CN103985123A - Abdominal aortic aneurysm outer boundary segmentation method based on CTA images - Google Patents

Abstract

The invention discloses an abdominal aortic aneurysm outer boundary segmentation method based on CTA images. The method includes the following steps that a CTA image sequence is obtained; an aorta blood vessel inner cavity is extracted from the CTA image sequence; according to the first geometrical characteristic of the extracted blood vessel inner cavity, aortic aneurysm laminas are positioned; a segmentation result of a previous layer of image servers as an initial boundary for segmentation on the current image, and all the aortic aneurysm laminas obtained through positioning are segmented though a CV model to extract an aortic aneurysm outer boundary. The method can be used for rapidly and accurately obtaining the aortic aneurysm outer boundary from the CTA image sequence, so that working efficiency of a doctor is improved, workloads are reduced, and cost is lowered.

Description

Abdominal aneurvsm outer boundary dividing method based on CTA image

Technical field

The invention belongs to computer vision and image processing field, be specifically related to a kind of abdominal aneurvsm outer boundary dividing method based on CTA image.

Background technology

Abdominal aneurvsm is the expansion of degenerating of abdominal aorta limitation, finally cannot bear that blood flow impacts and a kind of high-risk property disease that causes knurl wall to break.It is generally acknowledged that local abdominal aorta is expanded to diameter and increases more than 50%, can be diagnosed as abdominal aneurvsm.The incidence of disease of abdominal aneurvsm is along with the age constantly increases.In clinical examination, the CT image (contrast CT angiography, CTA) that radiography strengthens is formation method common in aneurysm imaging and Measurement accuracy.CTA provides aortal detailed anatomic information, can realize the visual of intravascular space, calcification part and thrombus.

The abdominal aortic aneurysms that carries out blood vessel intervention repairing and treating need to continue abdominal aneurvsm to carry out imaging, to find in time aneurysmal PD.An important monitor control index is exactly capsule sex index.Than the capsule sex index of calculating with diameter more accurately and reliably, yet this needs, doctor is manual successively to be delineated aneurysm the capsule sex index of calculating with volume ratio.

To the accurate quantification of blood vessel structure parameter for the diagnosis of disease and treatment important in inhibiting.By the measurement of abdominal aneurvsm diameter volume is calculated, the danger that can break to it is assessed, and the danger that the variation of operative treatment front and back abdominal aneurvsm diameter and volume has been broken after having shown.In addition, for the feasibility of internal blood vessel reparation and select suitable cantilever type and size, aneurysmal diameter, volume and knurl body temporal evolution parameter are important foundations.The shape of interventional therapy support and parameter be take CTA and MRA and are checked to be foundation.The parameter of preoperative assessment comprises: abdominal aneurvsm is far away, near-end knurl neck length degree and diameter, relation with important internal organ branch vessel, the thickness of tube wall and calcification situation, the type of abdominal aneurvsm, size, the situation of graft lead-in path (narrow and degreeof tortuosity).Yet the parameter measurement of two dimension is inaccurate, and method is to carry out three-dimensional parameter measurement more accurately.The reconstructing three-dimensional model of abdominal aneurvsm needs a large amount of man-machine interactivelies.In practical application, most aortic aneurysm data are by observing each section and manually irising out area-of-interest and obtain, and manual method is very consuming time, and along with the increase of data volume, it becomes more and more unactual.

Utilize the method for Computer Image Processing, extract abdominal aneurvsm in CT image sequence, the geometric configuration and the size that calculate abdominal aneurvsm are the important evidence of prediction knurl risk of rupture.Comprise the cutting apart of abdominal aneurvsm that inner chamber and outside thrombus cut apart two parts, adopt the method efficiency of cutting apart manually low in the past, can not meet the needs of clinical treatment.In order to reduce analysis time and to increase repeatability, need to adopt auto Segmentation or semi-automatic cutting apart.The outer boundary of abdominal aneurvsm, due to obscurity boundary, is cut apart difficulty larger.Because aneurysm is close with surrounding tissue gray scale, the edge of thrombus is very fuzzy, and has the tissue and the abdominal aneurvsm outer wall that much close on quite approaching, makes aneurysmal border be difficult to extract.And between different patients, identical patient is in different, aneurysmal size shape has very large difference, and same patient is in aneurysm CTA imaging, and the size shape of the different synusia of aneurysm has a long way to go.Aneurysm is cut apart simple threshold method cannot correctly distinguish thrombus and surrounding tissue, cannot distinguish shade of gray, for the thickness of tube wall and the performance of calcification situation, cannot accurately show.

