CN103049901A - Magnetic resonance diffusion tensor imaging fiber bundle tracking device - Google Patents

Abstract

The invention relates to a magnetic resonance diffusion tensor imaging fiber bundle tracking device. The process is respectively completed by each component through the following steps of (1) collecting magnetic resonance diffusion tensor images; (2) carrying out brain issue dividing on a sequence without a diffusion gradient magnetic field and any sequence with the diffusion gradient magnetic field, and using the sequence without the diffusion gradient magnetic field as an image reference template of the brain tissues; (3) carrying out three-dimensional affine conversion on the two image sequences of the brain tissues after being extracted, to obtain a space conversion relationship; (4) carrying out space conversion on the remained diffusion tensor images by an optimum conversion relationship; (5) calculating tensor fields and feature vectors; (6) setting the interested areas and tracking conditions; and (7) carrying out the bidirectional tracking and displaying based on a fiber bundle with variable step size, so as to quickly and effectively carry out fiber bundle tracking and displaying on the white matters of human brain. In the diffusion tensor imaging process, the image deviation caused by space positions can be corrected, and the elastic step size is adopted in the tracking process, so as to ensure the reliable fiber bundle tracking.

Description

Diffusion tensor imaging fibrous bundle follow-up mechanism

Technical field

The present invention relates to the diffusion tensor imaging fibrous bundle follow-up mechanism that a kind of problem for solving the demonstration of human brain diffusion tensor imaging brain white matter integrity bundle also can be applied to the fields such as clinic diagnosis of people's cerebral disease.

Background technology

The multi-sequence image of diffusion tensor imaging produces the inconsistent situation in sequence image locus easily because acquisition time is different, thereby causes the tensor of each voxel to produce error, and final so that brain white matter integrity bundle is followed the trail of and is difficult to the effect that reaches satisfied.Even used head fixing device in the actual scanning process, it is moving to eliminate head fully, thereby the locus of multi-sequence image is proofreaied and correct extremely important.Among the tracing process of cranial nerve fibrous bundle, brain tissue border template is extremely important.The obtaining of brain tissue border template can be shortened tensor field computing time, and can determine the border that brain tissue is followed the trail of, and guarantees the reliability of following the trail of.The fibrous bundle tracing process is an iterative process, and the selection of step-length is even more important in the iteration, complicated out of shape between the white matter of brain nerve fibre bundle, adopt variable step size describe its path then tool more have superiority, can accurately express complicated fibrous bundle.

Yet existing diffusion tensor imaging fibrous bundle tracer technique all is difficult to solve well whole issue.Traditional method is when carrying out the locus correction, adopting original image to carry out the locus corrects, and contain a large amount of interference pixels in the original image, such as scalp soft tissue etc., this is so that the locus error-correcting effect is limited, therefore obtain by the brain tissue template and carry out the locus correction, then have strong robustness.It is generally acknowledged that for the expression of different fibrous bundles, its tracing process just is equivalent to the integral process that begins from initial point, the step-length that sets during this time is impartial, when describing the larger fibrous bundle of deflection angle, distortion must occur etc. the step-length iteration like this, finally cause the increase of error.

