavg=\frac{1}{N_{ROI}}\underset{i=1}{\overset{N_{ROI}}{\Σ}}{e}_i---\left(10\right)$

In formula (10), N_ROIIt is the number of all pixels, e in ROIarea_iIt it is the elastic mould value that pixel is corresponding. ROIavg characterizes the average hardness of ROIarea.

In the embodiment that the present invention provides, the quantization method of the standard deviation definition of ROIarea elastic modelling quantity is:

$ROIstd=\sqrt{\frac{1}{N_{ROI}-1}\underset{i=1}{\overset{N_{ROI}}{\Σ}}{(e_i-ROIavg)}^2}---\left(11\right)$

In formula (11), N_ROIIt is the number of all pixels, e in ROIarea_iIt it is the elastic mould value that pixel is corresponding. ROIstd characterizes the uniformity coefficient of elasticity distribution in ROIarea, and ROIstd is the least, then elasticity distribution is the most uniform.

In the embodiment that the present invention provides, the quantization method of TUMORarea elastic modelling quantity meansigma methods is:

$TUMORavg=\frac{1}{N_{TUMOR}}\underset{i=1}{\overset{N_{TUMOR}}{\Σ}}{e}_i---\left(12\right)$

In formula (12), N_TUMORIt it is the number of all pixels in TUMORarea；eⁱIt it is the elastic modelling quantity that pixel is corresponding Value.TUMORavg characterizes the average hardness of tumor.

In the embodiment that the present invention provides, the quantization method of the standard deviation of TUMORarea elastic modelling quantity is:

$TUMORstd=\sqrt{\frac{1}{N_{TUMOR}-1}\underset{i=1}{\overset{N_{TUMOR}}{\Σ}}{(e_i-ROIavg)}^2}---\left(13\right)$

In formula (13), N_TUMORIt it is the number of all pixels in TUMORarea；e_iIt it is the elastic modelling quantity that pixel is corresponding Value.TUMORstd characterizes the uniformity coefficient of tumor elasticity distribution, and TUMORstd is the least, then elasticity distribution is the most uniform.

In the embodiment that the present invention provides, the elastic modelling quantity meansigma methods quantization method of SUDarea is:

$SUDavg=\frac{1}{N_{SUD}}\underset{i=1}{\overset{N_{SUD}}{\Σ}}{e}_i---\left(14\right)$

In formula (14), N_SUDIt it is the number of all pixels in SUDarea；e_iIt it is the elastic mould value that pixel is corresponding. SUDavg characterizes the average hardness of normal surrounding tissue.

In the embodiment that the present invention provides, elastic ratio quantization method is:

$Eratio=\frac{TUMORavg}{SUDavg}---\left(15\right)$

In formula (15), TUMORavg is the average hardness of TUMORarea；SUDavg is the average hardness of SUDarea. Eratio characterizes the relative resilient intensity of variation of tumor and normal surrounding tissue region.

Elastic characteristic population of parameters also includes elastic image parametric texture.Elastic image parametric texture includes: lesion region The elastic image parametric texture of elastic image parametric texture and gray level co-occurrence matrixes image-region.The elastic image stricture of vagina of lesion region Reason parameter includes: rectangular histogram normalization variance, the rectangular histogram degree of bias describe son, rectangular histogram kurtosis describes son, rectangular histogram concordance is retouched State son and entropy of histogram.The elastic image parametric texture of gray level co-occurrence matrixes image-region include co-occurrence matrix energy describe son, Co-occurrence matrix contrast describes son, co-occurrence matrix unfavourable balance square and co-occurrence matrix entropy.

