System and method for 3D statistical shape model for the left ventricle of the heart

1. A method for generating a 3D statistical shape model for a left ventricle of a heart comprises the steps of:

generating an average left ventricle by aligning a plurality of datasets, selecting landmarks, and setting weights expressing confidence for each landmark; and

determining a statistical model from a set of aligned shape model by principle component analysis.

2. A method for generating a 3D statistical shape model for a left ventricle of a heart comprises the steps of:

creating a template model;

aligning a plurality of left ventricle contour datasets;

assigning a plurality of landmarks to the datasets;

averaging the landmarks; and

determining a statistical model from a plurality of aligned shape models.

3. The method of claim 2 , wherein the step of creating the template model further comprises the steps of:

determining a pose of the left ventricle according to a short axis and a lateral-to-septal-wall direction;

positioning a plurality of models in the dataset according to the pose;

fitting each model to a contour data, wherein the pose and a parametric component can change; and

averaging the parametric component for each model to create the template model.

4. The method of claim 2 , wherein the step of aligning further comprises the step of aligning the dataset to the template model.

5. The method of claim 4 , wherein the step of aligning the dataset to the template model further comprises the steps of:

associating a plurality of individual contour points in each dataset with a corresponding model;

isotropically scaling each model to fit the template model; and

scaling the dataset to fit the template model.

6. The method of claim 2 , wherein the step of assigning a plurality of landmarks to the datasets further comprises the step of associating model surfaces using the template model.

7. The method of claim 2 , wherein the step of assigning landmarks further comprises the steps of:

fitting, completely, a model to each dataset; and

determining a point on each of a plurality of surfaces of the model which is closest to each of a plurality of nodes of the template model.

8. The method of claim 7 , wherein the nodes are points wherein at least three polygons comprising the template model meet.

9. The method of claim 2 , wherein averaging the landmarks further comprise the steps of:

determining a confidence of each landmark;

scoring each confidence; and

averaging the landmarks according to a membership function of a scored value of the corresponding confidences.

10. The method of claim 2 , wherein the statistical model is determined according to a principle component analysis.

11. A program storage device readable by machine, tangibly embodying a program of instructions executable by the machine to perform method steps for generating a 3D statistical shape model for a left ventricle of a heart, the method steps comprising:

creating a template model;

aligning a plurality of left ventricle contour datasets;

assigning a plurality of landmarks to the datasets;

averaging the landmarks; and

determining a statistical model from a plurality of aligned shape models.

12. The method of claim 11 , wherein the step of creating the template model further comprises the steps of:

determining a pose of the left ventricle according to a short axis and a lateral-to-septal-wall direction;

positioning a plurality of models in the dataset according to the pose;

fitting each model to a contour data, wherein the pose and a parametric component can change; and

averaging the parametric component for each model to create the template model.

13. The method of claim 11 , wherein the step of aligning further comprises the step of aligning the dataset to the template model.

14. The method of claim 13 , wherein the step of aligning the dataset to the template model further comprises the steps of:

associating a plurality of individual contour points in each dataset with a corresponding model;

isotropically scaling each model to fit the template model; and

scaling the dataset to fit the template model.

15. The method of claim 11 , wherein the step of assigning a plurality of landmarks to the datasets further comprises the step of associating model surfaces using the template model.

16. The method of claim 11 , wherein the step of assigning landmarks further comprises the steps of:

fitting, completely, a model to each dataset; and

determining a point on each of a plurality of surfaces of the model which is closest to each of a plurality of nodes of the template model.

17. The method of claim 16 , wherein the nodes are points wherein at least three polygons comprising the template model meet.

18. The method of claim 11 , wherein averaging the landmarks further comprise the steps of:

determining a confidence of each landmark;

scoring each confidence; and

averaging the landmarks according to a membership function of a scored value of the corresponding confidences.

19. The method of claim 11 , wherein the statistical model is determined according to a principle component analysis.