CN103021022A - Vascular remodeling method based on parameter deformation model energy optimization - Google Patents

Abstract

The invention provides a vascular remodeling method based on parameter deformation model energy optimization. A convenient tool is provided for directed treatment of vascular diseases. The method comprises the following steps of: 1. establishing a parameter deformation model; 2. performing generalized gradient vector current (GGVF) space back projection synthesis, namely performing back projection synthesis on a GGVF energy field vector in two contrastographic pictures in a three-dimensional space according to an adaptive weighting coefficient, and taking the synthesized vector as an external force field of the parameter deformation model; 3. performing parameter deformation curve iterative approximation, namely allowing the deformation curve to continuously and iteratively approximate to a real vascular center line due to the action of the external force and self-elasticity bending internal force of the GGVF synthesis model, and obtaining the result with minimum energy under the constraint conditions with minimum distance potential energy; 4. introducing a motion estimation matrix to represent the relative displacement of blood vessels in two contrastographic pictures due to the breathing movement and heartbeat; and 5. optimizing the reconstruction result.

Description

Based on the energy-optimised reconstructing blood vessel method of parameter deformation model

Technical field

The present invention relates to a kind ofly based on the energy-optimised reconstructing blood vessel method of parameter deformation model, be suitable for diagnosis and the treatment of clinical vascular diseases.

Background technology

The principle of coronary arteriography imaging is: coronary arteriography (Coronary Angiography, CAG) refer to behind the percutaneous puncture artery optionally left and to the right coronary opening insert conduit and injection of contrast medium, thereby show a kind of angiocardiography method of coronary anatomy structure and pathology.Coronarography is usually referred to as coronary angiography, its concrete operation method is: send into a radical center conduit from the femoral artery of patient upper leg root or the elbow artery of upper arm, at x-ray imaging (X-Ray Angiography, XRA) under the guidance of image aortic root is delivered at the tip of conduit always placed the coronary artery entrance, then inject the highdensity contrast preparation of doses, high-speed photograph obtains the coronary angiography image continuously.Coronary artery XRA image can clearly demonstrate the anatomical structure of coronary artery, the narrow positions of blood vessel and stenosis, and the function situation of side Zhi Xunhuan and left ventricle etc.Check by coronary artery XRA, can effectively estimate the order of severity of coronary artery disease.Formulate best therapeutic scheme according to check result, and guide the interventional therapy of coronary heart disease.

The process of XRA system imaging is actually emissive source and launches the pencil-beam X ray, after ray passes illuminated object, through the amplification of receiving end image intensifier and imaging on image planes.In the X ray penetrating material process, owing to be subject to the modulation of material chemical constitution, density, thickness etc., the two-dimensional space that can form a Transmission X transmitted intensity distributes, and this distributes through after a series of conversion, transmitting, and finally forms visible radioscopic image.The following feature of XRA image ubiquity of obtaining in the clinical course:

The XRA image has higher ground unrest.X ray has seen through muscle, rib, conduit in penetrating the process of human body, the materials such as contrast preparation, and these materials have different attenuation coefficients, thus cause the intensity profile of image extremely inhomogeneous.Simultaneously, contrast preparation development exuberant in the conduit is very obvious, and closely similar with the gray scale of blood vessel, and severe jamming the identification of blood vessel and diagnosis.Blood-vessels is intricate, intersects and serious shielding, and there is larger difference in the projected image that different directions obtains, and correlativity is less.Exist a large amount of distortion in the image.For common contrastographic picture, with the naked eye be difficult to the distortion that identification wherein exists.

As previously mentioned, XRA technology itself exists some problems.And the three-dimensional visualization technique of blood vessel can remedy these problems effectively.This technology obtains and recovers the three-dimensional spatial information of blood vessel from limited two dimensional image, under the guidance of priori, can measure and assess the narrow of blood vessel, coronary artery is carried out motion simulation, and provide best radiography angle etc. for the doctor.Under the guiding of this technology, the doctor can position and process focus fast, thereby can effectively reduce operating time, improves operation efficient, reduces the radiant quantity of patient and the suffered X ray of doctor.The blood vessel three-dimensional visualization technique can both play an important role at aspects such as auxiliary diagnosis, Surgery Simulation, guiding treatments for the doctor provides demonstration means true to nature and the instrument of quantitative test.