Summary of the invention

The object of the invention is to high precision to finish the problem that the diffusion tensor imaging fibrous bundle is followed the trail of for prior art, providing a kind of result who the scanning of human brain is formed according to the brain function image capture device to carry out the diffusion tensor imaging fibrous bundle follow-up mechanism that the diffusion tensor imaging fibrous bundle follows the trail of comprises: gather different sequence diffusion tensor imaging images at least ten two directions of human brain by acquisition matrix from the brain function image capture device, get any one sequence in the different sequence diffusion tensor imaging images as the collection section of reference image; Different sequence diffusion tensor imaging images is carried out normalized so that the equal consistent normalized section of the pixel size of all sequences diffusion tensor imaging image and physical dimension; Thereby the reference image is extracted the cutting part of cutting apart the benchmark Brain Tissues Image that obtains the corresponding sequence that does not apply disperse magnetic field and the Brain Tissues Image to be corrected of the sequence that applies diffusion gradient magnetic field, and this benchmark Brain Tissues Image is as brain tissue standard masterplate; The benchmark Brain Tissues Image that goes out with reference to image according to the spatial alternation matrix computations reaches in the locus one to one with reference to image rectification section with reference to the pixel of the benchmark Brain Tissues Image of image and the pixel of Brain Tissues Image to be corrected as the realization of brain tissue standard masterplate with the spatial alternation relation of Brain Tissues Image to be corrected and with the benchmark Brain Tissues Image; Except forming the corresponding remedial frames for the treatment of with reference to all sequences diffusion tensor imaging image the image, remedial frames carries out three-dimensional affine transformation so that the pixel of all sequence diffusion tensor imaging images reaches one to one image registration correction section in the locus thereby treat according to the spatial alternation relation; Go out the tensor field of all sequences diffusion tensor imaging image and the disperse information of each voxel of all sequences diffusion tensor imaging image tensor with 3 * 3 matrixes is represented by matrix computations, thereby each voxel is carried out the calculating part that matrix decomposition obtains eigenwert and the proper vector of each voxel.By region of interest the region of interest that module select to need is followed the trail of at brain tissue standard masterplate is set, thus the region of interest selection portion of the Seed Points of the region of interest that the even interpolation of region of interest is obtained; The trace parameters that needed trace parameters when following the trail of is set arranges section, and trace parameters has brain tissue anisotropy threshold value, follows the trail of angle threshold, iterative steps; According to the trace parameters that sets, thereby begin the tracking part of following the trail of the fibrous bundle that obtains region of interest from positive and negative both direction along the proper vector of each voxel of region of interest from the Seed Points of region of interest; The storage part of the coordinate of the point in the number of the fibrous bundle that tracks of storage and every the fibrous bundle; The be connected display part of the fibrous bundle that forms of the intrafascicular point of display fibers.

Further, diffusion tensor imaging fibrous bundle follow-up mechanism provided by the invention also has such feature: tracking part represents to follow the trail of the point on the path of described fibrous bundle with the partial differential equation integration,

Point P _nWith adjacent iteration point P _N+1As follows in the relational expression of following the trail of on the path:

P _n+1=P _n+S·V _Pn （n=0,1,2…）

In this formula, V _PnExpression point P _nProper vector;

Thereafter iteration point P _N+1Proper vector with the some P _nThe proper vector at place has following relation, is expressed as with following formula:

$V_p=\left\{\begin{array}{cc}{V}_t& \mathrm{if}\cos <V_t\&CenterDot;V_i>>{\mathrm{\&theta;}}_{\mathrm{threshod}},V_t=V_s\\ -V_t& \mathrm{if}\cos <V_t\&CenterDot;V_i>>-{\mathrm{\&theta;}}_{\mathrm{threshod}},V_t=V_s\end{array}\right.$

In this formula, V _sBe a P _N+1Tensor resolution before direction, V _iBe a P _nDirection, θ _ThreshodBe described tracking angle threshold, if the tracking angle that calculates in the tracing process, is then followed the trail of direction for negative 180 degree counter-rotatings occur;

S is described iteration step length, and the definition of S is relevant with described anisotropy threshold value, and described anisotropy threshold value comprises: linear anisotropic coefficient C _LExpression, in-plane anisotropy coefficient C _PExpression, isotropy coefficient C _SExpression is formulated as follows:

C _L＝(λ ₁-λ ₂)/(λ ₁+λ ₂+λ ₃)

C _P=2(λ ₂-λ ₃)/(λ ₁+λ ₂+λ ₃)

C _S=3λ ₃/(λ ₁+λ ₂+λ ₃)

