# Method and apparatus for three-dimensional interactive tools for semi-automatic segmentation and editing of image objects

A system and method for segmenting and editing anatomical objects from medical images is disclosed. The system may be a medical diagnostic imaging system. A computer unit may execute computer software for segmenting anatomical objects from medical images. The computer software may extract an anatomical object from planar curves. Additionally, the computer software may correct the shape of an existing three-dimensional anatomical object from planar curves. The planar curves may be orthogonal to each other. A user may contour of an anatomical object on a plurality of slices, such as an axial slice a sagittal slice, a coronal slice, or some combination thereof. The contour may be drawn using a tracing pen on a display unit. The display unit may receive touch screen input from the tracing pen. The display unit may display the three-dimensional segmented anatomical object.

## Latest Patents:

- Low friction instrument driver interface for robotic systems
- Combination device for endoscopic and arthroscopic surgical procedures
- Inspection management system, inspection management apparatuses, and inspection management method
- Battery control for safeguarding lower voltage integrated circuits
- Work machine

**Description**

**RELATED APPLICATIONS**

[Not Applicable]

**FEDERALLY SPONSORED RESEARCH OR DEVELOPMENT**

[Not Applicable]

**MICROFICHE/COPYRIGHT REFERENCE**

[Not Applicable]

**BACKGROUND OF THE INVENTION**

The present invention generally relates to a system and method for improved medical imaging. Particularly, the present invention relates to a more efficient system and method for segmenting anatomical objects and correcting the segmentation of anatomical objects.

Medical diagnostic imaging systems encompass a variety of imaging modalities, such as x-ray systems, computerized tomography (CT) systems, ultrasound systems, electron beam tomography (EBT) systems, magnetic resonance (MR) systems, and the like. Medical diagnostic imaging systems generate images of an object, such as a patient, for example, through exposure to an energy source, such as x-rays passing through a patient, for example. The generated images may be used for many purposes. For instance, internal defects in an object may be detected. Additionally, changes in internal structure or alignment may be determined. Fluid flow within an object may also be represented. Furthermore, the image may show the presence or absence of objects in an object. The information gained from medical diagnostic imaging has applications in many fields, including medicine and manufacturing.

One application of utilizing the information gained from medical diagnostic imaging systems in the field of medicine is the segmentation of anatomical objects. The segmentation of anatomical objects and/or structures from two and three-dimensional images is important to allow the analysis of those anatomical objects and/or structures. For example, a particular organ or tissue may be extracted from the surrounding organs or tissues. The extracted organ or tissue may then be viewed independent of other objects that are not of interest. Such extraction allows a physician to focus only on the objects or structures of interest and develop a more accurate diagnosis and treatment strategy.

Anatomical segmentation, however, is a complex problem. Manual segmentation is a tedious, time consuming process. Fully automatic segmentation, although ideal, currently does not yield acceptable results. A combination of manual segmentation and automatic segmentation has yielded a number of interactive segmentation techniques.

Currently, a “live wire” technique allows a user to select a seed point on a contour, and while dragging and moving the mouse, the optimal line between the seed point and the current position may be computed. The live wire algorithm, however, may only be used in the two dimensional space and hence, this technique is a slice-by-slice segmentation, where the drawn contour on one slice becomes the initial contour on the next slice and this initial contour can be deformed.

Another strategy models edges in a surface mesh as semi-elastic linked elements in a chain. The surface mesh vertices connect the edges, so when a vertex is edited, the displacement stretches or compresses its neighboring edges. The difficulty in this strategy is to define the required extent of displacement. Another strategy that may be used is to use some medical information about the to-be-segmented object. The information may be a model, a map that can be verified to the actual medical images and then modified by the user. Since some medical objects may assume several forms, it is very hard to choose the correct model for the actual medial images or the degree of deformation allowable.

In addition, once an object has been segmented, a user still has the problem of manually editing the object. Usually, only minor changes are required after the display of the results of the algorithm. These changes, however, generally have to be made on several slices and on complex target shapes. Present editing tools are time consuming and difficult to use, making post-segmentation editing a disproportionately long and difficult task.

