Method and system for automatic quantification of aortic valve function from 4D computed tomography data using a physiological model

• US 8,009,887 B2
• Filed: 10/17/2008
• Issued: 08/30/2011
• Est. Priority Date: 11/02/2007
• Status: Active Grant

• Current Assignee: Siemens Healthineers AG (Siemens AG)
• Original Assignee: Siemens Corp. (Siemens AG)
• Inventors: Ionasec, Razvan, Houle, Helene C., Georgescu, Bogdan, Comaniciu, Dorin, Scheuering, Michael, Camus, Estelle, Vogt, Sebastian
• Primary Examiner(s): Tang; Minh N
• Application Number:
  US12/288,215
• Publication Number:
  US 20090123050A1
• Time in Patent Office: 1,047 Days
• Field of Search: None
• US Class Current: 382/128
• CPC Class Codes: G06T 17/20 Finite element generation, e.g. wire-frame surface description, tesselation
• CPC Class Codes: G06T 2207/10076 4D tomography; Time-sequential 3D tomography
• CPC Class Codes: G06T 2207/10136 3D ultrasound image
• CPC Class Codes: G06T 2207/20076 Probabilistic image processing
• CPC Class Codes: G06T 2207/20081 Training; Learning
• CPC Class Codes: G06T 2207/30048 Heart; Cardiac
• CPC Class Codes: G06T 7/0016 involving temporal comparison
• CPC Class Codes: G06T 7/251 involving models
• CPC Class Codes: G06V 10/471 using approximation functions
• CPC Class Codes: G06V 10/755 Deformable models or variational models, e.g. snakes or active contours
• CPC Class Codes: G06V 2201/031 of internal organs

Abstract

A method and system for modeling the aortic valve in 4D image data, such as 4D CT and echocardiography, is disclosed. An initial estimate of a physiological aortic valve model is determined for at least one reference frame of a 4D image sequence based on anatomic features in the reference frame. The initial estimate is refined to generate a final estimate in the reference frame. A dynamic model of the aortic valve is then generated by estimating the physiological aortic valve model for each remaining frame of the 4D image sequence based on the final estimate in the reference frame. The aortic valve can be quantitatively evaluated using the dynamic model.

Claims

1. A method for generating a dynamic aortic valve model, comprising:

receiving a 4D cardiac image sequence comprising a plurality of frames, each frame comprising 3D image data;

determining an initial estimate for a physiological aortic valve model in at least one reference frame of the plurality of frames based on anatomical landmarks in the at least one reference frame;

refining the initial estimate for the physiological aortic valve model in the at least one reference frame using boundary detection to generate a final estimate for the physiological aortic valve model in the at least one reference frame; and

generating a dynamic aortic valve model by estimating the physiological aortic valve model for remaining frames of the plurality of frames based on the final estimate for the physiological aortic valve model in the at least one reference frame.

2. The method of claim 1, wherein the physiological aortic valve model represents an aortic root and leaflets of the aortic valve by non-uniform rational B-splines (NURBS), and applies topological and geometrical constraints to the NURBS to define a physiologically compliant model of the aortic valve.

3. The method of claim 1, wherein said step of determining an initial estimate for a physiological aortic valve model in at least one reference frame of the plurality of frames based on anatomical landmarks in the at least one reference frame comprises:

detecting a plurality of anatomic features in the at least one reference frame; and

calculating a thin-plate-spline transformation based on the plurality of detected anatomic features; and

mapping the physiological aortic valve model to the at least one reference frame using the thin-plate-spline transformation.

4. The method of claim 3, wherein said step of detecting a plurality of anatomic features in the at least one reference frame comprises, for each of the plurality of anatomic features:

detecting candidates for a center of gravity of the anatomic feature at a low resolution in the at least one reference frame using a first trained detector;

detecting candidates for the center of gravity of the anatomic feature at a high resolution in the at least one reference frame using a second trained detector; and

determining a location of the anatomic feature in the at least one reference frame by calculating the mean of the candidates for the center of gravity of the anatomic feature at the high resolution.