C wherein _L+ C _P+ C _S=1, λ in the following formula ₁, λ ₂, λ ₃Represent each described voxel three eigenwerts from big to small;

The expression formula of iteration step length S is:

$S=\frac{1}{2*\max \left\{\right|V_{\mathrm{Pn}}\left(0\right)|,|V_{\mathrm{Pn}}\left(1\right)|,|V_{\mathrm{Pn}}\left(2\right)\left|\right\}}*\left\{\begin{array}{cc}1& \mathrm{if}{C}_L>C_P\&GreaterEqual;C_S\\ 1/2& \mathrm{if}{C}_P\&GreaterEqual;C_L>C_S\\ 0& \mathrm{if}{C}_S\&GreaterEqual;C_P>C_L\end{array}\right.$

Further, diffusion tensor imaging fibrous bundle follow-up mechanism provided by the invention also has such feature: acquisition matrix is greater than 256 * 256.

Further, diffusion tensor imaging fibrous bundle follow-up mechanism provided by the invention also has such feature: three-dimensional affine transformation be with mutual information as similarity measure, use quasi-Newton method to realize.

Further, diffusion tensor imaging fibrous bundle follow-up mechanism provided by the invention also has such feature: the interpolation ratio of interpolation is 3:1, also can be 4:1.

Further, diffusion tensor imaging fibrous bundle follow-up mechanism provided by the invention also has such feature: region of interest can for one also can be for a plurality of.

Invention effect and effect

Diffusion tensor imaging fibrous bundle follow-up mechanism provided by the invention can combine with various brain function image capture devices, can be the brain diffusion tensor imaging and provide accurate, tracing algorithm fast, for the cognition of cranial nerve fibrous bundle and the surgical guidance of carrying out subsequently provides reliable information, the impact that it has overcome the original image differences in spatial location and has followed the trail of step-length error in the iterative process, thereby obtain more accurately brain white matter fiber tract, be clinical diagnosis, radiotherapy localization, cerebral surgery operation and therapeutic evaluation improve comprehensively reliably information, a kind of reliable, accurate diffusion tensor imaging fibrous bundle follow-up mechanism is brought into play significant role for it and is laid the first stone in clinic diagnosis.

Description of drawings

Fig. 1 is the workflow diagram of the present invention's diffusion tensor imaging fibrous bundle follow-up mechanism in an embodiment;

Fig. 2 is the present invention's benchmark Brain Tissues Image in an embodiment;

Fig. 3 be the present invention adopt in an embodiment treat remedial frames;

Fig. 4 is the in an embodiment corpus callosum that forms of the shown fibrous bundle that tracks of display of the present invention.

Embodiment

The diffusion tensor imaging fibrous bundle follow-up mechanism that a kind of result who people's brain scanning is formed according to the superconducting magnetic resonance imaging device carries out the tracking of diffusion tensor imaging fibrous bundle passes through:

Collection section gathers different sequence diffusion tensor imaging images on 12 directions of human brain by acquisition matrix from the superconducting magnetic resonance imaging device, get any one sequence in the different sequence diffusion tensor imaging images as the reference image, in order to ensure precision, the sampling matrix of setting does not need to do figure image intensifying and image denoising greater than 256 * 256 in this process.

Normalized section carries out normalized so that the pixel size of all sequences diffusion tensor imaging image and physical dimension are all consistent to different sequence diffusion tensor imaging images, when setting the gray threshold of brain tissue and non-brain tissue (gray scale normalization is 0-1), select 0.1-0.4.