Accordingly, a system and method is needed for easier segmentation and editing of anatomical objects. Specifically, a need exists for a three-dimensional semi-automatic segmentation tool and a correction tool that is easier to use and more accurate than present tools. Such a system and method may allow a user to be more efficient and effective in diagnosing and treating medical conditions.

**SUMMARY OF THE INVENTION**

Certain embodiments of the present invention may include a method for segmenting anatomical objects from medical images. The method may include acquiring at least one axial curve and at least one orthogonal curve from a set of medical images. The orthogonal curve may include a sagittal curve or coronal curve. The method may also include optionally modifying the orthogonal curve to intersect the axial curve. The option is exercised if the axial curve and the orthogonal curve, as acquired, do not intersect.

Next, template shapes for axial slices may be generated using shape-based interpolation. The template shapes are generated for axial slices lacking user-drawn curves. Attractor points are then located by the intersections of the orthogonal curves and the axial planes. The transformation parameters for the axial slices lacking user-drawn curves are then computed. Various transformation methods may be used. In an embodiment, a two-dimensional affine transformation may be used. The points of the curves drawn on orthogonal slices may be used as attractors to guide the deformation. The transformation parameters may be computed by optionally complementing the input if the input does not account for sagittal or coronal input; computing the closest contour points to the attractor points for the slices and iteratively computing the parameters of the two-dimensional affine transformation. The transformation parameters are then optionally smoothed in the case of two-dimensional—slice-by-slice—transformation. Finally, the transformation is executed on the template shapes for the axial slices.

Moreover, certain embodiments of the present invention may include a method for correcting the shapes of existing three-dimensional anatomical objects from medical images. The method may include acquiring at least one axial curve and at least one orthogonal curve from a set of medical images. The orthogonal curve may include a sagittal curve or coronal curve. The method may also include optionally modifying the orthogonal curve to intersect the axial curve. The modification allows the orthogonal curves to intersect with the axial curves.

Next, template shapes for axial slices may be generated using shape-based interpolation. The generated template shapes may be modified on the axial slices to reach the to be corrected object. The template shapes are generated for axial slices lacking user-drawn curves. Attractor points are then located by the intersections of the orthogonal curves and the axial planes. The transformation parameters for the axial slices lacking user-drawn curves are then computed. Various transformation methods may be used. In an embodiment, a two-dimensional affine transformation may be used. The points of the curves drawn on orthogonal slices are used as attractors to guide the deformation. The transformation parameters may be computed by optionally complementing the input if the input does not account for sagittal or coronal input; computing the closest contour points to the attractor points for each slice; and iteratively computing the parameters of the two-dimensional affine transformation. The transformation parameters are then optionally smoothed in the case of two-dimensional—slice-by-slice—transformation. Next, the transformation is executed on the template shapes for the axial slices. Finally, a plurality of joints on the original object and the corrected object are smoothed.

Certain embodiments of the present invention include a medical diagnostic imaging system. The medical diagnostic imaging system may include a computer unit for manipulating data. The computer unit executes computer software for segmenting anatomical objects from medical images. The computer software may extract an anatomical object from planar curves. Alternatively, the computer software may correct the shape of an existing three-dimensional anatomical object from planar curves. The planar curves may be orthogonal to each other.

The medical diagnostic imaging system may also include an input unit for receiving input from a user. The input from a user being a contour of the anatomical object on a plurality of slices. The plurality of slices may comprise an axial slice, a sagittal slice, or a coronal slice, or some combination thereof. The medical diagnostic imaging system may also include a display unit for displaying a three-dimensional segmented anatomical object. The input unit may receive input from a tracing pen. The display unit may receive touch screen input.

The system and method described above may be carried out as part of a computer-readable storage medium including a set of instructions on a computer. The set of instructions may include an acquisition routine for acquiring at least one axial curve and one orthogonal curve from user input. An optional modification routine for modifying the orthogonal curve, the modification allowing the orthogonal curve to intersect with the axial curve. A generation routine for generating template shapes for axial slices. A location routine for locating attractor points. A computation routine for computing transformation parameters for axial slices lacking user-drawn curves. An optional smoothing routine for smoothing said transformation parameters. Finally, an execution routine for executing the transformation on the template shapes for the axial slices.

