CN102682449A - Automatic segmentation method based on self-adaptive external force level set for magnetic resonance images (MRIs) of soft tissue and realization method thereof - Google Patents

Abstract

The invention provides an automatic segmentation method based on self-adaptive external force level set for magnetic resonance images (MRIs) of soft tissue and a realization method thereof. The automatic segmentation method comprises the steps that object region tissue satisfying a gray threshold is selected under the circumstance of interactive operation; gradient information of a smoothened image is calculated; a curve evolution guiding function is set; a boundary stopping function is set; and if an evolutional curve exceeds the boundary of an object region, parameters of the boundary stopping function are adjusted, then curve evolution is re-conducted and the iteration is conducted again until the evolutional curve is converged within the boundary of the object region. The automatic segmentation method has the characteristics of high real-time performance, high operation efficiency, capability of segmenting multiple discrete regions at the same time, capability of accurately recognizing fuzzy boundaries of the soft tissue of a human body, high segmentation precision, clear image detail characteristics, high intelligent degree, no need of manual intervention, stable and reliable operation, and the like.

Description

Soft tissue nuclear-magnetism image adaptive external force level set is cut apart and implementation method automatically

Technical field

The present invention relates to a kind of medical image cutting method.Particularly relating to the soft tissue nuclear-magnetism image adaptive external force level set that a kind ofly has the manual intervention of need not, can be partitioned into the characteristic of destination organization rapidly and accurately automatically cuts apart and implementation method automatically.

Background technology

Image segmentation is one of classical research topic in fields such as Flame Image Process, graphical analysis and computer vision; It also is one of difficult point; Its objective is image is divided into the zone of each tool characteristic and target separated from background and noise, thereby infrastructural support is provided for follow-up quantitative, qualitative analysis.Because various images; Especially the mode of medical image is different; Introduce certain random noise in the process again inevitably, the form and the intensity difference of adding interesting target in the image are very big, also do not have a kind of method in common to be suitable for cutting apart of all images so far.MRI (nuclear magnetic resonance, Magnetic Resonance Imaging) is present very important a kind of clinical diagnostic imaging method.MRI utilizes nuclear magnetic resonance principle; According to the energy that is discharged different decay in the inner different structure environment of material; Detect the electromagnetic wave of being launched through adding gradient magnetic; Can learn to constitute nuclear position of this object and kind, can be depicted as the structural images of interior of articles in view of the above.Since the MRI technology be used for the inside of human body structure since, produced a series of revolutionary diagnostic methods in medical circle, greatly promoted developing rapidly of medical science, neuro-physiology and Cognitive Neuroscience.The MRI technology has resolving power preferably to soft tissue, become clinical in indispensable a kind of means.Yet because MRI is long sweep time, spatial resolution is not ideal enough, and the doctor is difficult to delineate out the three-D space structure of focus through two-dimentional original image; Characteristics such as simultaneously non-linear large deformation, contrast are low because human body soft tissue has, obscurity boundary etc., out-of-shape; Make lesion tissue such as with the naked eye differentiating soft tissue neoplasm clinically become difficult, take place owing to the unclear sing misdiagnosis and mistreatment that causes of identification through regular meeting; And also very complicated for the medical image segmentation of soft tissue, need the dissection doctor of specialty to carry out manual described point drafting, time and effort consuming can not adapt to clinical actual requirement, has limited the development of MRI The Application of Technology.Therefore, the research of human body soft tissue MRI image being carried out the fast automatic dividing method of medical image is very necessary and need to be resolved hurrily.

Traditional medical image cutting method be by the clinician according to dissect knowledge and experience manually point get the marginal point of area-of-interest, promptly manually cut apart, simulate smooth contour of object through interpolation etc. then.This method precision is higher, but takes time and effort, and with doctor's experience very big relation is arranged.Along with the development of computing machine and image processing techniques, a lot of computing machine partitioning algorithms have appearred, like the image partition method of classics.At present, the image segmentation algorithm of main flow mainly contains threshold method, watershed method, region growing algorithm and active contour model algorithm etc., and these methods are the special-purpose partitioning algorithm to concrete a certain or a certain type of sclerous tissues's image mostly.Segmented extraction goes out destination organization because characteristics such as human body soft tissue MRI image exists out-of-shape, non-linear large deformation, contrast is low, noise is big, classic method are difficult to therefrom.