Cutting part adopts brain tissue extracting tool BET(Brain Extraction Tool) thus the reference image extracted cut apart the benchmark Brain Tissues Image that obtains the corresponding sequence that does not apply disperse magnetic field and the Brain Tissues Image to be corrected that applies the sequence in diffusion gradient magnetic field, in the leaching process, surperficial by distorted pattern evolution match brain tissue under the effect of local auto-adaptive model power, estimate the threshold value of brain tissue and non-brain tissue by grey level histogram, come the barycenter and the approximate size that obtains brain tissue in the image of computed image according to threshold value, then the spherome surface of brain tissue inner setting is initialized as triangle gridding, arbitrary moment only has the slow mobile distortion that realizes of a vertex position of triangle gridding, iteration moves to the brain tissue surface under the effect of self-adaptation power, and guarantee accurately with smooth, if do not reach optimum solution, then change level and smooth restriction and continue iteration, until meet the demands, estimate the brain tissue surface with this, thereby obtain the benchmark Brain Tissues Image and the Brain Tissues Image to be corrected that applies the sequence in diffusion gradient magnetic field of the corresponding sequence that does not apply disperse magnetic field, this benchmark Brain Tissues Image can be used as brain tissue standard masterplate.

With reference to image rectification section with the benchmark Brain Tissues Image as brain tissue standard masterplate, origin is (0,0,0), pixel size and interval are constant, with mutual information for estimating, adopt the affined transformation of 12 parameters, thereby obtain with reference to the benchmark Brain Tissues Image of image and the spatial alternation matrix relationship of Brain Tissues Image to be corrected, after the conversion of spatial alternation matrix relationship, it is maximum that the similarity measure of above-mentioned two images reaches, thereby realize reaching corresponding one by one with reference to the pixel of the benchmark Brain Tissues Image of image in the locus with the pixel of Brain Tissues Image to be corrected.

Image registration correction section is except forming the corresponding remedial frames for the treatment of with reference to all sequences diffusion tensor imaging image image, thereby concerns according to spatial alternation and to treat that remedial frames carries out three-dimensional affine transformation so that the pixel of all sequence diffusion tensor imaging images reaches one by one correspondence in the locus.

Calculating part goes out the tensor field of all sequences diffusion tensor imaging image and the disperse information of each voxel of all sequences diffusion tensor imaging image is represented with the tensor of 3 * 3 matrixes by matrix computations, thereby each voxel is carried out to characteristic value and the characteristic vector that matrix decomposition obtains each voxel, the computational process of calculating part is as follows: at first the apparent disperse factor (ADC) is calculated: employing does not apply the diffusion gradient magnetic direction and forms the dispersion tensor data fields with the image that applies the diffusion gradient magnetic direction

ADC=log(DWI _g/DWI _g0)/-b _value

Wherein g represents the direction of bipolarity gradient axes, DWI _gRepresent disperse signal measured when the bipolarity gradient applies, DWI _G0Expression does not apply the measured disperse signal of bipolarity gradient, b _ValueWhat represent is the intensity of bipolarity gradient pulse;

Then carry out the assembling of gradient matrix H:

$H=\left[\begin{array}{c}{h}_{g1}\\ .\\ .\\ .\\ h_{\mathrm{gn}}\end{array}\right]$

$h_g=(g_x^2,g_y^2,g_z^2,{2g}_x{g}_y,{2g}_x{g}_z,{2g}_y{g}_z)$

Wherein g represents the direction of bipolarity gradient axes, g _xThe component of expression x direction, g _yThe component of expression y direction, g _zThe component of expression z direction; h _G1The gradient component that represents the 1st direction, h _GnThe gradient component that represents n direction, its gradient axes component of different MR equipment is different;

Then the ADC matrix S is assembled:

$S=\left[\begin{array}{c}{\mathrm{ADC}}_{g1}\\ .\\ .\\ .\\ {\mathrm{ADC}}_{\mathrm{gn}}\end{array}\right]$

In this formula, ADC _G1Be illustrated in the apparent diffusion coefficient of the 1st direction, ADC _GnRepresent n the apparent diffusion coefficient on the direction;

Then degree of freedom factor d is calculated again:

d=(H ^TH) ^-1H ^TS

Wherein, d=(D _Xx, D _Yy, D _Zz, D _Xy, D _Xz, D _Yz) ^T

Thereby formation tensor:

$D=\left[\begin{array}{ccc}{D}_{\mathrm{xx}}& D_{\mathrm{xy}}& D_{\mathrm{xz}}\\ D_{\mathrm{xy}}& D_{\mathrm{yy}}& D_{\mathrm{yz}}\\ D_{\mathrm{xz}}& D_{\mathrm{yz}}& D_{\mathrm{zz}}\end{array}\right]$

By matrix decomposition, obtain eigenwert and the proper vector of each voxel at last.

