CN100423029C - Method of analysis of local patterns of curvature distributions - Google Patents

Abstract

A method of detecting lesions and polyps in a digital image, wherein said image comprises a plurality of 3D volume points, is provided. The method includes identifying (101) a surface in the image, and for each point in the image, calculating (102) a first curvature measure, forming (103) a set of rings of points about each point, each ring being of equal geodesic distance from a center point for the ring, calculating (104), for each ring, a standard deviation of the first curvature measure, and selecting (105) those rings with a minimum standard deviation for the first curvature measure. A first curvature slope is calculated (107) for the selected rings, those points where the curvature slope departs from the pattern expected for a polyp or lesion are deleted (108) from the surface.

Description

Be used to analyze the method for the local mode of curvature distribution

Cross reference to relevant U. S. application

The temporary patent application NO.60/494 that the application requires Anna Jerebko to submit on August 13rd, 2003, the right of priority of 909 " Analysis of Local Patterns of CurvatureDistributions ", the content of this application is combined as reference here.

Background of invention

Senior information makes it possible to detecting potential problems in the stage early and that more can treat in the available diagnosis from the current data that imaging system obtained.Suppose the detailed data that can obtain enormous quantity, must work out different algorithms so, with image data processing effectively and accurately from imaging system.By means of computing machine, on digital picture or digitized image, realize the improvement of Flame Image Process usually.

The digital collection system that is used to produce digital picture comprises digital X-ray film pick-up, computed tomography (" CT ") imaging, magnetic resonance imaging (" MRI "), ultrasonic (" US ") and such as the nuclear medicine technology of positron emission x-ray tomography (" PET ") and single photon emission computed x-ray tomography (" SPECT ").Also can be by for example producing digital picture for digitized forms by analog image such as the analog image scanning of typical X ray.Yet, the mass data in the digital picture for the difficulty that such as personnel such as doctor, under the situation of not adding help, makes an explanation normally and be dull.Computer-aided diagnosis (" CAD ") system help human aspect, especially visual, cut apart, detection, registration and medical pathologies journal play a crucial role in accusing.

Digital picture be by a series of expressions with produce by the numerical value of the relevant characteristic (such as gray-scale value or magnetic field intensity) of the anatomical location points of specific array position institute reference.This group anatomical location points comprises image area.In 2D digital picture or section district, discrete array position is called as pixel.Three-dimensional digital image can be made of the section district of piling up by different constructing technologies well known in the prior art.3D rendering is made up of discrete volume element (being also referred to as voxel), and described volume element is made up of the pixel in the 2D image.Pixel or voxel properties can be processed, to determine the different qualities about the patient anatomical model relevant with this pixel or voxel.

In case anatomic region and structure are fabricated and evaluate by analyzing pixel and/or voxel, just can be with subsequently the processing that utilizes region characteristic and feature and analytical applications in the relevant range, thus the precision and the efficient of raising imaging system.

In the more crucial CAD task one comprises from volume data (for example CT volume data) screening and detects dissimilar cancers early.For example, demonstrated the reduction of mortality ratio owing to detecting early and excising of cancer knurl such as the cancer of colon cancer.Pathology is spherical or hemispheric usually on geometric configuration.Under many circumstances, these spherical pathologies are attached to linearity or piecewise-linear surface.For example polyp of colon is given prominence to from colon wall and from fold in inner chamber.Unfortunately, before the late period of disease, existing method can not detect the characteristic symptom of various cancers usually.Therefore, provide early detecting of characteristic symptom in the primary goal aspect the preventative cancer screening that shifts to an earlier date.

Summary of the invention

In one aspect of the invention, provide a kind of method that is used for detecting in digital picture damage and polyp, wherein said image comprises a large amount of 3D volumetric point.This method comprises: identified surface in described image; For each point in this surface calculates the first curvature amount; Encircle for each point in this surface forms around one group of point of each described point, wherein each described point is the central point for its that group ring, and each ring has the geodesic distance that equates to the central point of this ring; Calculate the standard deviation of first curvature amount for each ring; Select those to have the ring of the lowest standard deviation of first curvature amount; For selected ring calculates the first curvature slope; And the deletion curvature slope departs from those points of the expectancy model of polyp or damage from this surface.