Summary of the invention

The object of this invention is to provide a kind of abdominal aneurvsm outer boundary dividing method based on CTA image, to solve conventional images dividing method, cannot accurately extract the technical matters of aneurysm outer boundary.

Concrete technical scheme of the present invention is:

An aortic aneurysm outer boundary dividing method based on CTA image, it comprises the following steps:

Obtain CTA image sequence;

From CTA image sequence, extract aortic blood tube cavity;

According to the first geometric properties location aortic aneurysm synusia of extracted intravascular space;

The segmentation result of a tomographic image was as the initial boundary that present image is cut apart in the past, and each aortic aneurysm synusia automatically location being obtained by CV model is cut apart, and extracted aortic aneurysm outer boundary.

In the above-mentioned aortic aneurysm outer boundary dividing method based on CTA image, preferably, the method of extracting aortic blood tube cavity from CTA image sequence comprises: according to the CT value setting threshold of CTA image angiosomes, utilize this threshold value to cut apart every tomographic image of CTA image sequence, obtain the 2 dimensional region of similar blood vessel; Calculate the second geometric properties of described 2 dimensional region, with setting value comparison, exclude part 2 dimensional region; According to interlayer continuity, remaining 2 dimensional region is carried out to interlayer merging, obtain at least one 3D region; And the 3D region that is communicated with Seed Points of extraction, as aortic blood tube cavity.

In the above-mentioned aortic aneurysm outer boundary dividing method based on CTA image, preferably, described the second geometric properties comprises one or more in area, girth, average, variance, circularity and excentricity.

In the above-mentioned aortic aneurysm outer boundary dividing method based on CTA image, preferably, described threshold value comprises first threshold θ _hwith Second Threshold θ _l, utilize the method that this threshold value cuts apart to every tomographic image of CTA image sequence the 2 dimensional region that obtains similar blood vessel to comprise: from CTA image sequence, to read a tomographic image; The pixel gray-scale value of lining by line scan, and be labeled as F (x, y), marking convention is and same a line internal labeling is worth to identical pixel merges into a set, the row labels of going forward side by side; The pixel point set that the line style of lining by line scan is arranged, when the pixel point set locus 8 between adjacent lines in abutting connection with time, the pixel collection that line style is arranged is merged into 2 dimensional region.

In the above-mentioned aortic aneurysm outer boundary dividing method based on CTA image, preferably, described the first geometric properties comprises that the minimax footpath of each layer of extracted intravascular space distributes and area value distributes.

In the above-mentioned aortic aneurysm outer boundary dividing method based on CTA image, preferably, described CV model comprises shape constraining energy, and this shape constraining energy definition is: $E_{\mathrm{shape}}\left(X_i\right)=\left\{\begin{array}{c}d(X_i,C_i)\mathrm{if}{X}_i\∈\mathrm{outside}\left(c\right)\\ -d(X_i,C_i)\mathrm{if}{X}_i\∈\mathrm{inside}\left(c\right)\end{array}\right.,$ Wherein, d (X _i, C _i) an expression point X _idistance along boundary method vector to border, outside (c) represents the region beyond border, inside (c) represents Yi Nei region, border.

In the above-mentioned aortic aneurysm outer boundary dividing method based on CTA image, preferably, described CV model comprises border indicator function $g(x,y)=\frac{1}{1+\mathrm{\λ}|\▿G*I(x,y)|},$ Wherein, for the partial gradient value after gaussian filtering, the weighting function that described border indicator function is controlled as boundary shape, in image, the value of sharpness of border place weighting function is less, larger in the value of obscurity boundary place weighting function.