The region of interest selection portion arranges module need to select tracking at brain tissue standard masterplate region of interest by region of interest, thereby the Seed Points of the region of interest that the even interpolation of region of interest is obtained, this process is for using ITK-SNAP software (deriving from) to select region of interest, and the number of region of interest is one or more.Read in standard brain tissue standard picture (storage format is * .VTK), in the transversal section, coronal-plane, sagittal plane use polygon tool to delineate the zone, delineate complete rear accept and delineate the zone, and update mesh, store at last the GIPL form, the interpolation ratio can be chosen as 3:1 or 4:1, just obtains dense Seed Points after the interpolation.

Trace parameters arranges the trace parameters that section can arrange needed trace parameters when following the trail of section is set:

1. brain tissue anisotropy threshold value: want to carry out fibrous bundle in specific zone and follow the trail of, represented tensor must be anisotropic, must satisfy minimum anisotropy value during tracking, not satisfying the zone of setting minimum value, then thinks and can't follow the trail of;

2. follow the trail of angle threshold: normal nerve fibre angle in adjacent voxel is very little, and the white matter in the human brain is relatively straight, if angle surpasses setting value in tracing process, must stop to follow the trail of;

3. iterative steps: set maximum and minimum iterative steps threshold value in the tracing process.Iterative steps is larger during tracking, and fibrous bundle is longer; Vice versa.Short fibrous bundle does not then store and demonstration as not satisfying the shortest iteration threshold of setting; If fibrous bundle is oversize, just think excessive tracking, do not store and demonstration yet.

Tracking part is according to the trace parameters that sets, thereby begins to follow the trail of the fibrous bundle that obtains region of interest from positive and negative both direction along the proper vector of each voxel of region of interest from the Seed Points of region of interest.

Adopt the Linear tracing method to adopt the tensor field short-cut method in the present embodiment, tensor field can be reduced to a vector field, and each tensor replaces with its maximal eigenvector, realizes the disperse Information Simplification.In the linear anisotropic zone, main proper vector has defined the direction of linear structure.The extraction of fibrous bundle is exactly to extract data to carry out the direction indication of fiber from tensor field, and has represented the main direction of disperse.Fibrous bundle extracts and at first selects Seed Points, then follows the trail of main proper vector, just can obtain the fibrous bundle in region of interest.By being carried out integration, a partial differential equation carries out the expression of fibrous bundle.Follow the trail of basic process for from each Seed Points in the region of interest, along the proper vector (namely following the trail of direction) of each voxel, follow the tracking criterion that sets, according to certain step-length, progressively stack is followed the trail of from positive and negative both direction.

During tracking initial point is carried out the integration of partial differential equation, suppose P ₀Initial point, some P _nAnd P _N+1Be formulated as follows in the path:

P _n+1=P _n+S·V _pn （n=0,1,2…）

Point P ₀Expression initial point, then V _PnExpression point P _nThe proper vector at place, iteration point P _N+1Proper vector with the some P _nThe main proper vector at place has following relation, is expressed as with following formula:

$V_p=\left\{\begin{array}{cc}{V}_t& \mathrm{if}\cos <V_t\&CenterDot;V_i>>{\mathrm{\&theta;}}_{\mathrm{threshod}},V_t=V_s\\ -V_t& \mathrm{if}\cos <V_t\&CenterDot;V_i>>-{\mathrm{\&theta;}}_{\mathrm{threshod}},V_t=V_s\end{array}\right.$