In the embodiment that the present invention provides, lesion region TUMORarea rectangular histogram normalization variance quantization method is:

$H\mathrm{var}=\frac{\underset{i=0}{\overset{L-1}{\Σ}}{(z_i-m)}^{2}p\left(z_i\right)}{{(L-1)}^2}---\left(16\right)$

In formula (16), z_iExpression elastic mould value is mapped to a stochastic variable after [0,255]；M represents TUMORarea Average after the elastic modelling quantity mapping that interior all pixels are corresponding；p(z_i) it is the grey level histogram of TUMORarea；L is possible Number of greyscale levels, typically takes 256.Hvar normalizes to scope [0,1], characterizes the Discrete Distribution situation of intra-tumor elastic mould value.

In the embodiment that the present invention provides, the rectangular histogram degree of bias describes the quantization method of son and is:

$Hskew=\frac{\underset{i=0}{\overset{L-1}{\Σ}}{(z_i-m)}^{3}p\left(z_i\right)}{{(L-1)}^2}---\left(17\right)$

Z in formula (17)_iExpression elastic mould value is mapped to a stochastic variable after [0,255], in m represents lesion region Average after the elastic modelling quantity mapping that all pixels are corresponding, p (z_i) it is the grey level histogram of lesion region, L is possible ash Degree progression, Hskew phenogram is as the asymmetric degree of histogram distribution.

Hskew is the biggest, represents that histogram distribution is the most asymmetric, otherwise the most symmetrical.

In the embodiment that the present invention provides, rectangular histogram kurtosis describes the quantization method of son and is:

$Hkurt=\frac{\underset{i=0}{\overset{L-1}{\Σ}}{(z_i-m)}^{4}p\left(z_i\right)}{{(L-1)}^2}---\left(18\right)$

Z in formula (18)_iExpression elastic mould value is mapped to a stochastic variable after [0,255], in m represents lesion region Average after the elastic modelling quantity mapping that all pixels are corresponding, p (z_i) it is the grey level histogram of lesion region, L is possible ash Degree progression, Hkurt characterizes elastic image and is distributed in close to approximate state during average, in order to whether to judge the elasticity distribution of image Concentrate near average elasticity value.Hkurt is the least, represents and more concentrates；Otherwise more dispersion.

In the embodiment that the present invention provides, TUMORarea rectangular histogram concordance describes sub-definite and is:

$Henergy=\underset{i=0}{\overset{L-1}{\Σ}}{p}^2\left(z_i\right)---\left(19\right)$

In formula (19), z_iExpression elastic mould value is mapped to a stochastic variable after [0,255]；p(z_i) be The grey level histogram of TUMORarea；L is possible number of greyscale levels.Henergy characterizes the uniformity coefficient of tumor elasticity distribution, point The more uniform duration of cloth is relatively big, otherwise less.

In the embodiment that the present invention provides, TUMORarea entropy of histogram is defined as:

$Hentropy=-\underset{i=0}{\overset{L-1}{\Σ}}p\left(z_i\right){\log }_{2}p\left(z_i\right)---\left(20\right)$

In formula (20), z_iExpression elastic mould value is mapped to a stochastic variable after [0,255]；p(z_i) be The grey level histogram of TUMORarea；L is possible number of greyscale levels.Hentropy also characterizes the uniformity of tumor elasticity distribution.

Referring to Fig. 6, schematic diagram is chosen in the gray level co-occurrence matrixes zoning provided for the embodiment of the present invention.In Fig. 6, in vain Color solid-line rectangle region is tumor minimum enclosed rectangle region, is used for calculating characteristic parameter based on gray level co-occurrence matrixes.It is designated as MBBarea, calculating distance is 2, from top to bottom the gray level co-occurrence matrixes G in direction.

In the embodiment that the present invention provides, the choosing method of gray level co-occurrence matrixes image-region is: according to merging tumor The tumor's profiles curve of the elastic image of marginal information calculates minimum enclosed rectangle region as gray level co-occurrence matrixes image district Territory.