Additionally, the system and method described above may be carried out as part of a computer-readable storage medium including a set of instructions for a computer. The set of instructions may include an acquisition routine for acquiring at least one axial curve and one orthogonal curve from user input. The set of instructions may also include an optional modification routine for modifying the orthogonal curve, the modification allowing the orthogonal curve to intersect with the axial curve. The set of instructions may also include an generation routine for generating template shapes for axial slices. The set of instructions may also include an location routine for locating attractor points. The set of instructions may also include an computation routine for computing transformation parameters for axial slices lacking user-drawn curves. The set of instructions may also include an optional smoothing routine for smoothing said transformation parameters. The set of instructions may also include an execution routine for executing the transformation on the template shapes for said axial slices. Finally, the set of instructions may also include a second smoothing routine for smoothing the joints of said axial slices.

**BRIEF DESCRIPTION OF THE DRAWINGS**

*a*-*d *illustrates an example of the method described in

*a*-*d *illustrates, as an example of the method demonstrated in

**DETAILED DESCRIPTION OF THE INVENTION**

**100** for controlling the display and segmentation of medical images. The system **100** includes a computer unit **110**. The computer unit **110** may be any equipment or software that permits electronic medical images, such as x-rays, ultrasound, CT, MRI, EBT, MR, or nuclear medicine for example, to be electronically acquired, stored, or transmitted for viewing and operation. The computer unit **110** may be connected to other devices as part of an electronic network.

The system **100** also includes an input unit **120**. The input unit **120** may be a console having a track ball **122** and keyboard **124**. The input unit **120** may also have a tracing pen **126**. Other input devices may be used to receive input from a user as part of the input unit **120**. For example a microphone may be used to receive verbal input from a user. The tracing pen **126** may communicate with the input unit **120** through a wire. The tracing pen **126** may also communicate with the input unit **120** in a wireless fashion.

The system **100** also includes at least one display unit **130**. The display unit **130** may be a typical computer display unit. The display unit **130** may be in electrical communication with the computer unit **110** and input unit **120**. The display unit **130** may have the capability of transmitting touch screen input from the tracing pen **126** to either the input unit **120** or the computer unit **110**. For example, a user may use the tracing pen **126** to trace a curve on an image displayed on the display unit **130**. The location of the curve may then be transmitted to the computer unit **110** for processing.

In an embodiment, the display unit **130** may represent multiple display units or display regions of a screen. Accordingly, any number of display units may be utilized in accordance with the present invention. Additionally, the computer unit **110**, input unit **120**, and display unit **130** may be separate units or be part of a single unit. Accordingly, the components of the system **100** may be single units, separate units, may be integrated in various forms, and may be implemented in hardware and/or in software.

In operation, the system **100** may be a medical diagnostic imaging system. The computer unit **110** may execute computer software for segmenting anatomical objects from medical images. The computer software may extract an anatomical object from planar curves. Alternatively, the computer software may correct the shape of an existing three-dimensional anatomical object from planar curves. The planar curves may be orthogonal to each other. The input unit **120** may receive input from a user. In an embodiment, the input may be a contour of the anatomical object to be segmented. The contour may be drawn on a plurality of slices, for example on an axial slice, a sagittal slice, a coronal slice, or some combination thereof. The contour may be drawn using a tracing pen **126** on the display unit **130**. The display unit **130** may receive touch screen input from the tracing pen **126**. The display unit may display the three-dimensional segmented anatomical object.

**200** for segmenting anatomical objects from medical images. More specifically, the method **200** illustrates a three-dimensional interpolation method that may be used to speed up manual tracing of three-dimensional image structures or objects. The user may contour the anatomical structure on some of the axial, sagittal and/or coronal slices. A surface close to these curves may then be approximated. If the result is not acceptable, the user may contour on more slices at the critical area. The user may then rerun the extraction algorithm, which will provide more precise result. Thus the user is enabled to balance between precision and speed and can control the quality of result.