In this formula, V _sBe a P _N+1Tensor resolution before direction, V _iBe a P _nDirection, θ _ThreshodBe described tracking angle threshold, if the tracking angle that calculates in the tracing process, is then followed the trail of direction for negative 180 degree counter-rotatings occur;

S is described iteration step length, and the definition of S is relevant with described anisotropy threshold value, and described anisotropy threshold value comprises: linear anisotropic coefficient C _LExpression, in-plane anisotropy coefficient C _PExpression, isotropy coefficient C _SExpression is formulated as follows:

$C_L=\frac{{\mathrm{\&lambda;}}_1-{\mathrm{\&lambda;}}_2}{{\mathrm{\&lambda;}}_1+{\mathrm{\&lambda;}}_2+{\mathrm{\&lambda;}}_3}$

$C_P=\frac{2({\mathrm{\&lambda;}}_2-{\mathrm{\&lambda;}}_3)}{{\mathrm{\&lambda;}}_1+{\mathrm{\&lambda;}}_2+{\mathrm{\&lambda;}}_3}$

$C_S=\frac{{3\mathrm{\&lambda;}}_3}{{\mathrm{\&lambda;}}_1+{\mathrm{\&lambda;}}_2+{\mathrm{\&lambda;}}_3}$

C wherein _L+ C _P+ C _S=1, λ in the following formula ₁, λ ₂, λ ₃Represent each described voxel three eigenwerts from big to small;

The expression formula of iteration step length S is:

$S=\frac{1}{2*\max \left\{\right|V_{\mathrm{Pn}}\left(0\right)|,|V_{\mathrm{Pn}}\left(1\right)|,|V_{\mathrm{Pn}}\left(2\right)\left|\right\}}*\left\{\begin{array}{cc}1& \mathrm{if}{C}_L>C_P\&GreaterEqual;C_S\\ 1/2& \mathrm{if}{C}_P\&GreaterEqual;C_L>C_S\\ 0& \mathrm{if}{C}_S\&GreaterEqual;C_P>C_L\end{array}\right.$

Simultaneously as seen, the iteration step length in this tracing process is variable, has good robustness, can reduce the error in the tracing process;

The PRELIMINARY RESULTS of following the trail of is a point set, represent the point set on the path that fibrous bundle follows the trail of, comprising fibrous bundle number and contained the counting thereof of following the trail of, each its three-dimensional point coordinate of some storage that point is concentrated, the point of every fibrous bundle is used for connecting, then having consisted of needs the fibrous bundle that shows, the coordinate of the point in the number of this fibrous bundle that tracks of storage portion stores and every the fibrous bundle, and display section shows the corpus callosum fibrous bundle that point in these fibrous bundles is connected and forms.

Diffusion tensor imaging fibrous bundle follow-up mechanism and various brain function image capture device combine, and normalized section has avoided the position difference that causes because of the difference of brain function image capture device.

Fig. 2 is the benchmark Brain Tissues Image in the present embodiment; Fig. 3 is the remedial frames for the treatment of that adopts in the present embodiment; Fig. 4 is the corpus callosum that the shown fibrous bundle that tracks of display forms in the present embodiment.Such as Fig. 2,3, shown in 4, in a routine Healthy People encephalopathic people magnetic resonance examination, use the diffusion tensor imaging fibrous bundle follow-up mechanism in the present embodiment to carry out the fibrous bundle tracking, here adopt and do not apply first sequence in diffusion gradient magnetic field as the benchmark Brain Tissues Image, arbitrary sequence that applies diffusion gradient magnetic field is Brain Tissues Image sequence to be corrected, obtain the optimal spatial transformational relation, afterwards all sequences being carried out the locus corrects, and the corpus callosum fibrous bundle of human brain followed the trail of and show, wherein region of interest is selected the central area, joint portion in two brains, the parameter of selecting is: the background threshold of tracking is 15, and maximum step number is 4000, and minimum step number is 100, the anisotropy threshold value is 0.5, angle threshold is 0.7, and maximum fibre number is 10000, and calculating the b value is-1000.Follow the trail of through fibrous bundle, obtain following the trail of the result, can show well mouth, knee, dried, the splenium of corpus callosum fibrous bundle.