In the embodiment that the present invention provides, the co-occurrence matrix energy of MBBarea describes sub-quantization method and is:

$ASM=\stackrel{K}{\Σ}\stackrel{K}{\Σ}{\left\{G(i,j)\right\}}^2---\left(21\right)$

In formula (21), (i j) represents each element value of gray level co-occurrence matrixes G that size is K × K to G.ASM phenogram picture Gray scale (elastic) is evenly distributed degree and texture fineness degree.Elastic image distribution is the most uniformly worth the biggest, otherwise less.

In the embodiment that the present invention provides, the co-occurrence matrix contrast of MBBarea describes sub-quantization method and is:

$CON=\underset{i=1}{\overset{K}{\Σ}}\underset{j=1}{\overset{K}{\Σ}}{(i-j)}^{2}G(i,j)---\left(22\right)$

In formula (22), (i j) represents each element value of gray level co-occurrence matrixes G that size is K × K to G.CON phenogram picture Definition and the degree of the texture rill depth.Texture rill is the deepest, and CON is the biggest, and visual effect is the most clear；Otherwise it is less.

In the embodiment that the present invention provides, the co-occurrence matrix unfavourable balance square quantization method of MBBarea is:

$IDM=\underset{i=1}{\overset{K}{\Σ}}\underset{j=1}{\overset{K}{\Σ}}\frac{G(i,j)}{1+{(i-j)}^2}---\left(23\right)$

In formula (23), (i j) represents each element value of gray level co-occurrence matrixes G that size is K × K to G.IDM phenogram picture Homogeneity, tolerance image texture local intensity of variation.Change is lacked between the zones of different of IDM the biggest explanation image texture, The most highly uniform.

In the embodiment that the present invention provides, the co-occurrence matrix entropy quantization method of MBBarea is:

$ENT=-\underset{i=1}{\overset{K}{\Σ}}\underset{j=1}{\overset{K}{\Σ}}G(i,j){\log }_{2}G(i,j)---\left(24\right)$

In formula (24), (i j) represents each element value of gray level co-occurrence matrixes G that size is K × K to G.ENT phenogram picture The non-uniform degree of texture or complexity.In in co-occurrence matrix, all elements has the randomness of maximum, space co-occurrence matrix When all values is almost equal, in co-occurrence matrix during element dispersed and distributed, ENT is bigger.

Elastic characteristic population of parameters also includes quantization parameter.

Refer to Fig. 7, the quantization method flow chart of quantization parameter value provided for the embodiment of the present invention, including

Step S51: lesion region is done dilation operation and obtains tumor peripheral organization region.

Referring to Fig. 8, schematic diagram is chosen in the tumor peripheral organization region provided for the embodiment of the present invention, and wherein, white is real Region between line and grey filled lines is tumor peripheral organization region.Use morphological images processing method, set radius as 20 Collar plate shape structural element, lesion region TUMORarea is done dilation operation, obtains tumor peripheral organization region, be expressed as TUMORSUDarea。

Step S52: described lesion region is done erosion operation and obtains pathological changes central area.

Refer to Fig. 9, for the embodiment of the present invention provide tumor center region choose schematic diagram, wherein, solid white line with Region between grey filled lines is tumor center region.Set the circle that radius is as the 1/5 of TUMORarea boundary rectangle minor axis length Dish configuration element, does erosion operation to TUMORarea, obtains pathological changes central area, is expressed as CENTERarea.

Step S53: be defined as follows parameter:

$TUMORsoft=\frac{N_{soft}}{N},TUMORhard=\frac{N_{hard}}{N}---\left(25\right)$

In formula (25), N is the number of all pixels in representing TUMORarea；N_softRepresent elastic modelling quantity in TUMORarea The value pixel number less than ROIavg；Nhard represents the TUMORarea elastic mould value pixel number more than ROIavg.

$TUMORSUDsoft=\frac{M_{soft}}{M},TUMORSUDhard=\frac{M_{hard}}{M}---\left(26\right)$

In formula (26), M is the number of all pixels in representing TUMORSUDarea；M_softRepresent bullet in TUMORSUDarea Property modulus value less than the pixel number of ROIavg；M_hardThe TUMORSUDarea elastic mould value picture more than ROIavg in representing Vegetarian refreshments number.