The level set active contour model method synthesis that development in recent years is got up has utilized zone and boundary information, is widely used in fields such as image segmentation and computer vision with the continuity of its accuracy, automatism and final segmentation result.These class methods are easy to combine to cut apart prioris such as the shape constraint of object, become the hot research fields of cutting apart of medical image gradually, for the segmentation problem that solves human body soft tissue MRI image provides an approach.Level Set Method is a kind of method that Euler method is found the solution the implicit expression PDE, is proposed by people such as Osher at first, is used for the tracking at fluid motion and flame propagation interface, is introduced in computer vision and image processing field subsequently again.Its main thought be with passive curve C (p, t) be embedded into the high one-dimensional space level set function φ (x, y, t) in, ((p, t), the evolution of curve of pursuit is come in position t)=0 to C through calculated level set function zero level φ.Wherein, p is the arbitrary parameter variable, the t express time.In the practical medical image, the edge of object shows as local strength and changes the most significant part, occurs with the discontinuous form of local feature.Ask first order differential can get a unimodal function to image, peak value is corresponding with the edge of object.Unimodal function is carried out differential, and the differential value at peak value place is zero, peak value both sides opposite in sign, and original extreme point can extract edge of image corresponding to the zero cross point in the second-order differential through detecting zero cross point.At present,

In image partition method based on level set; In order to guarantee the stability of algorithm numerical evaluation; Need repeat symbolic distance function initial work consuming time in the curve evolvement process; To a great extent limit the real-time of algorithm use, can't realize cutting apart automatically of human body soft tissue MRI, can not satisfy the clinical practice requirement.People's such as Li quick variation level set model has fundamentally solved symbolic distance period of a function property initialization problem through adding a symbolic distance function punishment energy functional, has greatly improved the efficient and the robustness of algorithm.But, because this method is only utilized simple gradient information control curve evolvement, during the fuzzy or discontinuous image of segmenting edge the border leakage phenomenon often appears, can't satisfy the requirement of human body soft tissue MRI image segmentation equally.Therefore, the present invention is based on the level set theory, to human body soft tissue MRI picture characteristics, a kind of fast automatic self-adaptation external force level set algorithm and its implementation of cutting apart of human body soft tissue MRI image that be used for set up in exploitation.

Summary of the invention

Technical matters to be solved by this invention is, provide a kind of have real-time, operation efficiency is high, can cut apart simultaneously a plurality of zone of dispersions, can be accurately under the no artificial intervention situation automatically in the identification human body soft tissue MRI image soft tissue nuclear-magnetism image adaptive external force level set of the smeared out boundary of destination organization cut apart automatically and implementation method.

The technical scheme that the present invention adopted is: a kind of soft tissue nuclear-magnetism image adaptive external force level set is cut apart and implementation method automatically.The self-adaptation external force level set automatic division method of soft tissue nuclear-magnetism image comprises the steps:

1) under the interactive operation situation, human body soft tissue nuclear magnetic resonance sequence two-dimensional medical images is carried out the choosing of initial profile frame according to the gray threshold of image sequence and input, choose the target area tissue that meets gray threshold;

2) gaussian filtering that two-dimentional soft tissue nuclear magnetic resonance image is carried out in the two-dimensional space is level and smooth; And the gradient information of the level and smooth back of calculating image; At first adopt gaussian kernel function that original image is carried out smothing filtering; Remove the high frequency noise in the image background, calculate the gradient information of image behind the smothing filtering then;

3) the curve evolvement guidance function is set, makes the evolution curve select the form adaptive adjustment evolution direction of profile according to initial block;

4) border is set and stops function, the computing of beginning curve evolvement is if the evolution curve convergence is then cut apart end to object boundary; Otherwise return the curve evolvement of step 3), proceed iteration, if evolution curve directed overshoot zone boundary gets into next step;

5) if evolution curve directed overshoot zone boundary then stops function parameter adjustment to the border, return the step 3) curve evolvement then, iteration again, up to the evolution curve convergence till the border, target area.

The described initial profile of step 1) only need comprise arbitrary target area pixel point, and the shape of initial profile can arbitrarily be chosen.