Embodiment effect and effect

The diffusion tensor imaging fibrous bundle follow-up mechanism that present embodiment provides and various brain function image capture device combine, can be the brain diffusion tensor imaging and provide accurate, tracing algorithm fast, for the cognition of cranial nerve fibrous bundle and the surgical guidance of carrying out subsequently provides reliable information, the impact that it has overcome the original image differences in spatial location and has followed the trail of step-length error in the iterative process, thereby obtain more accurately brain white matter fiber tract, be clinical diagnosis, radiotherapy localization, cerebral surgery operation and therapeutic evaluation improve comprehensively reliably information, and its precision has obvious advantage than conventional art.

Claims (8)

1. a result who the scanning of human brain is formed according to the brain function image capture device carries out the diffusion tensor imaging fibrous bundle follow-up mechanism that the diffusion tensor imaging fibrous bundle is followed the trail of, and it is characterized in that, comprising:

Collection section, from described brain function image capture device, gather different sequence diffusion tensor imaging images at least ten two directions of described human brain by acquisition matrix, get any one sequence in the described different sequence diffusion tensor imaging image as the reference image;

Normalized section carries out normalized so that the pixel size of all described sequence diffusion tensor imaging images and physical dimension are all consistent to different described sequence diffusion tensor imaging images;

Cutting part, thereby described the extraction with reference to image cut apart the Brain Tissues Image to be corrected that obtains the corresponding benchmark Brain Tissues Image that does not apply the disperse field pulses and apply the diffusion gradient field pulses, and described benchmark Brain Tissues Image is as brain tissue standard masterplate;

With reference to image rectification section, the pixel that goes out the spatial alternation relation of described described benchmark Brain Tissues Image with reference to image and Brain Tissues Image described to be corrected and realize described described benchmark Brain Tissues Image with reference to image with described benchmark Brain Tissues Image as brain tissue standard masterplate according to the spatial alternation matrix computations reaches corresponding one by one with the pixel of Brain Tissues Image described to be corrected in the locus;

Image registration correction section, except forming the corresponding remedial frames for the treatment of with reference to all the described sequence diffusion tensor imaging images the image, thereby treat that to described remedial frames carries out three-dimensional affine transformation so that the pixel of all described sequence diffusion tensor imaging images reaches corresponding one by one in the locus according to described spatial alternation relation;

Calculating part, go out the tensor field of all described sequence diffusion tensor imaging images and the disperse information of each voxel of all the described sequence diffusion tensor imaging images tensor with 3 * 3 matrixes is represented by matrix computations, thereby each described voxel is carried out eigenwert and the proper vector that matrix decomposition obtains each described voxel.

The region of interest selection portion arranges module by region of interest and need to select the region of interest of following the trail of at described brain tissue standard masterplate, thus the Seed Points of the described region of interest that the even interpolation of described region of interest is obtained;

Trace parameters arranges section, and needed trace parameters when following the trail of is set, and described trace parameters has brain tissue anisotropy threshold value, follows the trail of angle threshold, iterative steps;

Tracking part according to the described trace parameters that sets, thereby begins to follow the trail of the fibrous bundle that obtains described region of interest from positive and negative both direction along the described proper vector of each described voxel of described region of interest from the described Seed Points of described region of interest;

Storage part, the coordinate of the point in the number of the described fibrous bundle that tracks of storage and every the described fibrous bundle;

Display part shows the described fibrous bundle that the point in the described fibrous bundle is connected and forms.