In recent years, the three-dimensional reconstruction of blood vessel and visualization technique become the focus of people's research, external a large amount of scholar has launched research to the three-dimensional reconstruction based on the vascular tree of XRA image, proposed a lot of blood vessel 3 D reconstructing methods, but there is the defective of several aspects in present method for reconstructing:

1. adopt the method for polar curve constraint to seek corresponding point pair in two width of cloth contrastographic pictures, and because systematic error and blood vessel translation, the point that the polar curve constraint obtains affects reconstruction precision to there being very large error.

2. adopt the real vessel centerline of parameter deformation model iterative approach, but the original corresponding point in two width of cloth contrastographic pictures not strict corresponding point pair after mobile under to the effect in interior external force, so the vessel centerline accuracy of rebuilding out is low.

3. in the contrast imaging of monoplane, because respiratory movement and heartbeat, there is certain displacement in the blood vessel in two width of cloth contrastographic pictures, needs the optimized reconstruction result to obtain more high accuracy three-dimensional blood vessel in process of reconstruction.

XRA is very powerful visualization of blood vessels instrument, and it is widely used in bing referred to as " goldstandard " of coronary heart disease diagnosis and treatment in the middle of the diagnosis and treatment of clinical cardiovascular disease.Yet owing to be the radioscopy projection imaging, lost a large amount of space multistory information in the process of imaging, the doctor can only judge focus according to a few width of cloth two-dimensional projection image of minority usually, thereby subjective.Clinical diagnostic process often needs the doctor to possess stronger professional knowledge and a large amount of operating experiences.Blood vessel 3 D reconstructing technology based on contrastographic picture can remedy XRA equipment effectively in the deficiency aspect the imaging.This technology can not only provide clearly blood-vessels and image, blood vessel three dimensions steric information intuitively for the doctor, and relevant parameters (such as diameter, length of vessel and sectional area, optimal viewing angle etc.) that can the subsidiary blood vessel, thereby help diagnosis and the treatment of heart disease.Therefore the three-dimensional framework according to angiographic image reconstruction blood vessel has good clinical meaning and very high using value.

Summary of the invention

The present invention proposes a kind of based on the energy-optimised reconstructing blood vessel method of parameter deformation model, be the coronary arterial vessel tree skeleton three-dimensional rebuilding method of the parameter deformation model of the three-dimensional stack of a kind of generalized gradient vector flow based on the monoplane radiography (GGVF) energy field, provide a kind of easily instrument for vascular diseases for treatment.

Should based on the energy-optimised reconstructing blood vessel method of parameter deformation model, may further comprise the steps:

The first step: make up the parameter deformation model: according to two width of cloth contrastographic picture system informations, comprise that the left and right directions anglec of rotation, cephalopodium direction are revolved, radial translation, and the rotary flat that calculates between two width of cloth contrastographic pictures is answered relative matrix;

Second step: GGVF space back projection is synthetic: the GGVF energy field vector in two width of cloth contrastographic pictures is carried out back projection according to adaptive weight coefficient synthesize in three dimensions, and the vector after will synthesizing is as the external force field of parameter deformation model;

The 3rd step: parameter inflection curves iterative approach: inflection curves is the continuous real vessel centerline of iterative approach under the effect of GGVF synthetic model external force and self elastic bending internal force, and under the constraint condition of distance potential energy minimum, obtain the result of energy minimum;

The 4th step: consider that there is relative displacement in the blood vessel in two width of cloth contrastographic pictures that monoplane contrast imaging system obtains because of respiratory movement and heartbeat, introduces the relative displacement that an estimation matrix represents that the blood vessel in two width of cloth contrastographic pictures causes because of respiratory movement and heartbeat;

The 5th step: the optimization of reconstructed results: after the blood vessel initial result and primitive vessel Optimum Matching of rebuilding, adopted sparse bundle to adjust optimization method and calculate kinematic matrix and final three-dimensional blood vessel.