At step **210**, the computer unit **110** acquires at least one axial curve and one orthogonal curve from user input. In order to provide the curves, a user is free to choose any kind of drawing tools such as a manual freehand tool, a live wire tool to more easily follow the visible borders, a polygon/spline drawing tool by clicking the vertex/control points or other tools available to a user. In general, the orthogonal slices, on which the orthogonal curves may be drawn, may be sagittal and/or coronal slices. Accordingly, a contiguous sequence of contours is created, and such sequence may define a surface.

*a *illustrates an example of acquiring at least one axial curve and one orthogonal curve. In *a*, the “egg” curve is the input sagittal curve **310** drawn by a user on a sagittal slice. The generally horizontal curves are the input axial curves **315** drawn by a user on the axial slices.

At step **220**, if the orthogonal curve(s) and the axial curve(s) do not intersect, the orthogonal curve(s) may be modified so the orthogonal curve(s) intersects with the axial curve(s). When contours of a structure are precisely drawn, the curves on axial and orthogonal slices intersect. In *a*, the axial curves **315** and the sagittal curve **310** intersect. In practice, this is rarely true. *a*, the axial curve **415***a *exceeds the orthogonal curve **410***a. *The arrows **420***a *illustrate the difference. In *b*, the axial curve **415***b *is smaller than the orthogonal curve **410***b. *The arrows **420***b *illustrate the difference. In an embodiment, when input correction is requested, an algorithm modifies the orthogonal curves **410***a *and **410***b. *The orthogonal curves **410***a *and **410***b *are modified because it is assumed that the axial curves **415***a *and **415***b *are more precise.

At step **230**, template shapes are generated for axial slices. The template shapes are generated using shape-based interpolation. The template shapes are generated for axial slices lacking user-drawn curves. *b *illustrates step **230**. The axial template shapes **320** illustrated in *b *are generated for axial slices lacking user drawn curves. In the example provided, axial curves having user drawn curves are axial curves **315**. The axial curves **315** are shown in *b *along with the axial template shapes **320**, as is the orthogonal curve **310**. Accordingly, the axial template shapes **320** are an approximation of the axial curves **315**.

Next, at step **240**, the attractor points are located. The attractor points are the intersections of the orthogonal curves and the axial planes. In the example of *c*, the attractor points are pointed by the arrowheads **350**.

At step **250**, the transformation parameters are computed for axial slices lacking user-drawn curves. Various transformation methods may be used. For example, affine or spline transformation either slice by slice or directly in three-dimensions, which may provide a global transformation. In another embodiment, an active contour/deformable surface approach may be used either slice by slice or directly in three-dimensions, which may provide locally varying deformation. The points of the curves drawn on orthogonal slices are may be used as attractors to guide the deformation. In order to compute the transformation parameters, first, the input is complemented. If the sagittal or coronal input is missing, information about the missing input is gathered. For example, if there is only a sagittal input, information about the coronal direction is collected. Second, the closest contour points to the attractor points are computed on a slice-by-slice basis. In an embodiment, the algorithm computing the transformation parameters requires attractor point—closest contour point pairs. Finally, the parameters of the two-dimensional affine transformation, which moves the closest contour points into the attractor points, are computed. In an embodiment, the parameters are computed with the Iterative Closest Point algorithm. In the Iterative Closest Point algorithm, the closest points between two data sets are identified as corresponding points. A cost function depending on the distance between the closest points between the two data sets is minimized with respect to rotation, translation and shearing parameters. After determining these parameters, the resulting transformation can be applied to the respective data set, so that both data sets move closer to each other. In *c*, the arrows **350** illustrate the direction of deformation.

