Method of quantifying asymmetry of an object

Abstract

The method is performed in a computer and comprises acts of importing medical image data sets of an object, selecting regions of interest of the object in each medical image data set, processing the medical image data and constructing an optimal symmetry plane. The act of selecting the regions of interest of the object in each medical image data set defines the regions of interest of the object in each medical image data set and locates the regions of interest of the object in each medical image data set. The act of processing the medical image data executes an error equation and obtains optimal parameters of the optimal symmetry plane with a symmetry value. The act of constructing an optimal symmetry plane constructs an optimal symmetry plane with the obtained optimal parameters. Base on the discovered Optimal Symmetry Plane, the slanted images from medical imaging can be rectified.

Description

BACKGROUND OF THE INVENTION

1. Field of the Invention

The present invention relates to a quantitative asymmetry method, and particularly to a method of quantifying asymmetry of an object. A symmetry value is developed from the quantified symmetry method. Base on the discovered Optimal Symmetry Plane (OSP), the slanted images from medical imaging can be rectified.

2. Description of Related Art

Due to advancement in the medical field, medical images have become an important tool for pre-surgical planning for oral, maxillofacial, and plastic surgery. Medical imaging may be two-dimensional (2-D) or three-dimensional (3-D) images, and the images assist physicians in obtaining a full picture of the nidus of each individual patient. In addition, a symmetric face of an individual is much more attractive esthetically than an asymmetric face. However, a completely symmetric face is virtually impossible in nature.

In general, medical images are obtained by x-ray, ultrasound, Computer Tomography (CT) or Magnetic Resonance Imaging (MRI), and the obtained medical images are 2-D images. Furthermore, multiple 2-D images are used to construct a 3-D image, and more 2-D images increase visual spatial realism of the 3-D image. A traditional method of constructing a symmetry plane of a 3-D image model for surgeon is to manually draw an approximated central plane on the 3-D image model. The 3-D image is presumed to have a symmetry plane based on the central lines of the skull, and further, the lateral half of the 3-D image is mirrored accordingly. That means, the lateral half of the 3-D image is constructed based on the presumed symmetry plane generated from the processed 2-D images.

In general, the central symmetry line of one 3-D image model is determined subjectively by an individual surgeon. However, each patient is unique and various parts of the body are asymmetric to varying extents, and further, each surgeon has a slightly different but distinct esthetic sense. Furthermore, a single surgeon may draw different central lines on the same medical image in different circumstances. Therefore, how to determine a unique reasonable and reliable symmetry plane of a 3-D image model is crucial for plastic and maxillofacial surgery.

A few publications relate to determining symmetry of an object and are particularly applicable to digital images. Josef Bigun proposed a method to find the best symmetry region in certain patterns. Bigun's proposed method uses a symmetry equation and derives the best symmetry region by transforming the symmetry equation into a specific spectrum with a least square method. In another publication, Hagit Zabrodsky, Shmuel Peleg and David Avnir proposed an edge verification method to quantify the symmetry level of the object.

These methods described above are able to determine the symmetry of a specific object but require extensive and complex computations in data processing. Even though promoting computational performance during the determination of the symmetry of an object and developing a novel method to evaluate the degree of symmetry of an object are stated objectives in the field, actually obtaining a reliable and consistent degree of asymmetry is the key issue.

The present invention provides a way of quantitatively determining symmetry of an object to overcome the aforementioned problems.

SUMMARY OF THE INVENTION

The main objective of the invention is to provide a method of quantifying asymmetry of an object by locating an optimal symmetry plane associated with a symmetry value of the object in a 3-D medical image model. A symmetry value indicates the symmetry degree, or in another word, the asymmetry degree. The optimal symmetry plane helps medical personnel to correct slanted original images for successive multiple applications.

The method in accordance with the present invention is performed in a computer and comprises acts of importing medical image data set of an object, selecting regions of interest of the object in each medical image data set, processing the medical image data in each data set and constructing an optimal symmetry plane associated with a best symmetry value. The act of importing medical image data sets of an object imports multiple medical image data sets of an object into a computer. The act of selecting the regions of interest of the object in each medical image data set comprises steps of selecting contrast and threshold arguments and processing the image data sets. The step of selecting contrast and threshold arguments defines the regions of interest of the object in each medical image data. The step of processing the image data sets locates the regions of interest of the object in each medical image data set. The act of processing the medical image data executes an error equation and obtains parameters corresponding to the error equation. The act of constructing an optimal symmetry plane comprises constructing an optimal symmetry plane with a best symmetry value.