Step 2) gradient information of the level and smooth back of described calculating image is:

The original image function is that (x, y), gaussian kernel function is G to I _σ,

Wherein σ is a standard variance; Image behind the smothing filtering is I ', I '=G _σ* (x, y), its gradient information does I

The described curve evolvement guidance function of step 3) is according to both sides, object edge place second-order differential opposite in sign in the real image; The extreme point of first order differential proposes corresponding to the characteristic of the zero cross point in the second-order differential; According to this characteristic, definition level set curve evolvement guidance function:

f _indicator(I)＝kgsign(I″)g(const+|I′|)

Wherein, (the second-order differential symbol of I ") expression original image I, second is used to accelerate near the evolution speed of curve object boundary to sign, even constant term const >=1 is used to guarantee in absolute level and smooth zone evolution speed still greater than zero.

The described border of step 4) stops the formula that is provided with of function:

Wherein, G _σ* I representes to use gaussian kernel function G _σImage I is carried out smothing filtering,

Expression is asked gradient to the image after level and smooth, and threshold parameter ξ can adjust according to parameters such as the resolution of pending image, gray scales, and control evolution curve finally stops at correct position; After selected threshold xi, outside specifying between gradient zones (promptly ) boundary function value perseverance be zero.

Step 5) is described to stop function parameter adjustment to the border; Be under quick variation model framework; The curve evolvement guidance function is weighted to evolution external force item; And introduce new border and stop function, obtaining a kind of new automatic parted pattern of self-adaptation external force level set, algorithm is realized through minimizing following energy functional:

ε(φ)＝αggξgLength(C)+βgf _indicatorgg _ξgS(C)+vgP _SDF(φ)

Wherein φ is the level set function of high one dimension; Under quick variation model framework; The curve evolvement guidance function is weighted to evolution external force item; And introduce new border and stop function, obtaining a kind of new automatic parted pattern of self-adaptation external force level set, algorithm is realized through minimizing described energy functional;

Utilize the first variation of variational method calculating energy functional ε, then have with respect to φ:

$\frac{\∂\mathrm{\ϵ}}{\∂\mathrm{\φ}}=-\mathrm{\δ}\left(\mathrm{\φ}\right)(\mathrm{\α}g\mathrm{div}\left(g_{\mathrm{\ξ}}\frac{\▿\mathrm{\φ}}{|\▿\mathrm{\φ}|}\right)+{\mathrm{\βgg}}_{\mathrm{\ξ}}{\mathrm{gf}}_{\mathrm{indicator}})-\mathrm{\γg}(\mathrm{\Δ\φ}-\mathrm{div}\left(\frac{\▿\mathrm{\φ}}{|\▿\mathrm{\φ}|}\right))$

If functional ε gets extreme value; Then function phi satisfies Euler-Lagrange equation

interpolation time variable t, and the gradient katabatic drainage of minimization of energy functional ε is:

$\frac{\∂\mathrm{\ϵ}}{\∂t}=\mathrm{\αg\δ}\left(\mathrm{\φ}\right)g\mathrm{div}\left(g_{\mathrm{\ξ}}\frac{\▿\mathrm{\φ}}{|\▿\mathrm{\φ}|}\right)+{\mathrm{\βgg}}_{\mathrm{\ξ}}{\mathrm{gf}}_{\mathrm{indicator}}\mathrm{g\δ}\left(\mathrm{\φ}\right)+\mathrm{\γg}(\mathrm{\Δ\φ}-\mathrm{div}\left(\frac{\▿\mathrm{\φ}}{|\▿\mathrm{\φ}|}\right))$

Wherein, first is weighting arc length item, keeps the flatness of curve in the iterative process; Second is self-adaptation external force item, and the automatic direction of propulsion of adjusting curve of mutual alignment relation according to evolution curve and object boundary guarantees that the evolution curve approaches towards the border of target all the time; The symbolic distance function characteristic of the 3rd function that is used to keep to develop; α, beta, gamma are the weighting coefficient of each item, and δ (φ) is the De Lake function, in order to the zero level collection of detection level set function φ.

Owing to adopt level set function; Need use all space local derviation number averages of space partial derivative

and

and time partial derivative

when in the curve evolvement process, carrying out gradient calculation and adopt middle difference method to approach, the time partial derivative adopts forward difference to approach;

In concrete computation process, adopt the second-order differential of Laplace operator computed image, that is:

$I^{\′\′}={\▿}^{2}I(x,y)=\frac{{\∂}^{2}I(x,y)}{{\∂x}^2}+\frac{{\∂}^{2}I(x,y)}{{\∂y}^2}$

A kind of computer implemented method of self-adaptation external force level set automatic division method of soft tissue nuclear-magnetism image is to realize self-adaptation external force level set algorithm through the arrowband method, and concrete performing step is following:

1) with initial profile φ ⁰Zero cross point z ⁰Be the initial arrowband B of central configuration ⁰, put iterations k=0, turn to step 4);

2) identify level set function φ ^kIn zero cross point z ^k, construct new arrowband B ^k

3) the newly-increased element assignment of narrowband region is established arrowband B ^kIn newly-increased element be P, Q is arrowband B ^K-1The point that middle distance P is nearest, if

Put φ k (P)=d;

4) level set upgrades in the arrowband, according to formula $\frac{\∂\mathrm{\ϵ}}{\∂t}=\mathrm{\α\; G\δ}\left(\mathrm{\φ}\right)g\mathrm{Div}\left(g_{\mathrm{\ξ}}\frac{\▿\mathrm{\φ}}{|\▿\mathrm{\φ}|}\right)+{\mathrm{\β\; Gg}}_{\mathrm{\ξ}}{\mathrm{Gf}}_{\mathrm{Indicator}}\mathrm{G\δ}\left(\mathrm{\φ}\right)+\mathrm{\γ\; g}(\mathrm{\Δ\; \φ}-\mathrm{Div}\left(\frac{\▿\mathrm{\φ}}{|\▿\mathrm{\φ}|}\right))$ Upgrade arrowband B ^kInterior level set function when satisfying the condition of convergence, stops iteration, otherwise turns to step 2).

Soft tissue nuclear-magnetism image adaptive external force level set of the present invention is cut apart and implementation method automatically; Have real-time, operation efficiency is high, can cut apart, can accurately discern the characteristics such as smeared out boundary of human body soft tissue to a plurality of zone of dispersions simultaneously; And segmentation precision is high, clear, the intelligent degree of image detail characteristic is high, need not manual intervention, stable and reliable operation, and concrete manifestation is following:

(1) two-dimentional soft tissue MRI image segmentation efficient is high.

(2) choosing of initial profile frame is irrelevant with shape, and cutting procedure need not manual intervention, has reduced method and has realized difficulty.

(3) can carry out two dimensional image to a plurality of dispersive targets zone simultaneously cuts apart.

(4) curve evolvement efficient is high, and the evolution precision is accurate.

(5) the object area segmentation quality meets the medical image quality requirements fully.

(6) can carry out processing such as filtering, elimination noise to original image, also can carry out enhancement process the target area.

(7) the details segmentation effect to the area-of-interest of skeletal muscle soft tissue MRI image is good, separates with simultaneous adaptation property strong to the form in interesting target zone.

(8) all computings are carried out in unique two-dimensional image data structure, and the data space demand is less.

The present invention is mainly used in fields such as Medical Image Processing, clinical image engineering, digital medical technology, radiation therapy, surgery planning, surgical navigational.

Description of drawings

Fig. 1 is the process flow diagram of soft tissue nuclear-magnetism image adaptive external force level set automatic division method of the present invention;

Fig. 2 is computer implemented method of the present invention and process flow diagram;

Fig. 3 is the split-run test design sketch that the present invention is applied to different initial profile images,

Wherein, (a) being 140 iteration, (b) is 90 iteration, (c) is 160 iteration, (d) is 160 iteration;

Fig. 4 is the split-run test design sketch that self-adaptation external force level set model of the present invention is applied to different soft tissue MRI images,

Wherein, (a) being 480 iteration, (b) is 300 iteration, (c) is 650 iteration, (d) is 450 iteration.

Embodiment

Below in conjunction with embodiment with accompanying drawing is cut apart by soft tissue nuclear-magnetism image adaptive external force level set of the present invention automatically and implementation method is made detailed description.

Soft tissue nuclear-magnetism image adaptive external force level set automatic division method of the present invention; It is attribute according to the second-order differential opposite in sign of zero cross point both sides in the image area; Symbol with the second-order differential value is a boot entry, drive be positioned at target internal, outside or with the crossing evolution curve of target automatically, approach to the edge of target exactly.Simultaneously, model is introduced new boundary function, can constantly adjust its parameter according to the characteristic of image to be split, guarantees that the evolution curve finally rests on the edge of target exactly.