In another aspect of this invention, this method comprises: assemble the remaining point in the described surface, to form one or more a plurality of bunches; For each bunch calculates first curvature slope extreme value; Selection have each bunch first curvature slope extreme value extreme value (most extreme value) bunch, and discern selected have damage interested bunch; Insert the basal plane of damage interested.Described basal plane is opened described damage and the following separate tissue of damage.

In another aspect of this invention, provide a kind of method that detects damage and polyp in digital picture, wherein said image comprises a plurality of 3D volumetric point.This method comprises: identified surface in described image; For each point in this surface calculates Gaussian curvature and mean curvature; Encircle for each point in this surface forms around one group of point of each described point, wherein each described point is the central point for its that group ring, and each ring has the geodesic distance that equates to the central point of this ring; For each ring calculates the standard deviation of Gaussian curvature and the standard deviation of mean curvature; Select those to have the ring of the lowest standard deviation of Gaussian curvature and mean curvature; For selected ring calculates Gaussian curvature slope and mean curvature slope; And from this surface the deletion Gaussian curvature slope be greater than or equal to zero or mean curvature slope be less than or equal to those points of zero.

In another aspect of this invention, this method comprises: assemble the remaining point in the described surface, to form one or more a plurality of bunches; For each bunch calculates minimum Gaussian curvature slope and maximum mean curvature slope; Selection have minimum Gaussian curvature slope and maximum mean curvature slope bunch, and discern selected have damage interested bunch; And the basal plane that inserts damage interested.Described basal plane is opened described damage and the following separate tissue of damage.

In another aspect of this invention, a kind of computer-readable computer memory device is provided, it visibly comprises can be by the instruction repertorie of computing machine execution, this instruction repertorie is used for carrying out the method step that is used for detecting in digital picture damage and polyp, and wherein said image comprises a plurality of 3D volumetric point.Described method comprises: identified surface in described image; For each point in this surface calculates the first curvature amount; Encircle for each point in this surface forms around one group of point of each described point, wherein each described point is the central point for its that group ring, and each ring has the geodesic distance that equates to the central point of this ring; Calculate the standard deviation of first curvature amount for each ring; Select those to have the ring of the lowest standard deviation of first curvature amount; For selected ring calculates the first curvature slope; And the deletion curvature slope departs from those points of the expectancy model of polyp or damage from this surface.

In another aspect of this invention, this method comprises: assemble the remaining point in the described surface, to form one or more a plurality of bunches; For each bunch calculates first curvature slope extreme value; Selection have each bunch first curvature slope extreme value extreme value bunch, and identification has selected bunch of damage interested; And the basal plane that inserts damage interested.Described basal plane is opened described damage and the following separate tissue of damage.

Description of drawings

Fig. 1 describes the process flow diagram of the preferred method of the present invention.

Fig. 2 describes the example of the curvature distribution of cosine shape.

Fig. 3 a-b describe to have interpolation ring polyp of colon and have the center of interpolation, the polyp of colon of Zona transformans and the radial direction that curvature is calculated.

Fig. 4 presents the scatter diagram and the S of curvature points _KAnd S _HComputer curvature slopes.

Fig. 5 describes to be used to implement the exemplary computer system of the preferred embodiments of the invention.

Embodiment

Illustrative embodiment of the present invention is described below.For clear, all features of actual embodiment are not described in this manual.Certainly, it should be understood that in the improvement of any this actual embodiment, must make the specific decision of many embodiments, to realize that described constraint will have nothing in common with each other owing to different embodiments such as the developer's who defers to relevant system and constraint relevant commerce specific objective.In addition, it should be understood that this development effort may be complicated and time-consuming, but for those benefit from the those skilled in the art of present disclosure, will only be normal work to do.

Though the present invention can have different modifications and alternative form, its specific embodiment is for example shown in the drawings, and here describes in detail.Yet, it should be understood that, here the explanation of particular is not intended to limit the invention to disclosed special shape, but opposite, the present invention will be contained and be in as by all modifications scheme, equivalence or replacement scheme in the appended the spirit and scope of the present invention that claim limited.