Other objectives, advantages and novel features of the invention will become more apparent from the following detailed description when taken in conjunction with the accompanying drawings.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a functional block diagram of a first embodiment of a method for quantitatively determining symmetry of an object in accordance with the present invention;

FIG. 2 is a functional block diagram of a second embodiment of a method for quantitatively determining symmetry of an object in accordance with the present invention;

FIG. 3 is a functional block diagram of a third embodiment of a method for quantitatively determining symmetry of an object in accordance with the present invention;

FIG. 4 is a plane diagram illustrating the act of constructing an optimal symmetry line on image by using the minimum square method in FIG. 1;

FIG. 5 is a plane diagram illustrating the act of constructing an optimal symmetry plane and its corresponding parameters in FIGS. 2 and 3;

FIG. 6 is a diagram illustrating the act of voxel interpolation for medical image rectification in FIGS. 1, 2, and 3;

FIG. 7 is a series of medical chart used in the act of selecting the regions of interest of the object in medical image data and executing error function in FIG. 1;

FIG. 8(a) is a first medical image of a skull split into two optimal symmetry pieces by the acts in FIG. 1;

FIG. 8(b) is a second medical image of the skull split into two optimal symmetry pieces by the acts in FIG. 1;

FIG. 8(c) is a third medical image of the skull split into two optimal symmetry pieces by the acts in FIG. 1;

FIG. 8(d) is computer-generated image of the skull with an optimal symmetry plane constructed from FIGS. 8(a)-8(c) by the acts in FIG. 1;

FIG. 9(a) is a computer-generated perspective view of an image of a skull with an optimal symmetry plane obtained by the acts in FIG. 2;

FIG. 9(b) is another computer-generated perspective view of the skull with the optimal symmetry plane in FIG. 9(a);

FIG. 10(a) is a computer-generated front view of a skull with an optimal symmetry plane obtained by performing the acts in FIG. 3;

FIG. 10(b) is a computer-generated rear view of the skull with the optimal symmetry plane in FIG. 10(a);

FIG. 10(c) is a computer-generated perspective view of the skull in FIG. 10(a);

FIG. 11(a) is a medical image of a first slanted asymmetry object adjusted to a rectification image by the acts in FIG. 1;

FIG. 11(b) is a medical image of a second slanted asymmetric object adjusted to a rectification image by the acts in FIG. 2; and

FIG. 11(c) is a medical image of a third slanted asymmetric object adjusted to a rectification image by the acts in FIG. 3.

DETAILED DESCRIPTION OF PREFERRED EMBODIMENT

With reference to FIGS. 1 to 3, a method of quantifying asymmetry of an object in accordance with the present invention is performed in a computer and comprises acts of importing multiple medical image data sets of an object, selecting regions of interest of the object in each medical image data set, optionally selecting an exterior outline of the object, processing the medical image data in each data set, constructing an optimal symmetry plane associated with a best symmetry value and optionally constructing a corrected medical image with the optimal symmetry plane.

The act of importing medical image data sets of an object (11) comprises importing multiple medical image data sets of an object into a computer. The medical image data sets may be obtained from computer tomography (CT), ultrasonic imaging, magnetic resonance imaging (MRI) or the like. The medical image data sets may be 2-D images or 3-D images of the object.

The act of selecting regions of interest of the object in each medical image data set (21) locates the regions of interest (ROI) of the object in each medical image data set and comprises steps of selecting contrast and threshold arguments (211) and processing the image data (212).

With further reference to FIGS. 7 to 9, the contrast and threshold arguments relate to gray scale gradations of the image data. For example, pixels or voxels of a 2-D or 3-D image data set obtained by CT may have 4096 gray levels, and further, the gray levels of human bones in the image data may be in the range of 1323-4095. Therefore, the contrast and threshold arguments used in determining human bones are in the range of 1323-4095.