Soft tissue nuclear-magnetism image adaptive external force level set automatic division method of the present invention comprises the steps:

1) under the interactive operation situation; Human body soft tissue nuclear magnetic resonance sequence two-dimensional medical images is carried out the choosing of initial profile frame according to the gray threshold of image sequence and input; Choose the interesting target regional organization that meets gray threshold; Described initial profile only need comprise arbitrary target area pixel point, and the shape of initial profile can arbitrarily be chosen;

2) gaussian filtering that two-dimentional soft tissue nuclear magnetic resonance image is carried out in the two-dimensional space is level and smooth; And the gradient information of the level and smooth back of calculating image; At first adopt gaussian kernel function that original image is carried out smothing filtering; Remove the high frequency noise in the image background, calculate the gradient information of image behind the smothing filtering then; The gradient information of the level and smooth back of described calculating image is:

The original image function is that (x, y), gaussian kernel function is G to I _σ,

Wherein σ is a standard variance.Image behind the smothing filtering is I ', I '=G _σ* (x, y), its gradient information does I

3) the curve evolvement guidance function is set, makes the evolution curve can select the form adaptive adjustment evolution direction of profile according to initial block.In real image, the edge of object shows as local strength and changes the most significant part, shows the discontinuous characteristic of local feature.Through image being asked first order differential can get a unimodal function, peak value is corresponding with the edge of object.To unimodal function subdifferential again, the differential value at peak value place is zero, peak value both sides opposite in sign, and original extreme point is corresponding to the zero cross point in the second-order differential.The curve evolvement guidance function that the present invention proposes promptly is that the characteristic according to this zero cross point both sides opposite in sign proposes definition level set curve evolvement guidance function:

f _indicator(I)＝kgsign(I″)g(const+|I′|)(1)

Wherein, (the second-order differential symbol of I ") expression original image I, second is used to accelerate near the evolution speed of curve object boundary to sign, even constant term const >=1 is used to guarantee in absolute level and smooth zone evolution speed still greater than zero.

4) border is set and stops function, the computing of beginning curve evolvement.The border stops function and is directly controlling the position that the evolution curve stops, if the evolution curve convergence is then cut apart end to object boundary; Otherwise return the curve evolvement of step 3), proceed iteration, if evolution curve directed overshoot zone boundary gets into next step.

Border proposed by the invention stops function:

Wherein, G _σ* I representes to use gaussian kernel function G _σImage I is carried out smothing filtering, Expression is asked gradient to the image after level and smooth, and threshold parameter ξ can adjust according to parameters such as the resolution of pending image, gray scales, and control evolution curve finally stops at correct position.Its threshold parameter of human body soft tissue MRI image ξ of different parts is also different, generally chooses destination organization edge gradient value.After selected threshold xi; The boundary function value perseverance of (i.e.

) is zero outside specifying between gradient zones; Therefore, even the iteration curve that goes down to develop the border leakage phenomenon can not occur yet always.

5) if evolution curve directed overshoot zone boundary then stops function parameter adjustment to the border, return the step 3) curve evolvement then, iteration again, up to the evolution curve convergence till the border, target area.

Described the border is stopped function parameter adjustment; Be under quick variation model framework; The curve evolvement guidance function is weighted to evolution external force item; And introduce new border and stop function, obtaining a kind of new automatic parted pattern of self-adaptation external force level set, algorithm is realized through minimizing following energy functional:

$\mathrm{\ϵ}\left(\mathrm{\φ}\right)={\mathrm{\αgg}}_{\mathrm{\ξ}}\mathrm{gLength}\left(C\right)+\mathrm{\β}{\mathrm{gf}}_{\mathrm{indicator}}{\mathrm{gg}}_{\mathrm{\ξ}}\mathrm{gS}\left(C\right)+\mathrm{vg}{P}_{\mathrm{SDF}}\left(\mathrm{\φ}\right)$

$=\mathrm{\α}{\mathrm{gg}}_{\mathrm{\ξ}}g{\∫}_{\mathrm{\Ω}}\mathrm{\δ}\left(\mathrm{\φ}\right)|\▿\mathrm{\φ}|\mathrm{dxdy}+\mathrm{\β}{\mathrm{gf}}_{\mathrm{indicator}}{\mathrm{gg}}_{\mathrm{\ξ}}g{\∫}_{\mathrm{\Ω}}H(-\mathrm{\φ})\mathrm{dxdy}+\mathrm{vg}{\∫}_{\mathrm{\Ω}}\frac{1}{2}\left(\right|\▿\mathrm{\φ}|-1)\mathrm{dxdy}---\left(3\right)$

Wherein φ is the level set function of high one dimension, under quick variation model framework, the curve evolvement guidance function is weighted to evolution external force item, and introduces new border and stop function, obtains a kind of new automatic parted pattern of self-adaptation external force level set.Algorithm is realized through minimizing above-mentioned energy functional.