At step **260**, after the transformation parameters are computed, the transformation parameters are optionally smoothed in the case of two-dimensional—slice-by-slice—transformation. When the transformation parameters are computed slice-by-slice, results may be improved by smoothing the transformation parameters so that deformation of axial contours on subsequent slices vary smoothly. The smoothing weights may be 1, 2, 4, 2, 1, for example. For example, let the actual item be ai. The item ai gets weight 4, items ai−1 and ai+1 get 2 and item ai−2 and ai+2 get 1. Then we get the new ai value:

After the transformation parameters are computed and optionally smoothed, the transformation is executed at step **270** on the template shape for each axial slice. The axial templates **320** are deformed so that the axial templates **320** match the curve(s) **310** drawn on the orthogonal slice(s). In an embodiment, the deformation method is affine transformation. The results after the deformation of the axial curves are illustrated in *d. *

*a *illustrates user input curves. The two axial curves **510** are user input from axial slices. The coronal curve **520** is user input from a coronal slice. The sagittal curve **530** is user input from a sagittal slice. *b *illustrates the result of the segmentation of the bladder **550** after executing the three-dimensional interpolation method **200**.

**600** for editing segmented anatomical objects from medical images. More specifically, the method **600** illustrates a three-dimensional edit correction tool. In general, fully automatic segmentation methods usually do not give perfect results. There are cases, when the segmented surface does not go to the edges of the organ, or it may flow into other organs. Using this tool the user can correct the shape of an existing structure or object by providing some contours on some of the axial, sagittal and/or coronal slices. A surface close to the user drawn curves may then be approximated. If the result is not acceptable, the user may contour on some more slices at the critical area, and rerun the algorithm, which will provide more precise result. Thus the user is enabled to balance between precision and speed—can control the quality of result.

*a *illustrates an example of a liver segmentation that has missed the left lobe. Image **710***a *represents an axial image with the results of the liver segmentation outlined. Image **720***a *represents a coronal image with the results of the liver segmentation outlined. As is shown in both image **710***a *and **720***a*, the segmentation outline does not encompass the entire liver.

At step **610**, the computer unit **110** acquires at least one axial curve and one orthogonal curve from user input. In order to provide the curves, a user is free to choose any kind of drawing tools such as a manual freehand tool, a live wire tool to more easily follow the visible borders, a polygon/spline drawing tool by clicking the vertex/control points or other tools available to a user. In general, the orthogonal slices, on which the orthogonal curves may be drawn, may be sagittal and/or coronal slices. Accordingly, a contiguous sequence of contours is created, and such sequence may define a surface.

In **710***b *illustrates a user drawn curve **730** on the axial slice indicating the correct border of the organ. Image **720***c *illustrates a user drawn curve **740** on the coronal slice indicating the correct border of the organ. Accordingly, step **610** is satisfied with one axial input and one orthogonal input as shown in FIGS. **710** (*b*) and **720** (*c*).

Step, **620**, is similar to step **220** above. At step **620**, if the orthogonal curve(s) and the axial curve(s) do not intersect, the orthogonal curve(s) may be modified so the orthogonal curve(s) intersects with the axial curve(s).

Step **630** is similar to step **230** in the method **200**, with the modification that the generated shapes may be modified on the axial slices to reach the to be corrected object, resulting in an extended template shape. Step **640** is similar to step **240** in that the attractor points are located; the attractor points are the intersections of the orthogonal curves and the axial planes. In step **640**, however, two more attractor points are located as well, these are the closest points on the surface of the to be segmented object to the end-curve points of the axial curves.

**640** and method **600**. The closed line **810** is the segmented contour of the organ. The solid line **820** is the template curve that is generated from the user drawn axial input with shape-based interpolation. The dots **830** are the attractor points. The dots **840** are the contour points. The contour points **840** are the closest points on the template curve **820** to the attractor points **830**. In the example shown here, two of the **830** dots are on the closed line **810** and are the closest points to the curve-end points **840**. When the transformation is executed, the dashed line **850** becomes the new border of the organ.

Steps **650**, **660**, and **670** are similar to the steps **250**, **260**, and **270** as described above. In the method **600** after the transformation algorithm executes, an additional step of **680** is used. In step **680**, the joints of each slice are smoothed for the smooth transition between the original to be corrected object and the generated correcting object. Accordingly, after the execution of the method **600**, a corrected organ is generated as shown in *d. **d *illustrates a corrected axial image. *d *illustrates a corrected coronal image. In both **710***d *and **720***d*, the corrected images are outlined.