Method described below relates to the anomaly sxtructure that finds in the 3D volumetric medical images (for example, but being not limited to CT or MRI image).This method is damaged applicable to the lip-deep polyp of the hollow organ in the human body that detects and characterize as air flue, colon, bladder and gall-bladder or other, finds the platelet aggregation in the artery etc.Before using the inventive method, the 3D volumetric image has been carried out pre-service, to detect the surface of organ of interest.Can use for example any method that is used for the sense organ surface of 3D Canny edge detector, described organ surface is the border between hollow organ's tissue and air or fluid or other material (for example ight soil in the colon).

What the existing method that is used to analyze the curvature of polyp of colon and shape was considered polyp or polyp surface has that part of of sphere or elliptical shape.Yet curved transition appears in the border outstanding along polyp from colon wall.Method disclosed herein is based on the analysis of curvature distribution, and the analysis of this curvature distribution is to carry out to the geodesic distance on that part of surface of polyp and surrounding tissue merging along the polyp centre of surface from expection.These methods are based on a kind of hypothesis, and the voxel that promptly approaches into most the polyp center in the damage on plane will have such curvature, the numerical value maximum of this curvature, and when a voxel when this lesion center is removed, the numerical value of this curvature reduces.

Can be jointly or consider individually such as different curvature measure and its patterns such as Gaussian curvature, principal curvatures, mean curvature, minimum curvature, maximum curvatures.For example, Gaussian curvature is along with from the increase of the geodesic distance of polyp centre of surface and reduce, and becomes negative value or equal zero in polyp and surrounding tissue merging place.Mean curvature is followed opposite pattern.

The preferred method of the present invention is included at least one that calculate in (voxel) in average and the Gaussian curvature of having a few on surface, especially preferably calculates average and Gaussian curvature.It should be noted, under the situation that does not depart from the scope of the invention, can adopt other measures of surface curvature.Though will the present invention be described according to the gaussian sum mean curvature, method disclosed herein is not limited to disclosed curvature measure.What the existing method that is used to damage curvature and shape analysis was considered damage or its surface has that part of of sphere or elliptical shape.Here suppose that interested damage is outstanding from organ surface, and more or less has symmetric shape, focuses on symmetry simultaneously.To have minimum variance in this ring and will follow special pattern with the curvature of concentric ring on the injured surface aimed at of central point at damage top: curvature little by little changes from the damage top to that part of surface of damage and surrounding tissue merging.So when curvature was followed reverse mode, these methods also can be used to find the chamber in the surface of organ (for example colonic diverticula).

The analysis of curvature distribution is along carrying out from the radial distance of lesion center of expection, and comprises that part of surface of polyp and surrounding tissue merging.The voxel that approaches most the polyp center on the injured surface will have maximum Gaussian curvature, and this Gaussian curvature is along with reducing from the increase of the distance of polyp center, and become negative value or approach zero (Fig. 1) at polyp and surrounding tissue merging place.Mean curvature is followed opposite pattern.

When considering Gaussian curvature, the general mode of the curvature distribution from the center along the radius on all directions can be described roughly by cosine function: have maximum curvature at central point, along with distance reduces gradually, in Zona transformans, may become negative value, increase again then.The more or less circular candidate polyp that the curved transition pattern-recognition of cosine shape is possible.Figure 2 illustrates the example of the curvature distribution of cosine shape.The damage of the injustice as polyp of colon and air flue damage is given prominence to from colon wall and from the fold of healthy surface in inner chamber.Curved transition appears along outstanding border.Uneven injured surface, possible that even up or elongation or irregular polyp shape and the fold-like structures of adhering to cause non-ideal characteristic.When the surface was different from desirable hemispherical shape further, the pattern of curvature will little by little be launched from cosine shape.The actual curvature value may be scattered on quite wide angle.But most damages will be followed " cosine " pattern that surface curvature distributes, and make it possible to the health tissues in homolog the feature of " cosine " curvature distribution pattern as automatic damage check.With focus on sphere or comprise that the method for curvature itself of Gaussian curvature, mean curvature and principal curvatures is different, the curvature slope that combines with the standard deviation of curvature in the Zona transformans is constant the measuring of scale of damage circularity and symmetry.