2. diffusion tensor imaging fibrous bundle follow-up mechanism according to claim 1 is characterized in that:

Described tracking part represents to follow the trail of the point on the path of described fibrous bundle, some P with the partial differential equation integration _nWith adjacent iteration point P _N+1As follows in the relational expression of following the trail of on the path:

P _n+1=P _n+S·V _pn （n=0,1,2…）

In this formula, V _PnExpression point P _nProper vector;

Iteration point P _N+1Proper vector with the some P _nThe proper vector at place has following relation, is expressed as with following formula:

$V_p=\left\{\begin{array}{cc}{V}_t& \mathrm{if}\cos <V_t\&CenterDot;V_i>>{\mathrm{\&theta;}}_{\mathrm{threshod}},V_t=V_s\\ -V_t& \mathrm{if}\cos <V_t\&CenterDot;V_i>>-{\mathrm{\&theta;}}_{\mathrm{threshod}},V_t=V_s\end{array}\right.$

In this formula, V _sBe a P _N+1Direction, V _iBe a P _nDirection, θ _ThreshodBe described tracking angle threshold, if the tracking angle that calculates in the tracing process, is then followed the trail of direction for negative 180 degree counter-rotatings occur;

S is described iteration step length, and the definition of S is relevant with described anisotropy threshold value, and described anisotropy threshold value comprises: linear anisotropic coefficient C _LExpression, in-plane anisotropy coefficient C _PExpression, isotropy coefficient C _SExpression is formulated as follows:

$C_L=\frac{{\mathrm{\&lambda;}}_1-{\mathrm{\&lambda;}}_2}{{\mathrm{\&lambda;}}_1+{\mathrm{\&lambda;}}_2+{\mathrm{\&lambda;}}_3}$

$C_P=\frac{2({\mathrm{\&lambda;}}_2-{\mathrm{\&lambda;}}_3)}{{\mathrm{\&lambda;}}_1+{\mathrm{\&lambda;}}_2+{\mathrm{\&lambda;}}_3}$

$C_S=\frac{{3\mathrm{\&lambda;}}_3}{{\mathrm{\&lambda;}}_1+{\mathrm{\&lambda;}}_2+{\mathrm{\&lambda;}}_3}$

C wherein _L+ C _P+ C _S=1, λ in the following formula ₁, λ ₂, λ ₃Represent each described voxel three eigenwerts from big to small;

The expression formula of iteration step length S is:

$S=\frac{1}{2*\max \left\{\right|V_{\mathrm{Pn}}\left(0\right)|,|V_{\mathrm{Pn}}\left(1\right)|,|V_{\mathrm{Pn}}\left(2\right)\left|\right\}}*\left\{\begin{array}{cc}1& \mathrm{if}{C}_L>C_P\&GreaterEqual;C_S\\ 1/2& \mathrm{if}{C}_P\&GreaterEqual;C_L>C_S\\ 0& \mathrm{if}{C}_S\&GreaterEqual;C_P>C_L\end{array}\right.$

3. diffusion tensor imaging fibrous bundle follow-up mechanism according to claim 1, it is characterized in that: wherein, described acquisition matrix is greater than 256 * 256.

4. diffusion tensor imaging fibrous bundle follow-up mechanism according to claim 1 is characterized in that: described three-dimensional affine transformation be with mutual information as similarity measure, use quasi-Newton method to realize.

5. diffusion tensor imaging fibrous bundle follow-up mechanism according to claim 1, it is characterized in that: wherein, the interpolation ratio of described interpolation is 3:1.

6. diffusion tensor imaging fibrous bundle follow-up mechanism according to claim 1, it is characterized in that: wherein, the interpolation ratio of described interpolation is 4:1.

7. diffusion tensor imaging fibrous bundle follow-up mechanism according to claim 1, it is characterized in that: wherein, described region of interest is one.

8. diffusion tensor imaging fibrous bundle follow-up mechanism according to claim 1, it is characterized in that: wherein, described region of interest is a plurality of.

Images