The system **100** and method **200** described above may be carried out as part of a computer-readable storage medium including a set of instructions for a computer. The set of instructions includes an acquisition routine to acquire at least one axial curve and one orthogonal curve from user input. The set of instructions also includes an optional modification routine to modify the orthogonal curve(s) so the axial and orthogonal curve(s) intersect. The set of instructions also includes a generation routine to generate template shapes for axial slices. The set of instructions also includes a location routine to locate the attractor points. The set of instructions also includes a computation routine to compute the transformation parameters for axial slices lacking user-drawn curves. The set of instructions also includes an optional smoothing routine for smoothing the transformation parameters. The set of instructions also includes an execution routine for executing the transformation on the template shape for each axial slice.

Additionally, the system **100** and method **600** described above may be carried out as part of a computer-readable storage medium including a set of instructions on a computer. The set of instructions includes an acquisition routine to acquire at least one axial curve and one orthogonal curve from user input. The set of instructions also includes an optional modification routine to modify the orthogonal curve(s) so the axial and orthogonal curve(s) intersect. The set of instructions also includes a generation routine to generate template shapes for axial slices. The set of instructions also includes a location routine to locate the attractor points. The set of instructions also includes a computation routine to compute the transformation parameters for axial slices lacking user-drawn curves. The set of instructions also includes an optional smoothing routine for smoothing the transformation parameters. The set of instructions also includes an execution routine for executing the transformation on the template shape for each axial slice. The set of instructions also includes a second smoothing routine for smoothing the joints of each slice for a smooth transition between the original to be corrected object and the generated correcting object.

While the invention has been described with reference to certain embodiments, it will be understood by those skilled in the art that various changes may be made and equivalents may be substituted without departing from the scope of the invention. In addition, many modifications may be made to adapt a particular situation or material to the teachings of the invention without departing from its scope. Therefore, it is intended that the invention not be limited to the particular embodiment disclosed, but that the invention will include all embodiments falling within the scope of the appended claims.

## Claims

1. A method for segmenting anatomical objects from medical images, said method comprising:

- acquiring at least one axial curve and at least one orthogonal curve from a set of medical images;

- optionally modifying said orthogonal curve to intersect said axial curve if said axial curve and said orthogonal curve, as acquired, do not intersect;

- generating template shapes for axial slices of said set of medical images using shape-based interpolation, said template shapes are generated for axial slices lacking user-drawn curves;

- locating attractor points by the intersections of the orthogonal curves and the axial planes;

- computing transformation parameters for axial slices lacking user-drawn curves; and

- executing the transformation on said template shapes for said axial slices.

2. The method of claim 1, wherein the step of computing transformation parameters for axial slices lacking user-drawn curves is performed by carrying out the steps of:

- optionally complementing the input if said input does not account for sagittal or coronal input;

- computing the closest contour points to the attractor points for each slice;

- iteratively computing the parameters of the two-dimensional affine transformation; and

- optionally smoothing the transformation parameters.

3. The method of claim 2, wherein said step of iteratively computing the parameters of the two-dimensional affine transformation is performed with the Iterative Closest Point algorithm.

4. The method of claim 1, wherein said orthogonal curve is a sagittal curve.

5. The method of claim 1, wherein said orthogonal curve is a coronal curve.

6. A method for correcting the shapes of existing three-dimensional anatomical objects from medical images, said method comprising:

- acquiring at least one axial curve and at least one orthogonal curve from set of medical images;

- optionally modifying said orthogonal curve to intersect said axial curve if said axial curve and said orthogonal curve, as acquired, do not intersect;

- generating template shapes for axial slices of said data set using shape-based interpolation, wherein said template shapes may be modified on the axial slices to reach the to be corrected object, and said template shapes are generated for axial slices lacking user-drawn curves;

- locating end-curve and other attractor points, end-curve attractor points located by finding the closest points on the surface of said object to the end points of the axial curves, other attractor points are the intersections of the orthogonal curves and the axial planes;

- computing transformation parameters for axial slices lacking user-drawn curves;

- executing the transformation on said template shape for said axial slices; and

- smoothing a plurality of joints on said original object and said corrected object.