A kind of method for optimizing of the present invention is included in the curvature of calculating based on intensity along all points of the inwall of organ.In each position, calculate and analyze the scatter diagram of the curvature distribution in the radius of selected location.Referring now to Fig. 1, at first in step 101, the 3D edge detection method is applied to interested volume, to find the surface voxel of organ inwall.A non-limitative example of edge detection method is the Canny edge detector, and it produces thin and continuous edge.

In case a surface is identified, then calculate average (H) and/or Gaussian curvature (K) for each surface voxel in step 102.For gray level image I (z), Gauss (K) and average (H) curvature can be passed through for x, y:

$K=\frac{1}{h^2}\left[\begin{array}{c}{I}_x^2(I_{\mathrm{yy}}{I}_{\mathrm{zz}}-I_{\mathrm{yz}}^2)+{2I}_y{I}_z(I_{\mathrm{xz}}{I}_{\mathrm{xy}}-I_{\mathrm{xx}}{I}_{\mathrm{yz}})\\ +I_y^2(I_{\mathrm{xx}}{I}_{\mathrm{zz}}-I_{\mathrm{yz}}^2)+{2I}_x{I}_z(I_{\mathrm{yz}}{I}_{\mathrm{xy}}-I_{\mathrm{yy}}{I}_{\mathrm{xz}})\\ +I_z^2(I_{\mathrm{xx}}{I}_{\mathrm{yy}}-I_{\mathrm{xy}}^2)+{2I}_x{I}_y(I_{\mathrm{xz}}{I}_{\mathrm{yz}}-I_{\mathrm{zz}}{I}_{\mathrm{xy}})\end{array}\right]---\left(1\right)$

$H=\frac{1}{{2h}^{3/2}}\left[\begin{array}{c}{I}_x^2(I_{\mathrm{yy}}+I_{\mathrm{zz}})-{2I}_y{I}_z{I}_{\mathrm{yz}}\\ +I_y^2(I_{\mathrm{xx}}+I_{\mathrm{zz}})-{2I}_x{I}_y{I}_{\mathrm{xz}}\\ +I_z^2(I_{\mathrm{xx}}+I_{\mathrm{yy}})-{2I}_x{I}_y{I}_{\mathrm{xy}}\end{array}\right]---\left(2\right)$

Define, wherein I (x, y, z) presentation surface (x, y, z) ∈ R ³Summit intensity, I _xThe presentation video data are to the partial derivative of x, I _XzExpression is to mixed partial derivative of x and z etc., and $h=I_x^2+I_y^2+I_z^2.$

In step 103, the voxel that has equal geodesic distance from decentering forms ring around each surface voxel, and can calculate average (H) and Gauss (K) curvature for all surface voxel in each ring.For example, integer can be used as the distance that forms ring: first (center) ring by to when the distance D of anterior centrosome element less than 1mm (voxel of D＜=1mm) forms, and the next one encircles and formed by the voxel that is positioned at apart from 1mm＜D＜=2mm place, or the like.For better precision, can use ring with meticulousr gradient.Fig. 3 a has described the polyp of colon that has the ring of interpolation on the polyp surface.Then, in step 104, in each ring, can calculate the standard deviation of gaussian sum mean curvature.In step 105, select those rings of the standard deviation minimum of curvature.In this group ring, continue to select curvature to have extreme value, also promptly have a minimal Gaussian curvature (K then _Min) or maximum mean curvature (H _Max) or have minimal Gaussian curvature (K _Min) and maximum mean curvature (H _Max) those rings so that obtain the concentric ring that a group switching centre limits the center of structures of interest.

Given one group has minimal Gaussian curvature (K _Min) or maximum mean curvature (H _Max) ring, then can be in step 107 following calculating curvature slope:

S _K＝(K _c-k _min)/D _Kmin，

S _H＝(H _c-H _max)/D _Kmax，

Wherein:

S _KIt is Gaussian curvature slope;

S _HIt is mean curvature slope;

K _cIt is the Gaussian curvature in the center ring;

K _MinBe curvature with ring of minimal Gaussian curvature;

H _cIt is the mean curvature in the center ring;

H _MaxBe curvature with ring of maximum mean curvature;

D _KminIt is distance corresponding to ring with minimal Gaussian curvature; With

D _KmaxIt is distance corresponding to ring with maximum mean curvature.