In the above-mentioned aortic aneurysm outer boundary dividing method based on CTA image, preferably, the level set function value at the reference mark in the arrowband that a CV model iterative computation is defined by initialization area, arrowband is defined as the curve of S=± δ/2, the inside and outside gray average in border is the inside and outside gray average in border in narrowband region, and δ represents level set function to be normalized the threshold value of choosing after calculating.

In the above-mentioned aortic aneurysm outer boundary dividing method based on CTA image, preferably, this aortic aneurysm outer boundary dividing method also comprises the step of calculating the thickness of aortic aneurysm in radioactive ray mode.

The inventive method can obtain aortic aneurysm outer boundary from CTA image sequence quickly and accurately, thereby can improve doctor's work efficiency, alleviates workload, reduces costs.

Accompanying drawing explanation

Fig. 1 is the process flow diagram of the aortic aneurysm outer boundary dividing method of an embodiment based on CTA image;

Fig. 2 is for wherein extracting the process flow diagram of aortic blood tube cavity;

Fig. 3 is the design sketch of some aortic blood tube cavities of being divided in experiment;

Fig. 4 is the design sketch of some aortic aneurysm outer boundaries of being divided in experiment.

Embodiment

Below in conjunction with drawings and Examples, the present invention is further described.These more detailed descriptions are intended to help to understand the present invention, and should not be used to limit the present invention.According to content disclosed by the invention, it will be understood by those skilled in the art that and can not need some or all these specific detail can implement the present invention.And in other cases, for fear of by innovation and creation desalination, do not describe well-known operating process in detail.

As shown in Figure 1, this aortic aneurysm outer boundary dividing method based on CTA image comprises the following steps:

Step S1, obtains CTA image sequence.Patient passes through the mode injection of contrast medium of vein bolus, scanning computed tomography A view data, and image is DICOM form.

Step S2, initialization.Adopt anisotropic diffusion filtering mode to carry out filtering processing to image, manually draw a circle to approve the thrombus outer boundary region of the first width CTA image, and select in inner chamber a bit as Seed Points.

Step S3, extracts aortic blood tube cavity from CTA image sequence.Can adopt quick symmetrical region growth algorithm to extract intravascular space.In view of the part angiosomes after CT contrast imaging and bony areas very approaching in CT value, in preferred embodiment, one of the feature using the geometric properties in region in CT cross-sectional image as classification, adds in the discrimination standard of symmetrical region growth.Fig. 2 shows a kind of flow process of extracting aortic blood tube cavity.As shown in Figure 2, this method from CTA image sequence extraction aortic blood tube cavity comprises the following steps:

Step S31, reads a tomographic image from CTA image sequence.The pixel gray-scale value of lining by line scan, and be labeled as F (x, y), marking convention is as follows:

the gray-scale value that wherein G (x, y) is pixel, θ _hfor first threshold, θ _lfor Second Threshold, θ _hand θ _laccording to the CT value of CTA image angiosomes, set.And same a line internal labeling is worth to identical pixel merges into a set, the row labels of going forward side by side.

Step S32, the pixel point set that the line style of lining by line scan is arranged, when the pixel point set locus 8 between adjacent lines in abutting connection with time, the pixel collection that line style is arranged is merged into 2 dimensional region.

By above-mentioned steps S31 to step S32, can be at the 2 dimensional region of every layer of similar blood vessel of CTA Image Acquisition.

Step S33, gray feature is differentiated, and excludes the 2 dimensional region that part does not belong to blood vessel.Concrete grammar is gray average and the variance of zoning, with setting value comparison, excludes the 2 dimensional region that part does not belong to blood vessel.

Step S34, geometric properties is differentiated.Calculate the second geometric properties of 2 dimensional region, with setting value comparison, exclude the 2 dimensional region that part does not belong to blood vessel.The second geometric properties can be one or more in area, girth, average, variance, circularity and excentricity.Due in CT cross-sectional image, all there is very large difference in the geometric properties such as the circularity in the area and perimeter in the region of bony areas and angiosomes, the average in region and variance, region, excentricity, so can exclude the bony areas close with angiosomes CT value by this step.

Step S35, carries out interlayer merging according to interlayer continuity by remaining 2 dimensional region, obtains at least one 3D region.

Step S36, connective differentiation, extracts the 3D region being communicated with Seed Points, as aortic blood tube cavity.