$CENTERhard=\frac{L_{hard}}{L}---\left(27\right)$

In formula (27), L is the number of all pixels in representing CENTERarea；L_hardRepresent CENTERarea Elastic value Pixel number more than ROIavg.

Step S54: elastic scoring is as follows:

When TUMORsoft >=90%, quantization parameter value Escore is 1；

When TUMORsoft-TUMORhard >=10%, quantization parameter value Escore is 2；

When TUMORhard-TUMORsoft >=10%, quantization parameter value Escore is 3；

When TUMORhard >=80%, TUMORSUFsoft >=70%, quantization parameter value Escore is 4；

When TUMORhard >=90%, TUMORSUFhard >=50%, quantization parameter value Escore is 5.

Above-mentioned quantization parameter uses elastic point system to carry out quantization signifying lesion region elasticity Relative distribution situation, but needs To combine with elasticity number parameter, tumor could be made and diagnosing accurately.

The above embodiment of the present invention, gives the quantization method of above 16 elastic characteristic parameters, these characteristic parameters group Describe the attribute of tumor from different perspectives.In actual applications, by these identification abilitys strong, objective, repeatable by force Characteristic parameter carry out integrated decision-making, adjuvant clinical doctor diagnose, and draws accurate result.

Below by way of specific embodiment, the present invention is further described.

Method proposed by the invention 3rd hospital's Ultrasonography attached to Zhongshan University is used to pass through Supersonic The ultrasonic breast tumor image (including conventional B ultrasonic image and the elastic modelling quantity image of focus) that Aixplorer Ultrasound Instrument gathers enters The quantization of elastic characteristic and extraction are gone.This apparatus preparation has shearing wave elastogram module, uses SC15-4 linear array probe, frequently Rate scope is 4-15MHz.When gathering data, under two-dimensional ultrasound pattern, first carry out the scanning of horizontal, the vertical each tangent plane of mammary gland, find After breast tumor, observe the size of focus, form, border, internal echo and the Infiltrating of surrounding tissue, record focus Position.Then switching to ultrasonic elastograph imaging checking mode, sampling frame is about 2-3 times of size of tumor, if focus is relatively big, only Take being partially located in sampling frame of focus.Double width shows two dimension B ultrasonic image and elastic image simultaneously, keeps probe during operation as far as possible Stablize, maintain 2 seconds, Cryopreservation static state B ultrasonic image and elastic image.

See also Figure 10 (a), 10 (b) and 10 (c), Figure 11 (a), 11 (b) and 11 (c), table 1 and table 2.Be given respectively The lesion region segmentation of fibroadenoma and IDC and characteristic quantification extract results, it can be seen that optimum focus is hard Spending less, intralesional elasticity distribution is more uniform, and the contrast elastic with surrounding tissue is less；Pernicious focus hardness is relatively big, sick Elasticity distribution within stove is uneven, and the contrast elastic with surrounding tissue is bigger.From the point of view of clinical trial results, the present invention carries The extracting method of the tumor elastic characteristic of the elastogram gone out, can be used for instructing the differentiation good, pernicious of tumor, for the pathology of tumor Learn research and early diagnosis treats the foundation providing new.

Table 1 elastic characteristic parameter (1)

Table 2 elastic characteristic parameter (2)

The above, be only presently preferred embodiments of the present invention, and the present invention not makees any pro forma restriction, though So the present invention is disclosed above with preferred embodiment, but is not limited to the present invention, any technology people being familiar with this specialty Member, in the range of without departing from technical solution of the present invention, when the technology contents of available the disclosure above makes a little change or modification For the Equivalent embodiments of equivalent variations, as long as being without departing from technical solution of the present invention content, according to the technical spirit pair of the present invention Any simple modification, equivalent variations and the modification that above example is made, all still falls within the range of technical solution of the present invention.