Fig. 3 b has described the radial direction of curvature calculating and center and the negative curvature Zona transformans that calculates.Fig. 4 has shown the scatter diagram and the S of curvature points _KAnd S _HComputer curvature slopes.When center ring was positioned at the top of polypoid structure, having may be corresponding to the damage deformation band such as the ring of the curvature extreme value of minimal Gaussian curvature and/or maximum mean curvature.Have negative Gaussian curvature slope and (or) voxel of positive mean curvature slope may be corresponding to lip-deep polyp or another lump.Fold or other cylindrical objects also may have identical characteristic, but in fold, has curvature criteria deviation in the ring of minimal Gaussian curvature and/or maximum mean curvature obviously greater than the curvature criteria deviation in polyp (perhaps other oblong surface points).Under this hypothesis, in step 108, can detect all surface voxel, wherein when surface voxel belongs to flat surfaces or belong to depression outside the damage, S _K＞=0 and/or S _H＜=0.In addition S can be set _KAnd S _HThe threshold value of numerical value, further damage and other surface structures are made a distinction.

In step 109, remaining surface voxel is carried out cluster.At each bunch, in step 110, calculate curvature slope extreme value, be minimum S _K(or) maximum S _HThen, in step 111, in given sub volume of interest, in all bunches, having minimum S _KOr maximum S _HBunch be chosen as corresponding to damage interested bunch.Injured surface by belong to obtain bunch voxel represent.Then, in step 112, by the basal plane that inserts to determine following tissue and lesion volume are separated.Be considered to belong to damage by the voxel between bunch determined surface that obtains and the cut surface that inserted.

For example the method for optimizing of said method can be applied to the damage check such as colon, air flue and other organs in the different imaging form such as CT, MR, US, PET under hard-core situation.

It should be understood that the present invention can implement with the multi-form of hardware, software, firmware, dedicated process or its combination.In one embodiment, the present invention can be implemented as the application program that visibly is included on the computer-readable program storage device with form of software.This application program can be uploaded, and can be carried out by the machine that comprises any suitable architecture.

Referring now to Fig. 5,, be used to implement computer system 501 of the present invention and especially can comprise CPU (central processing unit) (CPU) 502, storer 503 and I/O (I/O) interface 504 according to one embodiment of the invention.This computer system 501 is coupled to display 505 via I/O interface 504 and usually such as various input equipments 506 such as mouse and keyboards.Support circuit can comprise circuit such as cache memory, power supply, clock circuit and communication bus.Storer 503 can comprise random-access memory (ram), ROM (read-only memory) (ROM), disc driver, tape drive etc. or its combination.The present invention can be implemented as and be stored in the storer 503 and routine 507 that carried out by CPU 502, to handle the signal from signal source 508.Similarly, computer system 501 is general-purpose computing systems, and it becomes dedicated computer system when carrying out routine 507 of the present invention.

Computer system 501 also comprises operating system and micro-instruction code.Various process as described herein and function can be the parts (perhaps its combination) of the part of micro-instruction code or the application program that is performed via operating system.In addition, various other peripherals, can be connected on the computer platform such as additional data storage device and printing device.

In addition, it should be understood that, because the assembly of the composition system of being described in the accompanying drawings and some in the method step can be implemented with software, so depend on the mode that the present invention is programmed, the actual relationship between the system component (perhaps method step) may be different.Under the situation of given the present invention provided here instruction, the those of ordinary skill in the association area can be considered these and similarly embodiment or configuration of the present invention.

Above-mentioned particular embodiment only is illustrative, because it is evident that for the those of skill in the art that benefit from instruction described here, can revise or put into practice the present invention in difference mode still of equal value.In addition, except described in the following claim, the structure shown in being not limited to here or the details of design.Therefore, clearly, can change or revise above-mentioned particular embodiment, and think that all this flexible programs all are in the scope and spirit of the present invention.Therefore, the protection of looking for here is described in claim below.

