Framework for segmentation of cylindrical structures using two dimensional hybrid models

• US 6,201,543 B1
• Filed: 12/17/1997
• Issued: 03/13/2001
• Est. Priority Date: 12/17/1997
• Status: Expired due to Term

• Current Assignee: Siemens Medical Solutions USA Incorporated (Siemens AG)
• Original Assignee: Siemens Corporate Research Incorporated (Siemens AG)
• Inventors: OʼDonnell, Thomas, Jolly, Marie Pierre, Gupta, Alok
• Primary Examiner(s): Vo, Cliff N.
• Application Number:
  US08/992,869
• Time in Patent Office: 1,182 Days
• Field of Search: 345/420, 345/423, 345/442, 345/425, 345/418, 345/419, 345/424
• US Class Current: 345/420
• CPC Class Codes: G06T 17/20 Finite element generation, e.g. wire-frame surface description, tesselation
• CPC Class Codes: G06T 2207/10081 Computed x-ray tomography [CT]
• CPC Class Codes: G06T 2207/30101 Blood vessel; Artery; Vein; Vascular
• CPC Class Codes: G06T 7/0012 Biomedical image inspection
• CPC Class Codes: G06T 7/12 Edge-based segmentation
• CPC Class Codes: G06T 7/149 involving deformable models, e.g. active contour models
• CPC Class Codes: G06V 10/755 Deformable models or variational models, e.g. snakes or active contours

Abstract

A method of modeling a structure comprises the steps of initializing a parametric component of a hybrid model from a volumetric image, the volumetric image comprising image planes, creating starting points for two dimensional segmentations at intersections between the parametric component and the image planes, segmenting a boundary of the structure by applying active contours, sampling the active contours to create a dataset and parametrically and locally fitting the hybrid model to the dataset to represent the structure.

Claims

1. A method of modeling a structure comprising the steps of:

initializing a parametric component of a hybrid model from a vohumetric image, the volumetric image comprising image planes;

creating starting points for two dimensional segmentations at intersections between the parametric component and the image planes;

segmenting a boundary of the structure by applying active contours;

sampling the active contours to create a dataset;

parametrically and locally fitting the hybrid model to the dataset to represent the structure;

reducing a number of iterations to fit the hybrid model by introducing a finite element shape function for distributing forces to draw the hybrid model to data of the dataset wherein the forces include data forces and smoothing forces;

introducing a scale factor to weight data locations from the dataset to provide a scaled force vector for more rapidly fitting the hybrid model to the dataset; and

combining the data forces and the smoothing forces by introducing a trade-off parameter for balancing data forces against smoothing forces and thereby providing increased stability to the fitting of the hybrid model.

2. The method of modeling a structure as recited in claim 1 wherein the shape function is represented by a formula:

3. The method of modeling a structure as recited in claim 1 wherein the shape function is represented by a formula:

4. The method of modeling a structure as recited in claim 1 wherein the scale factor is represented by a formula:

$K_{\mathrm{data}}=\underset{i}{\max }<br><br>\underset{j=i}{\overset{M}{\sum <br><br>}}<br><br>N_i<br><br>\left(s<br><br>\left(j\right)\right)$

where i is a set of all nodes and M is a number of datapoints associated with a node i,j which has a closest point to an element s(j) in element coordinates and N_i is a shape function for node i with respect to element s(j).

5. The method of modeling a structure as recited in claim 1 wherein the dataset represents three dimensions.

6. A method of modeling branching cylindrical structures comprising the steps of:

initializing a parametric component of a three dimensional hybrid model from a volumetric image of the branching structure;

the volumetric image comprising image planes;

creating starting points for two dimensional segmentations at intersections between the branching structure and the image planes;

segmenting the branching structure by applying two dimensional active contours to the image planes to more accurately depict features of the branching structure;

sampling the active contours to create a dataset;

parametrically and locally fitting the hybrid model to the dataset to represent the branching structure by applying forces to draw the hybrid model onto the data of the dataset;

reducing a number of iterations to fit the hybrid model by introducing a finite element shape function for distributing forces to draw the hybrid to data of the dataset;

introducing a scale factor to weight data locations from the dataset to provide a scaled force vector for more rapidly fitting the hybrid model to the dataset, wherein the scale factor is represented by the formula;

$K_{\mathrm{data}}=\underset{i}{\max }<br><br>\underset{j=i}{\overset{M}{\sum <br><br>}}<br><br>N_i<br><br>\left(s<br><br>\left(j\right)\right)$

 

where i is a set of all nodes and M is a number of datapoints associated with a node i,j which has a closest point to an element s(j) in element coordinates and N_i is a shape function for node i with respect to element s(j).