Claims (7)

1. the extracting method of a tumor elastic characteristic based on ultrasonic elastograph imaging, it is characterised in that comprise the steps:

(1) medical image and elastic image are gathered；

(2) described medical image is carried out pretreatment, medical image is carried out speckle noise Filtering Processing；Filter through speckle noise The medical image that ripple processes is smoothed；

(3) automatically extract the borderline tumor information of medical image after pretreatment, described in automatically extract employing based on Chan- The borderline tumor extracting method of Vese model；

(4) by the borderline tumor information fusion of extraction in corresponding elastic image；

(5) extract the elastic characteristic population of parameters of the elastic image merging described borderline tumor information, choose area-of-interest, feel emerging Interest region includes lesion region and surrounding tissue area, and wherein, elastic characteristic population of parameters includes elasticity number parameter, described elasticity number Parameter include the elastic modelling quantity meansigma methods of area-of-interest, the elastic modelling quantity standard deviation of area-of-interest, lesion region average Value, the standard deviation of lesion region, the meansigma methods of surrounding tissue area and elastic ratio；

The quantization method of the elastic modelling quantity meansigma methods of described area-of-interest is:

$ROIavg=\frac{1}{N_{ROI}}\underset{i=1}{\overset{N_{ROI}}{\Σ}}{e}_i$

N in formula_ROIFor the number of pixels all in described area-of-interest, e_iIt is the elastic mould value that pixel is corresponding, e_iI For pixel, ROIavg characterizes the average hardness of area-of-interest；

The quantization method of the elastic modelling quantity standard deviation of described area-of-interest is:

$ROIstd=\sqrt{\frac{1}{N_{ROI}-1}\underset{i=1}{\overset{N_{ROI}}{\Σ}}{(e_i-ROIavg)}^2}$

N in formula_ROIFor the number of pixels all in described area-of-interest, e_iIt is the elastic mould value that pixel is corresponding, ROIstd characterizes the uniformity coefficient of elasticity distribution in area-of-interest；

The quantization method of the meansigma methods of described lesion region is:

$TUMORavg=\frac{1}{N_{TUMOR}}\underset{i=1}{\overset{N_{TUMOR}}{\Σ}}{e}_i$

N in formula_TUMORNumber for pixels all in described lesion region；e_iFor the elastic mould value that pixel is corresponding, TUMORavg characterizes the average hardness of lesion region；

The quantization method of the standard deviation of described lesion region is:

$TUMORstd=\sqrt{\frac{1}{N_{TUMOR}-1}\underset{i=1}{\overset{N_{TUMOR}}{\Σ}}{(e_i-ROIavg)}^2}$

N in formula_TUMORNumber for pixels all in described lesion region；e_iFor the elastic mould value that pixel is corresponding, TUMORstd characterizes the uniformity coefficient of tumor elasticity distribution；

The quantization method of the meansigma methods of described surrounding tissue area is:

$SUDavg=\frac{1}{N_{SUD}}\underset{i=1}{\overset{N_{SUD}}{\Σ}}{e}_i$

N in formula_SUDNumber for pixels all in surrounding tissue area；e_iIt is the elastic mould value that pixel is corresponding, e_iI be Pixel, SUDavg characterizes the average hardness of surrounding tissue area；

The quantization method of described elastic ratio is:

$Eratio=\frac{TUMORavg}{SUDavg}$

In formula, TUMORavg is the average hardness of lesion region；SUDavg is the average hardness of normal surrounding tissue, Eratio table Levy the relative resilient intensity of variation of tumor and normal surrounding tissue region；

Described elastic characteristic population of parameters also includes that quantization parameter, the quantization method of described quantization parameter value are:

Described lesion region is done dilation operation and obtains tumor peripheral organization region；

Described lesion region is done erosion operation and obtains pathological changes central area；

It is defined as follows parameter:

$TUMORsoft=\frac{N_{soft}}{N},TUMORhard=\frac{N_{hard}}{N}$

The N number of all pixels, N in representing described lesion region in formula_softIn representing described lesion region, elastic mould value is little In the pixel number of described ROIavg, N_hardRepresent the described lesion region elastic mould value pixel more than described ROIavg Number,

$TUMORSUDsoft=\frac{M_{soft}}{M},TUMORSUDhard=\frac{M_{hard}}{M}$

The number of all pixels, M in formula, M represents described tumor peripheral organization region_softIn representing tumor peripheral organization region The elastic mould value pixel number less than described ROIavg, M_hardRepresent that tumor peripheral organization region elastic mould value is more than institute State the pixel number of ROIavg,

$CENTERhard=\frac{L_{hard}}{L}$

The number of all pixels, L in formula, L represents described pathological changes central area_hardRepresent described pathological changes central area Elastic The value pixel number more than described ROIavg；

Elastic scoring: when TUMORsoft >=90%, quantization parameter value Escore is 1；

When TUMORsoft-TUMORhard >=10%, quantization parameter value Escore is 2；

When TUMORhard-TUMORsoft >=10%, quantization parameter value Escore is 3；

When TUMORhard >=80%, TUMORSUDsoft >=70%, quantization parameter value Escore is 4；

When TUMORhard >=90%, TUMORSUDhard >=50%, quantization parameter value Escore is 5.

The extracting method of tumor elastic characteristic based on ultrasonic elastograph imaging the most according to claim 1, it is characterised in that Described area-of-interest includes described lesion region and described surrounding tissue area, and the choosing method of described area-of-interest is:

Tumor's profiles curve according to the elastic image merging described borderline tumor information calculates horizontal boundary rectangle；

By described boundary rectangle to four direction continuation, constitute one and comprise tumor and size is tumor 2～the rectangle of 3 times Region is as described area-of-interest.

The extracting method of tumor elastic characteristic based on ultrasonic elastograph imaging the most according to claim 1, it is characterised in that Described elastic characteristic population of parameters also includes that elastic image parametric texture, described elastic image parametric texture include: lesion region The elastic image parametric texture of elastic image parametric texture and gray level co-occurrence matrixes image-region.

The extracting method of tumor elastic characteristic based on ultrasonic elastograph imaging the most according to claim 3, it is characterised in that The elastic image parametric texture of described lesion region includes: rectangular histogram normalization variance, the rectangular histogram degree of bias describe son, rectangular histogram peak Degree describes son, rectangular histogram concordance describes son and entropy of histogram；The elastic image texture of described gray level co-occurrence matrixes image-region Parameter includes that co-occurrence matrix energy describes son, co-occurrence matrix contrast describes son, co-occurrence matrix unfavourable balance square and co-occurrence matrix entropy.

The extracting method of tumor elastic characteristic based on ultrasonic elastograph imaging the most according to claim 4, it is characterised in that The choosing method of described gray level co-occurrence matrixes image-region is:

Tumor's profiles curve according to the elastic image merging described borderline tumor information calculates minimum enclosed rectangle region and makees For described gray level co-occurrence matrixes image-region.

The extracting method of tumor elastic characteristic based on ultrasonic elastograph imaging the most according to claim 4, it is characterised in that The quantization method of described rectangular histogram normalization variance is:

$H\mathrm{var}=\frac{\underset{i=0}{\overset{L-1}{\Σ}}{(z_i-m)}^{2}p\left(z_i\right)}{{(L-1)}^2}$

Z in formula_iExpression elastic mould value is mapped to a stochastic variable after [0,255]；M is all pixels in representing lesion region Average after the elastic modelling quantity mapping that point is corresponding；p(z_i) it is the grey level histogram of lesion region；L is possible number of greyscale levels, takes 256, Hvar normalize to scope [0,1], characterize the Discrete Distribution situation of intra-tumor elastic mould value；