Claims (11)

1. method that is used for detecting damage and polyp in digital picture, wherein said image comprises a plurality of 3D volumetric point, said method comprising the steps of:

Identified surface in described image;

For each point in this surface calculates the first curvature amount;

Encircle for each point in this surface forms around one group of point of each described point, wherein each described point is the central point for its that group ring, and each ring has the geodesic distance that equates to the central point of this ring;

Calculate the standard deviation of first curvature amount for each ring;

Select those to have the ring of the lowest standard deviation of first curvature amount;

For selected ring calculates the first curvature slope; And

The deletion curvature slope departs from those points of the expectancy model of polyp or damage from this surface.

2. may further comprise the steps in addition in accordance with the method for claim 1:

Remaining point in the described surface is carried out cluster, to form one or more bunch;

For each bunch calculates first curvature slope extreme value;

Selection have each bunch first curvature slope extreme value extreme value bunch, and discern selected have damage interested bunch; And

Insert the basal plane of damage interested, described basal plane is opened the separate tissue of described damage below damage.

3. in accordance with the method for claim 2,

Wherein said first curvature amount is to be defined as

$K=\frac{1}{h^2}\left[\begin{array}{c}{I}_x^2(I_{\mathrm{yy}}{I}_{\mathrm{zz}}-I_{\mathrm{yz}}^2)+2I_y{I}_z(I_{\mathrm{xz}}{I}_{\mathrm{xy}}-I_{\mathrm{xx}}{I}_{\mathrm{yz}})\\ +I_y^2(I_{\mathrm{xx}}{I}_{\mathrm{zz}}-I_{\mathrm{yz}}^2)+2I_x{I}_z(I_{\mathrm{yz}}{I}_{\mathrm{xy}}-I_{\mathrm{yy}}{I}_{\mathrm{xz}})\\ +I_z^2(I_{\mathrm{xx}}{I}_{\mathrm{yy}}-I_{\mathrm{xy}}^2)+2I_x{I}_y(I_{\mathrm{xz}}{I}_{\mathrm{yz}}-I_{\mathrm{zz}}{I}_{\mathrm{xy}})\end{array}\right]$

Gaussian curvature, wherein I represents intensity, this intensity is to have coordinate (x, y, z) ∈ R ³The function of described lip-deep point, single subscript of I is represented I to the partial derivative by the indicated coordinate of described subscript, the double subscript of I represents the mixed partial derivative by the indicated coordinate of described subscript, and $h=I_x^2+I_y^2+I_z^2,$ And

Wherein the first curvature slope is by S _K=(K _c-K _Min)/D _Kmin, defined Gaussian curvature slope, wherein

S _KBe Gaussian curvature slope,

K _cIt is the Gaussian curvature in the center ring;

K _MinBe curvature with ring of minimal Gaussian curvature; And

D _KminIt is distance corresponding to ring with minimal Gaussian curvature; And

Wherein for polyp or damage, Gaussian curvature slope is less than zero, and the curvature slope extreme value is given bunch a minimum curvature slope, and the extreme value of each extreme value is minimum Gaussian curvature slope.

4. in accordance with the method for claim 2, wherein said first curvature amount be by

$H=\frac{1}{2h^{3/2}}\left[\begin{array}{c}{I}_x^2(I_{\mathrm{yy}}+I_{\mathrm{zz}})-2I_y{I}_z{I}_{\mathrm{yz}}\\ +I_y^2(I_{\mathrm{xx}}+I_{\mathrm{zz}})-2I_x{I}_y{I}_{\mathrm{xz}}\\ +I_z^2(I_{\mathrm{xx}}+I_{\mathrm{yy}})-2I_x{I}_y{I}_{\mathrm{xy}}\end{array}\right]$

Defined mean curvature,

Wherein I represents intensity, and this intensity is to have coordinate (x, y, z) ∈ R ³The function of described lip-deep point, single subscript of I is represented I to the partial derivative by the indicated coordinate of described subscript, the double subscript of I represents the mixed partial derivative by the indicated coordinate of described subscript, and $h=I_x^2+I_y^2+I_z^2,$ And

Wherein the first curvature slope is by S _H=(H _c-H _Max)/D _Kmax, defined mean curvature slope, wherein

S _HBe mean curvature slope,

H _cBe the mean curvature in the center ring,

H _MaxBe curvature with ring of maximum mean curvature, and

D _KmaxBe distance corresponding to ring with maximum mean curvature, and

Wherein for polyp or damage, mean curvature slope is greater than zero, and the curvature slope extreme value is given bunch a maximum curvature slope, and the extreme value of each extreme value is maximum mean curvature slope.