Utilize the first variation of variational method calculating energy functional ε, then have with respect to level set function φ:

$\frac{\∂\mathrm{\ϵ}}{\∂\mathrm{\φ}}=-\mathrm{\δ}\left(\mathrm{\φ}\right)(\mathrm{\α}g\mathrm{div}\left(g_{\mathrm{\ξ}}\frac{\▿\mathrm{\φ}}{|\▿\mathrm{\φ}|}\right)+{\mathrm{\βgg}}_{\mathrm{\ξ}}{\mathrm{gf}}_{\mathrm{indicator}})-\mathrm{\γg}(\mathrm{\Δ\φ}-\mathrm{div}\left(\frac{\▿\mathrm{\φ}}{|\▿\mathrm{\φ}|}\right))---\left(4\right)$

If functional ε gets extreme value; Then level set function φ satisfies Euler-Lagrange equation

interpolation time variable t, and the gradient katabatic drainage of minimization of energy functional ε is:

$\frac{\∂\mathrm{\ϵ}}{\∂t}=\mathrm{\αg\δ}\left(\mathrm{\φ}\right)g\mathrm{div}\left(g_{\mathrm{\ξ}}\frac{\▿\mathrm{\φ}}{|\▿\mathrm{\φ}|}\right)+{\mathrm{\βgg}}_{\mathrm{\ξ}}{\mathrm{gf}}_{\mathrm{indicator}}\mathrm{g\δ}\left(\mathrm{\φ}\right)+\mathrm{\γg}(\mathrm{\Δ\φ}-\mathrm{div}\left(\frac{\▿\mathrm{\φ}}{|\▿\mathrm{\φ}|}\right))---\left(5\right)$

Wherein, first is weighting arc length item, keeps the flatness of curve in the iterative process; Second is self-adaptation external force item, and the automatic direction of propulsion of adjusting curve of mutual alignment relation according to evolution curve and object boundary guarantees that the evolution curve approaches towards the border of target all the time; The symbolic distance function characteristic of the 3rd function that is used to keep to develop.α, beta, gamma are the weighting coefficient of each item, and δ (φ) is the De Lake function, in order to the zero level collection of detection level set function φ.

Owing to adopt level set function; Need use all space local derviation number averages of space partial derivative and

and time partial derivative

when in the curve evolvement process, carrying out gradient calculation and adopt simple middle difference method to approach, the time partial derivative adopts forward difference to approach.

In concrete computation process, adopt the second-order differential of Laplace operator computed image, that is:

$I^{\′\′}={\▿}^{2}I(x,y)=\frac{{\∂}^{2}I(x,y)}{{\∂x}^2}+\frac{{\∂}^{2}I(x,y)}{{\∂y}^2}---\left(6\right)$

Because Laplace operator is responsive especially to noise, can produce bilateral effect, can not be used to detect the direction at edge.Therefore, we carry out filtering with Gaussian function to image earlier, with Laplace operator filtered image are asked second derivative then.

For further improving counting yield, the Dirac function adopts regularization format to describe, and adopts deposited opening (smeared-out) method of single order precision to approach.

${\mathrm{\δ}}_{{\mathrm{\ϵ}}_3}\left(z\right)=\left\{\begin{array}{cc}0,& \left|z\right|>{\mathrm{\ϵ}}_3\\ (1+\cos \left(\frac{\mathrm{\π}\·z}{{\mathrm{\ϵ}}_3}\right))/2{\mathrm{\ϵ}}_3,& \left|z\right|\≤{\mathrm{\ϵ}}_3\end{array}\right.---\left(7\right)$

Wherein, ε ₃Determined the bandwidth that numerical value is level and smooth, usually value ε ₃=1.5 Δ x.

A kind of computing machine arrowband method proposed by the invention comes the automatic partitioning algorithm of above-mentioned human body soft tissue MRI image adaptive external force level set is carried out computer realization.In cutting apart the curve evolvement process, it is a very narrow zone at center that interesting areas is merely with the zero level collection, and only can move a very little segment distance inwards or outwards from current location through outline line new after the time step.Therefore, in each iterative process, need not to consider the remote point outside the narrowband region, only need to calculate the evolution speed of net point in the arrowband.After curve evolvement arrives the border, arrowband, be the new arrowband of central configuration with zero level collection curve again again.Computational fields is limited in the belt-like zone in the iterative process, and the computation complexity of algorithm is from O (n ³) be reduced to O (kn ²), wherein k is the width of arrowband, can improve the implementation efficiency of algorithm greatly.