Step S4, according to the first geometric properties location aortic aneurysm synusia of extracted intravascular space.The first geometric properties is minimax footpath and the area distributions of inner chamber.

Step S5, the segmentation result of a former tomographic image is as the initial boundary that present image is cut apart, and each aortic aneurysm synusia automatically location being obtained by CV model is cut apart, and extracts aortic aneurysm outer boundary.

In a kind of preferred embodiment, the CV model that adopts arrowband to control, its energy functional is:

$\begin{array}{c}E\left(C\right)=\\ {\mathrm{\λ}}_1{\∫}_{\mathrm{inside}\left(C\right)}{|u_0(x,y)-c_1|}^2\mathrm{dxdy}+{\mathrm{\λ}}_2{\∫}_{\mathrm{outside}\left(C\right)}{|u_0(x,y)-c_2|}^2\mathrm{dxdy}\\ +\mathrm{\μ}\·\mathrm{Length}\left(C\right)+\mathrm{\ν}\·\mathrm{Area}\left(\mathrm{inside}\left(C\right)\right)-\mathrm{\ω}\·{\∫E}_{\mathrm{shape}}(x,y)\mathrm{dxdy}\end{array}$

Wherein, E _shape(x, y) represents the shape constraining energy of current location, and shape constraining energy is determined by the dissection priori about aneurysm outer boundary.We by shape constraining energy definition are:

$E_{\mathrm{shape}}\left(X_i\right)=\left\{\begin{array}{c}d(X_i,C_i)\mathrm{if}{X}_i\∈\mathrm{outside}\left(c\right)\\ -d(X_i,C_i)\mathrm{if}{X}_i\∈\mathrm{inside}\left(c\right)\end{array}\right.$

Wherein, d (X _i, C _i) an expression point X _idistance along boundary method vector to border, outside (c) represents the region beyond border, inside (c) represents Yi Nei region, border.Shape constraining energy, as a part for image constraint external force, points to initialization border.

Further, for fear of boundary leakage in evolutionary process to adjacent tissue around, a border indicator function has been proposed in preferred embodiment:

$g(x,y)=\frac{1}{1+\mathrm{\λ}|\▿G*I(x,y)|}$

The weighting function of controlling as boundary shape.Wherein, for the partial gradient value after gaussian filtering, in image, the value of sharpness of border place weighting function is less, larger in the value of obscurity boundary place weighting function, thereby strengthens the effect of shape control performance, make border rest on preset shape near, avoided the leakage of border to adjacent tissue.

In order to make iterative process more efficient, in preferred embodiment, the level set function value at the reference mark in the arrowband that iterative computation is defined by initialization area, arrowband is defined as the curve of S=± δ/2, the threshold value of choosing after δ represents the level set function to be normalized, thus greatly reduce calculative data volume.For avoiding the impact of narrowband region external pixels on segmentation result, when the inside and outside gray average of computation bound, only calculate in narrowband region the gray average that border is inside and outside.

By above-mentioned steps S1, to step S5, can from CTA image, extract aortic aneurysm outer boundary quickly and accurately.

Workload for further minimizing personnel, further comprises step S6, calculates the thickness of aortic aneurysm in radioactive ray mode.Particularly, with aortic aneurysm lumen centers, point out outbreak radioactive ray, angle is these radioactive ray and aortic aneurysm inner and outer boundary intersect at respectively a N and some F, and the distance of putting between N and F is the thickness of the aortic aneurysm in direction, the sustainer inner and outer boundary here refers to respectively in endovascular border with in EV border.

The inventive method can be cut apart for abdominal aneurvsm outer boundary, but is not limited to this, also can be used for other aortic aneurysm outer boundary and cuts apart.

Fig. 3 shows the abdominal aorta intravascular space of the several cases that are divided into by the inventive method, image is that the abdominal aorta intravascular space to cutting apart carries out the front 3-D view of three-dimensional reconstruction by surface rendering method, and abdominal aorta intravascular space has obtained complete extraction.Fig. 4 shows the abdominal aneurvsm outer boundary result of several groups of case CTA images that are divided into by the inventive method, wherein white contours line represents to cut apart by the inventive method the abdominal aneurvsm outer boundary profile obtaining, the real border of can fitting well, accurately extracts abdominal aneurvsm outer boundary.