The step of processing the image data (212) uses the contrast and threshold arguments to separate the pixels or voxels of the medical image data into two groups (i.e. those conforming and those not conforming to the contrast and threshold arguments), and further, the pixels or voxels conforming to the contrast and threshold arguments are the regions of interest of the object. Therefore, the regions of interest of the object in each medical image data set are located.

With further reference to FIG. 10, the act of selecting an exterior outline of the object (31) comprises selecting the exterior outline of the object in the regions of interest by edge detection, such as the exterior outline of a skull.

The act of processing the medical image data in each data set (41) comprises steps of executing an error equation (411) and obtaining optimal parameters (422).

In the first embodiment the error equation and the corresponding parameters are obtained as follows.

1. Suppose f(x, y) is a gray level equation of the object in the medical image data, and then assume that L(θ, c)=(cos θ)x+(sin θ)y+c=0 is an symmetry line equation of the object in the medical image data, wherein θ represents an angle between the symmetry line L(θ,c) and the x-axis, and c is a constant.

2. If a pixel's coordinates are (x_i, y_i) and the pixel's corresponding symmetric coordinates relating to the symmetry line L(θ,c) are (x_si, y_si), then x_si=−cos(2θ)x_i−sin(2θ)y_i−2 cos(θ)c and y_si=−sin(2θ)x_i+cos(2θ)y_i−2 sin(θ)c. That is, if pixels exist such that (x_i, y_i)=(x_si, y_si), the symmetry line L(θ, c) is the optimal symmetry line of the object in the medical image data set.

3. The error equation is defined to be $E\left(\theta ,c\right)=\sum _{i=0}^{n-1}{\left[f\left(x_i,y_i\right)-f\left(x_{\mathrm{si}},y_{\mathrm{si}}\right)\right]}^2.$

4. Because the step of processing the image data (212) separates the object in the medical image data into two groups, the error equation can be simplified to $E\left(\theta ,c\right)=\sum _{i=0}^{n-1}\left(f\left(x_i,y_i\right)^f\left(x_{\mathrm{si}},y_{\mathrm{si}}\right)\right),$ wherein $f\left(x_i,y_i\right)^f\left(x_{\mathrm{si}},y_{\mathrm{si}}\right)=\{\begin{array}{cc}0,& f\left(x_i,y_i\right)=f\left(x_{\mathrm{si}},y_{\mathrm{si}}\right)\\ 1,& f\left(x_i,y_i\right)\ne f\left(x_{\mathrm{si}},y_{\mathrm{si}}\right).\end{array}$ Furthermore, a best symmetry value E(θ, c) is obtained when θ and c are the optimal parameters.

For example, a first embodiment of the method in accordance with the present invention analyzes a skull in a medical image data by first selecting and processing regions of interest in the medical image data set. With reference to FIG. 7, then a symmetry line is assumed to split the skull into a left half part and a right half part. If the left half and right half parts can be optimally overlapped, the symmetry line is optimal, and further, the left half and right half parts split by the symmetry line are most symmetric. Therefore, the optimal parameters are θ and c and are substituted into the equation L(θ,c) of the symmetry line obtained in the step of operating an error equation (411). Therefore, the symmetry line with equation L(θ,c) is an optimal symmetry line.

In the first embodiment, the act of constructing an optimal symmetry plane (51) is performed by the minimum square method (511) as follows.

1. With reference to FIG. 4, y=mx+c is the equation of the optimal symmetry line, d is the perpendicular distance between the optimal symmetry line and the center C of the object, and θ is the angle between the optimal symmetry line y and the x-axis.

2. Therefore, A_x=C_x+d sin(θ), A_Y=C_y−d cos(θ) and c=A_y−mA_x.

3. An optimal symmetry plane is assumed to be y=ax+bz+c.

4. Let $q=\sum _{i=1}^n[y_i-{\left(ax+bz+c\right]}^2.$ When q approaches zero, q′(a,b,c)=0. Furthermore, $\frac{\partial q}{\partial a}=\sum _{i=1}^n-2x_i\left(y_i-a{x}_i-b{z}_i\right)=0,\frac{\partial q}{\partial b}=\sum _{i=1}^n-2z_i\left(y_i-a{x}_i-b{z}_i\right)=0$ and $\frac{\partial q}{\partial c}=\sum _{i=1}^n-2\left(y_i-a{x}_i-b{z}_i\right)=0.$