As shown in Figure 2, the computer implemented method of the self-adaptation external force level set automatic division method of soft tissue nuclear-magnetism image of the present invention, concrete steps are following:

1) with initial profile φ ⁰Zero cross point z ⁰Be the initial arrowband B of central configuration ⁰, put primary iteration number of times k=0, turn to step 4);

2) identify level set function φ ^kIn zero cross point z ^k, construct new arrowband B ^k, the same step 1) of method;

3) the newly-increased element assignment of narrowband region is established arrowband B ^kIn newly-increased element be P, O is arrowband B ^K-1The point that middle distance P is nearest, if

Put φ ^k(P)=d;

4) level set upgrades in the arrowband, upgrades arrowband B according to formula (8) ^kInterior level set function φ when satisfying the condition of convergence, stops iteration, otherwise turns to step 2).

$\frac{\∂\mathrm{\ϵ}}{\∂t}=\mathrm{\αg\δ}\left(\mathrm{\φ}\right)g\mathrm{div}\left(g_{\mathrm{\ξ}}\frac{\▿\mathrm{\φ}}{|\▿\mathrm{\φ}|}\right)+{\mathrm{\βgg}}_{\mathrm{\ξ}}{\mathrm{gf}}_{\mathrm{indicator}}\mathrm{g\δ}\left(\mathrm{\φ}\right)+\mathrm{\γg}(\mathrm{\Δ\φ}-\mathrm{div}\left(\frac{\▿\mathrm{\φ}}{|\▿\mathrm{\φ}|}\right))---\left(8\right)$

Fig. 3 is the automatic partitioning algorithm of self-adaptation external force level set of human body soft tissue nuclear magnetic resonance image and the split-run test design sketch that computer implemented method is applied to different initial profile images of being used for of the present invention.Get ξ=20, β=2.0, initial profile lays respectively at the diverse location of target image, after the several times iteration, all converges to the edge of object.

Fig. 4 is applied to the split-run test design sketch of different soft tissue MRI images for self-adaptation external force level set model.By figure (a), (b), can know (c) He (d) that all there is edge fog phenomenon in various degree in each interesting target in the image, and the edge of initial profile and target intersects.Through each parameter of continuous adjustment model, the evolution curve finally all converges to the edge of target exactly.Among the figure first the row in closed curve be respectively initial profile, second the row in curve be the final evolution result of initial profile.Wherein, parameter ξ in the cutting apart of the first row image=2.25, β=4.0, ξ in the secondary series=7.25, β=-2.5, ξ=6.25 in the 3rd row, β=-2.5, ξ=7.20 in last row, β=2.0.

Claims (8)

1. the self-adaptation external force level set automatic division method of a soft tissue nuclear-magnetism image is characterized in that: comprise the steps:

1) under the interactive operation situation, human body soft tissue nuclear magnetic resonance sequence two-dimensional medical images is carried out the choosing of initial profile frame according to the gray threshold of image sequence and input, choose the target area tissue that meets gray threshold;

2) gaussian filtering that two-dimentional soft tissue nuclear magnetic resonance image is carried out in the two-dimensional space is level and smooth; And the gradient information of the level and smooth back of calculating image; At first adopt gaussian kernel function that original image is carried out smothing filtering; Remove the high frequency noise in the image background, calculate the gradient information of image behind the smothing filtering then;

3) the curve evolvement guidance function is set, makes the evolution curve select the form adaptive adjustment evolution direction of profile according to initial block;

4) border is set and stops function, the computing of beginning curve evolvement is if the evolution curve convergence is then cut apart end to object boundary; Otherwise return the curve evolvement of step 3), proceed iteration, if evolution curve directed overshoot zone boundary gets into next step;

5) if evolution curve directed overshoot zone boundary then stops function parameter adjustment to the border, return the step 3) curve evolvement then, iteration again, up to the evolution curve convergence till the border, target area.

2. the self-adaptation external force level set automatic division method of a kind of soft tissue nuclear-magnetism image according to claim 1 is characterized in that the described initial profile of step 1) only need comprise arbitrary target area pixel point, and the shape of initial profile can arbitrarily be chosen.