5. comprise in addition in accordance with the method for claim 2:

For each point in this surface calculates the torsion amount;

Calculate the standard deviation of torsion amount for each ring in this surface;

Select those to have the ring of the lowest standard deviation of torsion amount;

For selected ring calculates the torsion slope;

Deletion first curvature slope and torsion slope depart from those points of the expectancy model of polyp or damage from this surface;

For each bunch calculates torsion slope extreme value; And

Selection have the extreme value of first curvature slope extreme value and torsion slope extreme value extreme value bunch, with discern selected have damage interested bunch.

6. in accordance with the method for claim 5, wherein said first curvature amount is to be defined as

$K=\frac{1}{h^2}\left[\begin{array}{c}{I}_x^2(I_{\mathrm{yy}}{I}_{\mathrm{zz}}-I_{\mathrm{yz}}^2)+2I_y{I}_z(I_{\mathrm{xz}}{I}_{\mathrm{xy}}-I_{\mathrm{xx}}{I}_{\mathrm{yz}})\\ +I_y^2(I_{\mathrm{xx}}{I}_{\mathrm{zz}}-I_{\mathrm{yz}}^2)+2I_x{I}_z(I_{\mathrm{yz}}{I}_{\mathrm{xy}}-I_{\mathrm{yy}}{I}_{\mathrm{xz}})\\ +I_z^2(I_{\mathrm{xx}}{I}_{\mathrm{yy}}-I_{\mathrm{xy}}^2)+2I_x{I}_y(I_{\mathrm{xz}}{I}_{\mathrm{yz}}-I_{\mathrm{zz}}{I}_{\mathrm{xy}})\end{array}\right]$

Gaussian curvature, wherein I represents intensity, this intensity is to have coordinate (x, y, z) ∈ R ³The function of described lip-deep point, single subscript of I is represented I to the partial derivative by the indicated coordinate of described subscript, the double subscript of I represents the mixed partial derivative by the indicated coordinate of described subscript, and $h=I_x^2+I_y^2+I_z^2,$ And

Wherein the first curvature slope is by S _K=(K _c-K _Min)/D _Kmin, defined Gaussian curvature slope, wherein

S _KBe Gaussian curvature slope,

K _cIt is the Gaussian curvature in the center ring;

K _MinBe curvature with ring of minimal Gaussian curvature; And

D _KminIt is distance corresponding to ring with minimal Gaussian curvature; And

Wherein for polyp or damage, Gaussian curvature slope is less than zero, and the curvature slope extreme value is given bunch a minimum curvature slope, and the extreme value of each extreme value is minimum Gaussian curvature slope.

7. in accordance with the method for claim 5, wherein said torsion amount be by

$H=\frac{1}{2h^{3/2}}\left[\begin{array}{c}{I}_x^2(I_{\mathrm{yy}}+I_{\mathrm{zz}})-2I_y{I}_z{I}_{\mathrm{yz}}\\ +I_y^2(I_{\mathrm{xx}}+I_{\mathrm{zz}})-2I_x{I}_y{I}_{\mathrm{xz}}\\ +I_z^2(I_{\mathrm{xx}}+I_{\mathrm{yy}})-2I_x{I}_y{I}_{\mathrm{xy}}\end{array}\right]$

Defined mean curvature, wherein I represents intensity, this intensity is to have coordinate (x, y, z) ∈ R ³The function of described lip-deep point, single subscript of I is represented I to the partial derivative by the indicated coordinate of described subscript, the double subscript of I represents the mixed partial derivative by the indicated coordinate of described subscript, and $h=I_x^2+I_y^2+I_z^2,$ And

Wherein the torsion slope is by S _H=(H _c-H _Max)/D _Kmax, defined mean curvature slope, wherein

S _HBe mean curvature slope,

H _cBe the mean curvature in the center ring,

H _MaxBe curvature with ring of maximum mean curvature, and

D _KmaxBe distance corresponding to ring with maximum mean curvature, and

Wherein for polyp or damage, mean curvature slope is greater than zero, and the curvature slope extreme value is given bunch a maximum curvature slope, and the extreme value of each extreme value is maximum mean curvature slope.