The advantage of the method for this blood vessel 3 D reconstructing is:

1. adopt the parameter deformation model method of " top-down " to finish reconstruction, reconstruction precision is high;

2. adopt the GGVF energy field as the external force field of deformation model, the blood vessel of reconstruction is in detail more near real blood vessel;

3. the introducing kinematic matrix comes the relative displacement between simulated blood vessel, reduces reconstruction error;

4. make up GGVF space back projection synthetic model, avoided adopting non-strict coupling corresponding point to the error of method for reconstructing;

5. adopt formation Curve Matching method to finish the intervascular matching relationship that has most, increased the accuracy of rebuilding blood vessel;

6. finish reconstruction based on monoplane imaging contrastographic picture, have widely applicability.

Description of drawings

Fig. 1 is workflow diagram proposed by the invention;

Fig. 2 is video camera projection model synoptic diagram proposed by the invention;

Fig. 3 is non-strict matching double points synoptic diagram proposed by the invention;

Fig. 4 is the three-dimensional transformation model synoptic diagram of GGVF energy field proposed by the invention;

Fig. 5 is the structural representation of system of the present invention;

Fig. 6 is the system architecture synoptic diagram of a kind of preferred embodiment of the present invention.

Embodiment

Can be further understood by the following detailed description and accompanying drawings about the advantages and spirit of the present invention.

Accompanying drawing 1 is for rebuilding process flow diagram, and described blood vessel 3 D reconstructing comprises following step:

Step S101, according to two width of cloth contrastographic picture system informations, comprise the left and right directions anglec of rotation (LAO, RAO), cephalopodium direction rotation (CRAN, CAUD), radial translation (SID), and calculate rotary flat phase shift between two width of cloth contrastographic pictures to matrix.1 X on the three dimensions medium vessels center line _i(x _i, y _i, z _i) coordinate that projects on the contrast apparatus image planes is X _{1, i}=(x _{1, i}, y _1i) ^T, (x _{1, c}, y _{1, c}) ^TCoordinate for principal point.

Step S102, for two width of cloth contrastographic pictures that read in, the GGVF energy field vector of each pixel in the computed image is with the external force field of this vector as the parameter deformation model.GGVF is defined as vector field E _Image(x, y)=[u (x, y)+v (x, y)], f (x, y) is edge map, makes energy functional obtain following minimum value:

$\mathrm{\&epsiv;}=\&Integral;\&Integral;\mathrm{\&mu;}(u_x^2+u_{\mathrm{<figure-callout\; id="2"\; label="y"\; filenames="HDA00002447491000021.png"\; state="\{\{state\}\}">y</figure-callout>}}^2+v_x^2+v_y^2)+\left|\mathrm{\&Delta;f}{|}^2\right|E_{\mathrm{image}}-\mathrm{\&Delta;f}{|}^2\mathrm{dxdy}---\left(3\right)$

μ adjusts parameter, first and second of the integrand that is used for compromising.According to the noise content in the image this parameter (image noise is large, increases the value of μ) is set.Use the numerical solution (wherein, K is weighting factor) that the variational method and method of finite difference realize the GGVF deformation model:

$\left\{\begin{array}{c}{u}_t(x,y,t)=e^{-\left(\right|\&dtri;f|/K)}{\&dtri;}^{2}u(x,y,t)-(u(x,y,t)-f_x(x,y))(1-e^{-\left(\right|\&dtri;f|/K)})\\ v_t(x,y,t)=e^{-\left(\right|\&dtri;f|/K)}{\&dtri;}^{2}v(x,y,t)-(v(x,y,t)-f_y(x,y))(1-e^{-\left(\right|\&dtri;f|/K)})\end{array}\right.---\left(4\right)$

Step S103, make up the non-closed curve parameter of a three-dimensional deformation model, this model is near the three-dimensional non-closed curve X (s) with energy=[x (s) of the definition characteristics of image of 3D region interested, y (s), z (s)], s ∈ [0,1], S is arc length parameters, and the energy of parameter deformation model can use following functional definition:

$E_{\mathrm{energy}}=\underset{0}{\overset{1}{\&Integral;}}[E_{\mathrm{internal}}\left(X\left(s\right)\right)+E_{\mathrm{external}}\left(X\left(s\right)\right)]\mathrm{ds}---\left(5\right)$

Determining of parametric line is by interior energy E _InternalWith external enwergy E _ExternalThe weighted sum minimum realize.The extreme value of energy functional can be expressed as form:

$\mathrm{\&epsiv;}\left(X\left(s\right)\right)={\&Integral;}_0^1\left\{\right[{\mathrm{\&alpha;}\left|X^{\&prime;}\left(s\right)\right|}^2+\mathrm{\&beta;}{\left|X^{\&prime;\&prime;}\left(s\right)\right|}^2]+E_{\mathrm{exteranl}}\}\mathrm{ds}$

$={\&Integral;}_0^{1}F(X\left(s\right),X_s\left(s\right),X_{\mathrm{ss}}\left(s\right))\mathrm{ds}---\left(6\right)$

Step S104, size and the direction of the suffered GGVF synthetic model external force of each pixel and self elastic bending internal force on the calculating parameter inflection curves, computing formula is as follows:

F＝β·x _i-2+(-α-4β)·x _i-1+(2α+6β)·x _i+(-α-4β)·x _i+1

+β·x _i+2+F _external.x+β·y _i-2+(-α-4β)·y _i-1+(2α+6β)·y _i

+(-α-4β)·y _i+1+β·y _i+2+F _external.y+β·z _i-2+(-α-4β)·z _i-1+(2α+6β)·z _i （7）

+(-α-4β)·z _i+1+β·z _i+2+F _external.z

＝0

Wherein, $a=-\frac{\mathrm{\&beta;}}{h^4},$ $b=(\frac{\mathrm{\&alpha;}}{{\mathrm{<figure-callout\; id="2"\; label="h"\; filenames="HDA00002447491000021.png"\; state="\{\{state\}\}">h</figure-callout>}}^2}+\frac{4\mathrm{\&beta;}}{h^4}),$ $c=-(\frac{2\mathrm{\&alpha;}}{{\mathrm{<figure-callout\; id="2"\; label="h"\; filenames="HDA00002447491000021.png"\; state="\{\{state\}\}">h</figure-callout>}}^2}+\frac{6\mathrm{\&beta;}}{h^4}),$ H is step-length (constant).

Step S105, after obtaining the suffered GGVF synthetic model external force of each pixel on the inflection curves and self elastic bending internal force, under distance potential energy constraint condition, the direction that the point of inflection curves reduces towards energy (position of real blood vessels) is carried out the iteration translation, the each time change of position all to calculate reposition apart from potential energy, when the direction that reduces towards distance potential energy moves, accept this displacement, the computing formula of minor increment potential energy constraint:

$\arg \min F\left(I_{\mathrm{energy}}\left(X_i^n\right)\right)\text{=argminF}\left(I_{\mathrm{energy}}(X_{1,i}^n,{\mathrm{<figure-callout\; id="2"\; label="X"\; filenames="HDA00002447491000021.png"\; state="\{\{state\}\}">X</figure-callout>}}_{2,i}^n)\right)$

$=\arg \min F(E_d(X_{1,i}^n,X_{1,i}^{n-1})+E_d\left({\mathrm{<figure-callout\; id="2"\; label="X"\; filenames="HDA00002447491000021.png"\; state="\{\{state\}\}">X</figure-callout>}}_{2,i}^n{-\mathrm{<figure-callout\; id="2"\; label="X"\; filenames="HDA00002447491000021.png"\; state="\{\{state\}\}">X</figure-callout>}}_{2,i}^{n-1}\right)---\left(8\right)$

$=\arg \min F\left(w_d\exp \right[-d(X_{1,i}^n,X_{1,i}^{n-1})]+w_d\exp [-d({\mathrm{<figure-callout\; id="2"\; label="X"\; filenames="HDA00002447491000021.png"\; state="\{\{state\}\}">X</figure-callout>}}_{2,i}^n-{\mathrm{<figure-callout\; id="2"\; label="X"\; filenames="HDA00002447491000021.png"\; state="\{\{state\}\}">X</figure-callout>}}_{2,i}^{n-1})\left]\right)$

Expression is apart from force,

Represent the projection of three-dimensional sampled point in two width of cloth contrastographic pictures.