The described rectangular histogram degree of bias describes the quantization method of son:

$Hskew=\frac{\underset{i=0}{\overset{L-1}{\Σ}}{(z_i-m)}^{3}p\left(z_i\right)}{{(L-1)}^2}$

Z in formula_iExpression elastic mould value is mapped to a stochastic variable after [0,255], and m is all pixels in representing lesion region Average after the elastic modelling quantity mapping that point is corresponding, p (z_i) it is the grey level histogram of lesion region, L is possible number of greyscale levels, Hskew phenogram is as the asymmetric degree of histogram distribution；

Described rectangular histogram kurtosis describes the quantization method of son:

$Hkurt=\frac{\underset{i=0}{\overset{L-1}{\Σ}}{(z_i-m)}^{4}p\left(z_i\right)}{{(L-1)}^2}$

Z in formula_iExpression elastic mould value is mapped to a stochastic variable after [0,255], and m is all pixels in representing lesion region Average after the elastic modelling quantity mapping that point is corresponding, p (z_i) it is the grey level histogram of lesion region, L is possible number of greyscale levels, Hkurt characterizes elastic image and is distributed in close to approximate state during average；

Described rectangular histogram concordance describes the quantization method of son:

$Henergy=\underset{i=0}{\overset{L-1}{\Σ}}{p}^2\left(z_i\right)$

Z in formula_iExpression elastic mould value is mapped to a stochastic variable after [0,255], p (z_i) it is that the gray scale of lesion region is straight Fang Tu, L are possible number of greyscale levels, and Henergy characterizes the uniformity coefficient of tumor elasticity distribution；

The quantization method of described entropy of histogram is:

$Hentropy=-\underset{i=0}{\overset{L-1}{\Σ}}p\left(z_i\right){\log }_{2}p\left(z_i\right)$

Z in formula_iExpression elastic mould value is mapped to a stochastic variable after [0,255], p (z_i) it is that the gray scale of lesion region is straight Fang Tu, L are possible number of greyscale levels, and Hentropy characterizes the uniformity of tumor elasticity distribution；

Described co-occurrence matrix energy describes the quantization method of son:

$ASM=\underset{i=1}{\overset{K}{\Σ}}\underset{j=1}{\overset{K}{\Σ}}{\left\{G(i,j)\right\}}^2$

(i, j) represents each element value of gray level co-occurrence matrixes G that size is K × K, and ASM phenogram is equal as intensity profile for G in formula Even degree and texture fineness degree；

Described co-occurrence matrix contrast describes the quantization method of son:

$CON=\underset{i=1}{\overset{K}{\Σ}}\underset{j=1}{\overset{K}{\Σ}}{(i-j)}^{2}G(i,j)$

G in formula (i, j) represents each element value of gray level co-occurrence matrixes G that size is K × K, the definition of CON phenogram picture and The degree of the texture rill depth；

The quantization method of described co-occurrence matrix unfavourable balance square is:

$IDM=\underset{i=1}{\overset{K}{\Σ}}\underset{j=1}{\overset{K}{\Σ}}\frac{G(i,j)}{1+{(i-j)}^2}$

G in formula (i, j) represents each element value of gray level co-occurrence matrixes G that size is K × K, the homogeneity of IDM phenogram picture, The intensity of variation of tolerance image texture local；

The quantization method of described co-occurrence matrix entropy is:

$ENT=-\underset{i=1}{\overset{K}{\Σ}}\underset{j=1}{\overset{K}{\Σ}}G(i,j){\log }_{2}G(i,j)$

G in formula (i, j) represents each element value of gray level co-occurrence matrixes G that size is K × K, ENT characterize image texture non-all Even degree or complexity.

The extracting method of tumor elastic characteristic based on ultrasonic elastograph imaging the most according to claim 1, it is characterised in that Described medical image is traditional B hypergraph picture or CT image or MRI image or X-ray image.