Claims (9)

1. the aortic aneurysm outer boundary dividing method based on CTA image, is characterized in that, this dividing method comprises the following steps:

Obtain CTA image sequence;

From CTA image sequence, extract aortic blood tube cavity;

According to the first geometric properties location aortic aneurysm synusia of extracted intravascular space;

The segmentation result of a tomographic image was as the initial boundary that present image is cut apart in the past, and each aortic aneurysm synusia automatically location being obtained by CV model is cut apart, and extracted aortic aneurysm outer boundary.

2. the aortic aneurysm outer boundary dividing method based on CTA image according to claim 1, is characterized in that, the method for extracting aortic blood tube cavity from CTA image sequence comprises:

According to the CT value setting threshold of CTA image angiosomes, utilize this threshold value to cut apart every tomographic image of CTA image sequence, obtain the 2 dimensional region of similar blood vessel;

Calculate the second geometric properties of described 2 dimensional region, with setting value comparison, exclude part 2 dimensional region;

According to interlayer continuity, remaining 2 dimensional region is carried out to interlayer merging, obtain at least one 3D region;

Extract the 3D region being communicated with Seed Points, as aortic blood tube cavity.

3. the aortic aneurysm outer boundary dividing method based on CTA image according to claim 2, is characterized in that, described the second geometric properties comprises one or more in area, girth, average, variance, circularity and excentricity.

4. the aortic aneurysm outer boundary dividing method based on CTA image according to claim 2, is characterized in that, described threshold value comprises first threshold θ _hwith Second Threshold θ _l, utilize the method that this threshold value cuts apart to every tomographic image of CTA image sequence the 2 dimensional region that obtains similar blood vessel to comprise:

From CTA image sequence, read a tomographic image;

The pixel gray-scale value of lining by line scan, and be labeled as F (x, y), marking convention is

And same a line internal labeling is worth to identical pixel merges into a set, the row labels of going forward side by side;

The pixel point set that the line style of lining by line scan is arranged, when the pixel point set locus 8 between adjacent lines in abutting connection with time, the pixel collection that line style is arranged is merged into 2 dimensional region.

5. the aortic aneurysm outer boundary dividing method based on CTA image according to claim 1, is characterized in that, described the first geometric properties comprises minimax footpath and the area distributions of inner chamber.

6. the aortic aneurysm outer boundary dividing method based on CTA image according to claim 1, is characterized in that, described CV model comprises shape constraining energy, and this shape constraining energy definition is:

$E_{\mathrm{shape}}\left(X_i\right)=\left\{\begin{array}{c}d(X_i,C_i)\mathrm{if}{X}_i\∈\mathrm{outside}\left(c\right)\\ -d(X_i,C_i)\mathrm{if}{X}_i\∈\mathrm{inside}\left(c\right)\end{array}\right.$

Wherein, d (X _i, C _i) an expression point X _idistance along boundary method vector to border, outside (c) represents the region beyond border, inside (c) represents Yi Nei region, border.

7. the aortic aneurysm outer boundary dividing method based on CTA image according to claim 1, is characterized in that, described CV model comprises border indicator function $g(x,y)=\frac{1}{1+\mathrm{\λ}|\▿G*I(x,y)|},$ Wherein, for the partial gradient value after gaussian filtering, the weighting function that described border indicator function is controlled as boundary shape, in image, the value of sharpness of border place weighting function is less, larger in the value of obscurity boundary place weighting function.

8. the aortic aneurysm outer boundary dividing method based on CTA image according to claim 1, it is characterized in that, the level set function value at the reference mark in the arrowband that a CV model iterative computation is defined by initialization area, arrowband is defined as the curve of S=± δ/2, the inside and outside gray average in border is the inside and outside gray average in border in narrowband region, and δ represents level set function to be normalized the threshold value of choosing after calculating.

9. the aortic aneurysm outer boundary dividing method based on CTA image according to claim 1, is characterized in that, this aortic aneurysm outer boundary dividing method also comprises the step of calculating the thickness of aortic aneurysm in radioactive ray mode.