3. the self-adaptation external force level set automatic division method of a kind of soft tissue nuclear-magnetism image according to claim 1 is characterized in that step 2) gradient information of the level and smooth back of described calculating image is:

The original image function is that (x, y), gaussian kernel function is G to I _σ,

Wherein σ is a standard variance; Image behind the smothing filtering is I ', I '=G _σ* (x, y), its gradient information does I

4. the self-adaptation external force level set automatic division method of a kind of soft tissue nuclear-magnetism image according to claim 1; It is characterized in that; The described curve evolvement guidance function of step 3) is according to both sides, object edge place second-order differential opposite in sign in the real image; The extreme point of first order differential proposes corresponding to the characteristic of the zero cross point in the second-order differential, according to this characteristic, and definition level set curve evolvement guidance function:

f _indicator(I)＝kgsign(I″)g(const+|I′|)

Wherein, (the second-order differential symbol of I ") expression original image I, second is used to accelerate near the evolution speed of curve object boundary to sign, even constant term const >=1 is used to guarantee in absolute level and smooth zone evolution speed still greater than zero.

5. the self-adaptation external force level set automatic division method of a kind of soft tissue nuclear-magnetism image according to claim 1 is characterized in that the described border of step 4) stops the formula that is provided with of function:

Wherein, G _σ* I representes to use gaussian kernel function G _σImage I is carried out smothing filtering,

Expression is asked gradient to the image after level and smooth, and threshold parameter ξ can adjust according to parameters such as the resolution of pending image, gray scales, and control evolution curve finally stops at correct position; After selected threshold xi, outside specifying between gradient zones (promptly

) boundary function value perseverance be zero.

6. the self-adaptation external force level set automatic division method of a kind of soft tissue nuclear-magnetism image according to claim 1; It is characterized in that step 5) is described to stop function parameter adjustment to the border, is under quick variation model framework; The curve evolvement guidance function is weighted to evolution external force item; And introduce new border and stop function, obtaining a kind of new automatic parted pattern of self-adaptation external force level set, algorithm is realized through minimizing following energy functional:

ε(φ)＝αgg _ξgLength(C)+βgf _indicatorgg _ξgS(C)+vgP _SDF(φ)

Wherein φ is the level set function of high one dimension; Under quick variation model framework; The curve evolvement guidance function is weighted to evolution external force item; And introduce new border and stop function, obtaining a kind of new automatic parted pattern of self-adaptation external force level set, algorithm is realized through minimizing described energy functional;

Utilize the first variation of variational method calculating energy functional ε, then have with respect to φ:

If functional ε gets extreme value; Then function phi satisfies Euler-Lagrange equation

interpolation time variable t, and the gradient katabatic drainage of minimization of energy functional ε is:

Wherein, first is weighting arc length item, keeps the flatness of curve in the iterative process; Second is self-adaptation external force item, and the automatic direction of propulsion of adjusting curve of mutual alignment relation according to evolution curve and object boundary guarantees that the evolution curve approaches towards the border of target all the time; The symbolic distance function characteristic of the 3rd function that is used to keep to develop; α, beta, gamma are the weighting coefficient of each item, and δ (φ) is the De Lake function, in order to the zero level collection of detection level set function φ.

7. the self-adaptation external force level set automatic division method of a kind of soft tissue nuclear-magnetism image according to claim 6; It is characterized in that; Owing to adopt level set function; Need use all space local derviation number averages of space partial derivative

and

and time partial derivative

when in the curve evolvement process, carrying out gradient calculation and adopt middle difference method to approach, the time partial derivative adopts forward difference to approach;

In concrete computation process, adopt the second-order differential of Laplace operator computed image, that is:

。

8. the computer implemented method of the self-adaptation external force level set automatic division method of the described soft tissue nuclear-magnetism of claim 1 image is characterized in that, is to realize self-adaptation external force level set algorithm through the arrowband method, and concrete performing step is following:

1) with initial profile φ ⁰Zero cross point z ⁰Be the initial arrowband B of central configuration ⁰, put iterations k=0, turn to step 4);

2) identify level set function φ ^kIn zero cross point z ^k, construct new arrowband B ^k

3) the newly-increased element assignment of narrowband region is established arrowband B ^kIn newly-increased element be P, Q is arrowband B ^K-1The point that middle distance P is nearest, if

Put φ ^k(P)=d;

4) level set upgrades in the arrowband, according to formula Upgrade arrowband B ^kInterior level set function when satisfying the condition of convergence, stops iteration, otherwise turns to step 2).

Images