7. The method of claim 6, wherein the step of computing transformation parameters for axial slices lacking user-drawn curves is performed by carrying out the steps of:

- optionally complementing the input if said input does not account for sagittal or coronal input;

- computing the closest contour points to the attractor points for each slice;

- iteratively computing the parameters of the two-dimensional affine transformation; and

- optionally smoothing the transformation parameters.

8. The method of claim 7, wherein said step of iteratively computing the parameters of the two-dimensional affine transformation is performed with the Iterative Closest Point algorithm.

9. The method of claim 6, wherein said orthogonal curve is a sagittal curve.

10. The method of claim 6, wherein said orthogonal curve is a coronal curve.

11. A medical diagnostic imaging system, said system comprising:

- a computer unit for manipulating data, said computer unit executing computer software for segmenting anatomical objects from medical images, said computer software extracting an anatomical object from planar curves, said planar curves being orthogonal to each other;

- an input unit for receiving input from a user, said input being a contour of said anatomical object on a plurality of slices; and,

- a display unit for displaying a three-dimensional segmented anatomical object.

12. The system of claim 11, wherein said plurality of slices comprise an axial slice.

13. The system of claim 11, wherein said plurality of slices comprise a sagittal slice.

14. The system of claim 11, wherein said plurality of slices comprise a coronal slice.

15. The system of claim 11, wherein said input unit includes a tracing pen.

16. The system of claim 11, wherein said display unit may receive touch screen input.

17. A medical diagnostic imaging system, said system comprising:

- a computer unit for manipulating data, said computer unit executing computer software for segmenting anatomical objects from medical images, said computer software correcting the shape of an existing three-dimensional anatomical object from planar curves, said planar curves being orthogonal to each other;

- an input unit for receiving input from a user, said input being a contour of said anatomical object on a plurality of slices; and

- a display unit for displaying a three-dimensional segmented anatomical object.

18. The system of claim 17, wherein said plurality of slices comprise an axial slice.

19. The system of claim 17, wherein said plurality of slices comprise a sagittal slice.

20. The system of claim 17, wherein said plurality of slices comprise a coronal slice.

21. The system of claim 17, wherein said input unit includes a tracing pen.

22. The system of claim 17 wherein said display unit may receive touch screen input.

23. A computer-readable storage medium including a set of instructions for a computer, the set of instructions comprising:

- an acquisition routine for acquiring at least one axial curve and one orthogonal curve from user input;

- an optional modification routine for modifying said orthogonal curve, said modification allowing said axial curve to intersect with said orthogonal curve;

- a generation routine for generating template shapes for axial slices;

- a location routine for locating attractor points;

- a computation routine for computing transformation parameters for axial slices lacking user-drawn curves; and

- an execution routine for executing the transformation on the template shapes for said axial slices.

24. A computer-readable storage medium including a set of instructions for a computer, the set of instructions comprising:

- an acquisition routine for acquiring at least one axial curve and one orthogonal curve from user input;

- an optional modification routine for modifying said orthogonal curve, said modification allowing said axial curve to intersect with said orthogonal curve;

- a generation routine for generating template shapes for axial slices;

- a location routine for locating end curve and attractor points;

- a computation routine for computing transformation parameters for axial slices lacking user-drawn curves;

- an execution routine for executing the transformation on the template shapes for said axial slices; and

- a smoothing routine for smoothing the joints of said axial slices.

**Patent History**

**Publication number**: 20070116334

**Type:**Application

**Filed**: Nov 22, 2005

**Publication Date**: May 24, 2007

**Patent Grant number**: 7773786

**Applicant**:

**Inventors**: Marta Fidrich (Szeged), Imre Pap (Doc), Attila Tanacs (Zakanyszek)

**Application Number**: 11/286,540

**Classifications**

**Current U.S. Class**:

**382/128.000;**382/154.000; 382/173.000

**International Classification**: G06K 9/00 (20060101); G06K 9/34 (20060101);