5. Therefore, the following equation is obtained. $\left(\sum _{i=1}^n{x}_i^2\right)a+\left(\sum _{i=1}^n{x}_i{z}_i\right)b+\left(\sum _{i=1}^n{x}_i\right)c=\left(\sum _{i=1}^n{x}_i{y}_i\right)$$\left(\sum _{i=1}^n{x}_i{z}_i\right)a+\left(\sum _{i=1}^n{z}_i^2\right)b+\left(\sum _{i=1}^n{z}_i\right)c=\left(\sum _{i=1}^n{z}_i{y}_i\right)$$\left(\sum _{i=1}^n{x}_i\right)a+\left(\sum _{i=1}^n{z}_i\right)b+\left(n\right)c=\left(\sum _{i=1}^n{y}_i\right)$

6. Parameters a, b and c are eliminated with the Gauss elimination method to obtain an optimal symmetry plane.

With reference to FIGS. 8(a)-8(d), the skull in multiple medical image data obtained from doing CT is respectively split into two optimal symmetry pieces by an optimal symmetry line as previously described. After finishing the act of constructing an optimal symmetry plane (51), the optimal symmetry plane is composed of the multiple optimal symmetry lines.

In addition, the error equation and the optimal parameters in the second and third embodiments of the method in accordance with the present invention are obtained as follows.

1. f(x, y, z) is a gray level equation of the object in the medical image data set, and E(x, y, z)=ax+by+cz+d=0 is assumed to be a symmetry plane equation of the object in the medical image data set.

2. A pixel's coordinates are (x_i, y_i, z_i) and the pixel's symmetric coordinates assumed to relate to the symmetry plane are (x_si, y_si, z_si). Therefore $x_{\mathrm{si}}=x_i-\frac{2a}{\sqrt{a^2+b^2+c^2}}\left(a{x}_i+b{y}_i+c{z}_i+d\right),y_{\mathrm{si}}=y_i-\frac{2b}{\sqrt{a^2+b^2+c^2}}\left(a{x}_i+b{y}_i+c{z}_i+d\right)$$\mathrm{and}$$z_{\mathrm{si}}=z_i-\frac{2c}{\sqrt{a^2+b^2+c^2}}\left(a{x}_i+b{y}_i+c{z}_i+d\right).$

3. With reference to FIG. 5, L₁ is an intersection line between plane E and plane XY, L₂ is an intersection line between plane E and plane XZ, point P is an intersection point between the x-axis and the plane E, α is an angle between L₁ and the x-axis, β is an angle between L₂ and the x-axis and k_d is a distance between point P and the origin. Furthermore, the coordinates of point P are (k_d,0,0), the direction vector of L₁ is (cos α, sin α, 0), the direction vector of L₂ is (cos β, 0, sin β) and the plane E has a vertical vector (cos α,sin α,0)*(cos β,0,sin β)=(sin α sin β,sin β cos β,cos β sin α).

4. Therefore, a=sin α sin β, b=sin β cos β, c=cos β sin α and d=−k_d(sin α sin β), and further, the plane E can be simplified to (sin α sin β)x+(sin β cos β)y+(cos β sin α)z−(sin α sin β)k_d=0.

5. The error equation is defined as $E\left(\alpha ,\beta ,k_d\right)=\left(\sum _{i=0}^{n-1}{\left[f\left(x_i,y_i,z_i\right)-f\left(x_{\mathrm{si}},y_{\mathrm{si}},z_{\mathrm{si}}\right)\right]}^2\right)/n.$

6. Because the step of processing the image data (212) separates the object in the medical image data into two groups, the error equation can be simplified into $E\left(\alpha ,\beta ,k_d\right)=\left(\sum _{i=0}^{n-1}\left[f\left(x_i,y_i,z_i\right)^f\left(x_{\mathrm{si}},y_{\mathrm{si}},z_{\mathrm{si}}\right)\right]\right)/n,$ wherein ${f\left(x_{i,}{y}_i,z_i\right)}^{\bigwedge }f\left(x_{\mathrm{si}},y_{\mathrm{si}},z_{\mathrm{si}}\right)=\{\begin{array}{cc}1,& f\left(x_i,y_i,z_i\right)=f\left(x_{\mathrm{si}},y_{\mathrm{si}},z_{\mathrm{si}}\right)\\ 0,& f\left(x_i,y_i,z_i\right)\ne f\left(x_{\mathrm{si}},y_{\mathrm{si}},z_{\mathrm{si}}\right)\end{array}.$ Furthermore, a best symmetry value E(α,β,k_d) is obtained when parameters α, β and k_d are the optimal parameters.