Step S106, introduce kinematic matrix and represent that blood vessel in two width of cloth contrastographic pictures is because the relative displacement that respiratory movement and heartbeat cause, and by the theoretical Optimum Matching relation of setting up initial blood vessel and the primitive vessel of reconstruction of the optimal curve corresponding relation of formation curve, adopt the method for sparse bundle adjustment optimization to calculate more accurate reconstructed results.

Wherein, μ estimates for one on this mapping function, and function ψ (ξ) is the angle of the tangent line of formation curve and x axle, L, The expression curve C,

Length, h,

Represent two curve C to be matched, Arc length, κ,

Represent two curvature of a curves, R is a constant that depends on match curve sampled point quantity.

Step S107 by calculating blood vessel diameter, adopts the ray cast method to play up the three-dimensional blood-vessel image that obtains rebuilding.

System construction drawing of the present invention has added the said three-dimensional body drafting module as shown in Figure 5 in this example, and system architecture as shown in Figure 6.

Although present invention is described with reference to preferred embodiment; but the above example does not consist of the restriction of protection domain of the present invention; any in spirit of the present invention and principle modification, be equal to and replace and improvement etc., all should be included in the claim protection domain of the present invention.

Claims (4)

1. based on the energy-optimised reconstructing blood vessel method of parameter deformation model, it is characterized in that, may further comprise the steps:

The first step: make up the parameter deformation model: according to two width of cloth contrastographic picture system informations, comprise the left and right directions anglec of rotation, cephalopodium direction rotation angle, radial translation, calculate rotary flat phase shift between two width of cloth contrastographic pictures to matrix;

Second step: generalized gradient vector flow (GGVF) space back projection is synthetic: the GGVF energy field vector in two width of cloth contrastographic pictures is carried out back projection according to adaptive weight coefficient synthesize in three dimensions, and the vector after will synthesizing is as the external force field of parameter deformation model;

The 3rd step: parameter inflection curves iterative approach: inflection curves is the continuous real vessel centerline of iterative approach under the effect of GGVF synthetic model external force and self elastic bending internal force, and under the constraint condition of distance potential energy minimum, obtain the result of energy minimum;

The 4th step: consider that there is relative displacement in the blood vessel in two width of cloth contrastographic pictures that monoplane contrast imaging system obtains because of respiratory movement and heartbeat, introduces the relative displacement that an estimation matrix represents that the blood vessel in two width of cloth contrastographic pictures causes because of respiratory movement and heartbeat;

The 5th step: the optimization of reconstructed results: after the blood vessel initial result and primitive vessel Optimum Matching of rebuilding, adopt sparse bundle to adjust optimization method and calculate kinematic matrix and final three-dimensional blood vessel.

2. as claimed in claim 1 based on the energy-optimised reconstructing blood vessel method of parameter deformation model, it is characterized in that, wherein contrastographic picture GGVF space back projection is synthetic according to the vessel centerline of extracting in the contrastographic picture, calculate the GGVF energy field size and Orientation of all pixels on the vessel centerline, and according to contrastographic picture and original world coordinate system relativeness, GGVF energy field in the contrastographic picture is decomposed in the three dimensions, and superpose by weight.

3. as claimed in claim 2 based on the energy-optimised reconstructing blood vessel method of parameter deformation model, it is characterized in that, be subject in the process of interior external force generation deformation iteration convergence real blood vessels in inflection curves, by the constraint of distance potential energy, guarantee that curve moves towards the direction that potential energy reduces.

4. such as claim 1 or 2 or 3 described based on the energy-optimised reconstructing blood vessel method of parameter deformation model, it is characterized in that, wherein blood vessel parameter deformation model structure is to make up the parameter deformation model that original state is straight line in three dimensions, and by the fixing starting point and ending point coordinate of three-dimensional deformation curve of the vessel centerline information extracted in two width of cloth contrastographic pictures, iterative approach real blood vessels center line under the potential energy field constraint.

Images