7. The method of modeling branching cylindrical structures as recited in claim 6 wherein the dataset represents three dimensions.

8. The method of modeling branching cylindrical structures as recited in claim 11 wherein the shape function is represented by the formula:

9. The method of modeling branching cylindrical structures as recited in claim 11 wherein the shape function is represented by the formula:

10. The method of modeling branching cylindrical structures as recited in claim 6 wherein the forces include data forces and smoothing forces.

11. The method of modeling a structure as recited in claim 10 further comprising the step of combining the data forces and the smoothing forces by introducing a trade-off parameter τ

for balancing data forces against smoothing forces and thereby providing increased stability to the fitting of the hybrid model.

12. A method of modeling a structure, comprising the steps of:

initializing a parametric component of a hybrid model from a volumetric image, the volumetric image comprising image planes;

creating starting points for two dimensional segmentations at intersections between the parametric component and the image planes;

segmenting a boundary of the structure by applying active contours;

sampling the active contours to create a dataset;

parametrically and locally fitting the hybrid model to the dataset to represent the structure;

reducing a number of iterations to fit the hybrid model by introducing a finite element shape function for distributing forces to draw the hybrid model to data of the dataset, wherein the shape function is represented by a formula;

13. The method of modeling a structure as recited in claim 12 wherein the step of parametrically and locally fitting the hybrid model includes the step of applying forces to draw the hybrid model onto the data of the dataset.

14. The method of modeling a structure as recited in claim 13 wherein the forces include data forces and smoothing forces.

15. The method of modeling a structure as recited in claim 14, further comprising the step of combining the data forces and the smoothing forces by introducing a trade-off parameter for balancing data forces against smoothing forces and thereby providing increased stability to the fitting of the hybrid model.

16. The method of modeling a structure as recited in claim 12, further comprising the step of introducing a scale factor to weight data locations from the dataset to provide a scaled force vector for more rapidly fitting the hybrid model to the dataset.

17. The method of modeling a structure as recited in claim 16 wherein the scale factor is represented by a formula:

$K_{\mathrm{data}}=\underset{i}{\max }<br><br>\underset{j=i}{\overset{M}{\sum <br><br>}}<br><br>N_i<br><br>\left(s<br><br>\left(j\right)\right)$

where i is a set of all nodes and M is a number of datapoints associated with a node i,j which has a closest point to an element s(j) in element coordinates and N_i is a shape function for node i with respect to element s(j).

18. The method of modeling a structure as recited in claim 12 wherein the dataset represents three dimensions.

19. A method of modeling a structure comprising the steps of:

initializing a parametric component of a hybrid model from a volumetric image, the volumetric image comprising image planes;

creating starting points for two dimensional segmentations at intersections between the parametric component and the image planes;

segmenting a boundary of the structure by applying active contours;

sampling the active contours to create a dataset;

parametrically and locally fitting the hybrid model to the dataset to represent the structure;

reducing a number of iterations to fit the hybrid model by introducing a finite element shape function for distributing forces to draw the hybrid model to data of the dataset, wherein the shape function is represented by a formula;

20. The method of modeling a structure as recited in claim 19 wherein the step of parametrically and locally fitting the hybrid model includes the step of applying forces to draw the hybrid model onto the data of the dataset.

21. The method of modeling a structure as recited in claim 20 wherein the forces include data forces and smoothing forces.

22. The method of modeling a structure as recited in claim 21, further comprising the step of combining the data forces and the smoothing forces by introducing a trade-off parameter for balancing data forces against smoothing forces and thereby providing increased stability to the fitting of the hybrid model.

23. The method of modeling a structure as recited in claim 19, further comprising the step of introducing a scale factor to weight data locations from the dataset to provide a scaled force vector for more rapidly fitting the hybrid model to the dataset.

24. The method of modeling a structure as recited in claim 23 wherein the scale factor is represented by a formula:

$K_{\mathrm{data}}=\underset{i}{\max }<br><br>\underset{j=i}{\overset{M}{\sum <br><br>}}<br><br>N_i<br><br>\left(s<br><br>\left(j\right)\right)$

where i is a set of all nodes and M is a number of datapoints associated with a node i,j which has a closest point to an element s(j) in element coordinates and N_i is a shape function for node i with respect to element s(j).

25. The method of modeling a structure as recited in claim 19 wherein the dataset represents three dimensions.