In the second and third embodiments, the act of constructing an optimal symmetry plane (51) is performed as follows.

1. The optimal parameters α, β and k_d are substituted into the equation (sin α sin β)x+(sin β cos β)y+(cos β sin α)z−(sin α sin β)k_d=0 of the symmetry plane obtained in the steps of executing an error equation (411) and obtaining the optimal parameters (412).

2. As the value of f(x_i,y_i,z_i)ˆ(x_si,y_si,z_si) approaches to 1, the plane E is closer to optimal symmetry plane.

For example, the second and third embodiments of the method analyze a skull in a medical image data set and determine symmetry planes that pass through the symmetric center of the skull in the medical image data set.

The act of reconstructing a corrected medical image with the optimal symmetry plane (61) is performed by an interpolation method. With reference to FIG. 6, the corrected medical image is reconstructed as follows.

1. G(x,y.z) is a gray level equation for a point having coordinates (x,y,z) after performing the interpolation method, and V(i,j,k) is a gray level equation for eight points adjacent to the point (x,y,z). The equation G(x,y.z) is $G\left(x,y,z\right)=\sum _{i=0}^1\sum _{j=0}^1\sum _{k=0}^{1}V\left(i,j,k\right)\times Wx\left(k\right)\times Wy\left(j\right)\times Wz\left(i\right),\mathrm{wherein}$$Wx\left(n\right)=\{\begin{array}{cc}1-\frac{x}{dx}& \dots \mathrm{if}n=0;\\ \frac{x}{dx}& \dots \mathrm{if}n=1;\end{array}Wy\left(n\right)=\{\begin{array}{cc}1-\frac{y}{dy}& \dots \mathrm{if}n=0;\\ \frac{y}{dy}& \dots \mathrm{if}n=1;\end{array}Wz\left(n\right)=\{\begin{array}{cc}1-\frac{z}{dz}& \dots \mathrm{if}n=0;\\ \frac{z}{dz}& \dots \mathrm{if}n=1;\end{array}.$

The equation G(x,y.z) can be used to obtain a gray level value of each point in a new slice plane. Therefore, the new slice plane can be computed as described above.

With reference to FIGS. 11(a)-11(c), an asymmetric and slanted skull in the medical image data can be respectively corrected and re-sliced to be an optimal symmetry image with the first, the second and the third embodiments of the method.

Such a method helps people select a required part of an object in the medical image data set by inputting the least information to decrease computation time. Furthermore, the method executes an error equation to decrease the time to process and reconstruct the medical image data set. Therefore, people can obtain an optimal symmetry object as fast as possible. The method can assist medical personnel to correct the slanted original images more accurately and as fast as possible.

Even though numerous characteristics and advantages of the present invention have been set forth in the foregoing description, together with details of the structure and function of the invention, the disclosure is illustrative only. Changes may be made in detail especially in matters of arrangement of parts within the principles of the invention to the full extent indicated by the broad general meaning of the terms in which the appended claims are expressed.

Claims

1. A method of quantifying asymmetry of an object being operated in a computer and comprising acts of importing medical image data sets of an object into a computer; selecting regions of interest of the object in each medical image data set locating the regions of interest of the object in each medical image data set and comprising steps of selecting contrast and threshold arguments defining the regions of interest of the object in each medical image data set; and processing the image data by using the contrast and threshold arguments to locate the regions of interest of the object in each medical image data set; processing the medical image data in each data set by executing an error equation and obtaining optimal parameters corresponding to the error equation; and constructing an optimal symmetry plane with a best symmetry value, wherein the best symmetry value is a ratio value obtained from the error equation.