8. in accordance with the method for claim 1, wherein said first curvature amount is a Gaussian curvature, and may further comprise the steps in addition:

For each point in this surface calculates mean curvature;

For each ring calculates the standard deviation of Gaussian curvature and the standard deviation of mean curvature;

Select those to have the ring of the lowest standard deviation of Gaussian curvature and mean curvature;

For selected ring calculates Gaussian curvature slope and mean curvature slope; And

From this surface the deletion Gaussian curvature slope be greater than or equal to zero or mean curvature slope be less than or equal to those points of zero.

9. may further comprise the steps in addition in accordance with the method for claim 8:

Remaining point in the described surface is carried out cluster, to form one or more bunch;

For each bunch calculates minimum Gaussian curvature slope and maximum mean curvature slope;

Selection have minimum Gaussian curvature slope and maximum mean curvature slope bunch, and discern selected have damage interested bunch; And

Insert the basal plane of damage interested, described basal plane is opened the separate tissue of described damage below damage.

10. in accordance with the method for claim 9, wherein said Gaussian curvature is defined as

$K=\frac{1}{h^2}\left[\begin{array}{c}{I}_x^2(I_{\mathrm{yy}}{I}_{\mathrm{zz}}-I_{\mathrm{yz}}^2)+2I_y{I}_z(I_{\mathrm{xz}}{I}_{\mathrm{xy}}-I_{\mathrm{xx}}{I}_{\mathrm{yz}})\\ +I_y^2(I_{\mathrm{xx}}{I}_{\mathrm{zz}}-I_{\mathrm{yz}}^2)+2I_x{I}_z(I_{\mathrm{yz}}{I}_{\mathrm{xy}}-I_{\mathrm{yy}}{I}_{\mathrm{xz}})\\ +I_z^2(I_{\mathrm{xx}}{I}_{\mathrm{yy}}-I_{\mathrm{xy}}^2)+2I_x{I}_y(I_{\mathrm{xz}}{I}_{\mathrm{yz}}-I_{\mathrm{zz}}{I}_{\mathrm{xy}})\end{array}\right],$

Wherein I represents intensity, and this intensity is to have coordinate (x, y, z) ∈ R ³The function of described lip-deep point, single subscript of I is represented I to the partial derivative by the indicated coordinate of described subscript, the double subscript of I represents the mixed partial derivative by the indicated coordinate of described subscript, and $h=I_x^2+I_y^2+I_z^2,$ And Gaussian curvature slope is by S _K=(K _c-K _Min)/D _Kmin, it is defined,

Wherein

S _KBe Gaussian curvature slope,

K _cBe the Gaussian curvature in the center ring,

K _MinBe curvature with ring of minimal Gaussian curvature, and

D _KminIt is distance corresponding to ring with minimal Gaussian curvature.

11. in accordance with the method for claim 9,

Wherein said mean curvature be by

$H=\frac{1}{2h^{3/2}}\left[\begin{array}{c}{I}_x^2(I_{\mathrm{yy}}+I_{\mathrm{zz}})-2I_y{I}_z{I}_{\mathrm{yz}}\\ +I_y^2(I_{\mathrm{xx}}+I_{\mathrm{zz}})-2I_x{I}_y{I}_{\mathrm{xz}}\\ +I_z^2(I_{\mathrm{xx}}+I_{\mathrm{yy}})-2I_x{I}_y{I}_{\mathrm{xy}}\end{array}\right]$

Defined, wherein I represents intensity, and this intensity is to have coordinate (x, y, z) ∈ R ³The function of described lip-deep point, single subscript of I is represented I to the partial derivative by the indicated coordinate of described subscript, the double subscript of I represents the mixed partial derivative by the indicated coordinate of described subscript, and $h=I_x^2+I_y^2+I_z^2,$ And

Wherein mean curvature slope is by S _H=(H _c-H _Max)/D _Kmax, it is defined,

Wherein

S _HBe mean curvature slope,

H _cBe the mean curvature in the center ring,

H _MaxBe curvature with ring of maximum mean curvature, and

D _KmaxIt is distance corresponding to ring with maximum mean curvature.

Images