2. The method as claimed in claim 1, wherein the act of constructing an optimal symmetry plane further comprising steps of assuming an equation (sin α sin β)x+(sin β cos β)y+(cos β sin α)z−(sin α sin β)kd=0 being a symmetry plane of the object, wherein parameters α, β and kd are obtained from the step of obtaining the optimal parameters, α is an angle between an x-axis and an intersection line between the symmetry plane and the plane XY, β is an angle between the x-axis and an intersection line between the symmetry plane and the plane XZ and kd is a distance between a point on the symmetry plane and the origin; and obtaining the symmetry plane by substituting the optimal parameters α, β and kd into the equation (sin α sin β)x+(sin β cos β)y+(cos β sin α)z−(sin α sin β)kd=0 of the symmetry plane.

3. The method as claimed in claim 2, wherein the act of processing the medical image data sets further comprising steps of executing an error equation E⁡(α,β,kd)=(∑i=0n-1⁢[f⁡(xi,yi,zi)⋀⁢f⁡(xsi,ysi,zsi)])/n,whereinf⁡(xi,yi,zi)⋀⁢f⁡(xsi,ysi,zsi)={1,f⁡(xi, ⁢yi, ⁢zi)⁢ =f⁢ ⁢(xsi, ⁢ysi, ⁢zsi)0,f⁡(xi, ⁢yi, ⁢zi)⁢ ≠f⁢ ⁢(xsi, ⁢ysi, ⁢zsi);andobtaining the optimal parameters α, β and kd by executing the error equation.

4. The method as claimed in claim 1, wherein the act of constructing an optimal symmetry plane further comprising steps of assuming an equation L(θ, c)=(cos θ)x+(sin θ)y+c=0 to be a presumed symmetry line of the object in the medical image data, wherein θ represents an angle between the symmetry line L(θ,c) and the x-axis, and c is a constant; obtaining the symmetry plane by performing a minimum square method.

5. The method as claimed in claim 4, wherein the act of processing the medical image data sets further comprises steps of executing an error equation E⁡(θ,c)=∑i=0n-1⁢(f⁡(xi,⁢yi)⋀⁢f⁡(xsi,ysi)), wherein f⁡(xi,yi)⋀⁢f⁡(xsi,ysi)={0,f⁡(xi, ⁢yi)⁢ =f⁢ ⁢(xsi, ⁢ysi)1,f⁡(xi, ⁢yi)⁢ ≠f⁢ ⁢(xsi, ⁢ysi);andobtaining optimal parameters θ and c by executing the error equation.

6. The method as claimed in claim 1, wherein the symmetry value is

7. The method as claimed in claim 1, wherein the contrast and threshold arguments relate to gray scale gradations of the image data.

8. The method as claimed in claim 2 further comprising an act of selecting an exterior outline of the object, which selects the exterior outline of the object in the regions of interest by edge detection.

9. The method as claimed in claim 3 further comprising an act of selecting an exterior outline of the object, which selects the exterior outline of the object in the regions of interest by edge detection.

10. The method as claimed in claim 1 further comprising an act of reconstructing a corrected medical image with the optimal symmetry plane, which is performed with an interpolation method to correct slanted 2-D medical image data set.

11. The method as claimed in claim 2 further comprising an act of reconstructing a corrected medical image with the optimal symmetry plane, which is performed with an interpolation method to correct slanted 2-D medical image data set.

12. The method as claimed in claim 3 further comprising an act of reconstructing a corrected medical image with the optimal symmetry plane, which is performed with an interpolation method to correct slanted 2-D medical image data set.

13. The method as claimed in claim 4 further comprising an act of reconstructing a corrected medical image with the optimal symmetry plane, which is performed with an interpolation method to correct slanted 2-D medical image data set.

14. The method as claimed in claim 5 further comprising an act of reconstructing a corrected medical image with the optimal symmetry plane, which is performed with an interpolation method to correct slanted 2-D medical image data set.

15. The method as claimed in claim 6 further comprising an act of reconstructing a corrected medical image with the optimal symmetry plane, which is performed with an interpolation method to correct slanted 2-D medical image data set.

16. The method as claimed in claim 7 further comprising an act of reconstructing a corrected medical image with the optimal symmetry plane, which is performed with an interpolation method to correct slanted 2-D medical image data set.

17. The method as claimed in claim 8 further comprising an act of reconstructing a corrected medical image with the optimal symmetry plane, which is performed with an interpolation method to correct slanted 2-D medical image data set.