5. The method of claim 4, wherein each of the first and second detectors are trained based on training data using a probabilistic boosting tree.

6. The method of claim 4, wherein the plurality of anatomic features comprises three commissure points, three hinge points, two ostia, and three leaflet tips.

7. The method of claim 1, wherein said step of refining the initial estimate for the physiological aortic valve model in the at least one reference frame using boundary detection to generate a final estimate for the physiological aortic valve model in the at least one reference frame comprises:

detecting local boundaries in the at least one reference frame at boundary locations of the initial estimate of the physiological aortic valve model using a trained boundary detector;

evolving boundaries in the initial estimate of the physiological aortic valve model based on the detected local boundaries; and

fitting the physiological aortic valve model to the evolved boundaries.

8. The method of claim 7, wherein the boundary detector is trained using a probabilistic boosting tree with steerable features.

9. The method of claim 7, wherein said step of fitting the physiological aortic valve model to the evolved boundaries comprises:

fitting the physiological aortic valve model to the evolved boundaries using linear least squares approximation.

10. The method of claim 1, wherein said step of generating a dynamic aortic valve model by estimating the physiological aortic valve model for remaining frames of the plurality of frames based on the final estimate for the physiological aortic valve model in the at least one reference frame comprises:

determining an initial estimate for the physiological aortic valve model for each of the remaining frames in the plurality of frames based on the final estimate for the physiological aortic valve model in the at least one reference frame; and

refining the initial estimate for the physiological aortic valve model in each of the remaining frames in the plurality of frames using boundary detection to generate a final estimate for the physiological aortic valve model in each of the remaining frames.

11. The method of claim 10, wherein the at least one reference frame comprises a first reference frame and a second reference frame representing end-diastole and end-systole phases of a cardiac cycle, respectively, and said step of determining an initial estimate for the physiological aortic valve model for each of the remaining frames in the plurality of frames based on the final estimate for the physiological aortic valve model in the at least one reference frame comprises:

determining the initial estimate for the physiological aortic valve model for each of the remaining frames in the plurality of frames as a linear combination of the final estimate for the physiological aortic valve model in the first and second reference frames.

12. The method of claim 1, further comprising:

quantitatively evaluating the aortic valve using the dynamic aortic valve model.

13. The method of claim 12, wherein said step of quantitatively evaluating the aortic valve using the dynamic aortic valve model comprises:

dynamically measuring at least one of root diameter of the aortic valve, leaflet position of the aortic valve, root area of the aortic valve, coaptation distance relative to a hinge plane, inter-commissural distance, ostium height relative to the hinge plane, and commissure height relative to the hinge plane, over a cardiac cycle.

14. The method of claim 1, wherein said 4D image sequence is a 4D computed tomography (CT) sequence.

15. The method of claim 1, wherein said 4D image sequence is echocardiography image data.

16. An apparatus for generating a dynamic aortic valve model, comprising:

means for receiving a 4D cardiac image sequence comprising a plurality of frames, each frame comprising 3D image data;

means for determining an initial estimate for a physiological aortic valve model in at least one reference frame of the plurality of frames based on anatomical landmarks in the at least one reference frame;

means for refining the initial estimate for the physiological aortic valve model in the at least one reference frame using boundary detection to generate a final estimate for the physiological aortic valve model in the at least one reference frame; and

means for generating a dynamic aortic valve model by estimating the physiological aortic valve model for remaining frames of the plurality of frames based on the final estimate for the physiological aortic valve model in the at least one reference frame.

17. The apparatus of claim 16, wherein the physiological aortic valve model represents an aortic root and leaflets of the aortic valve by non-uniform rational B-splines (NURBS), and applies topological and geometrical constraints to the NURBS to define a physiologically compliant model of the aortic valve.

18. The apparatus of claim 16, wherein said means for determining an initial estimate for a physiological aortic valve model in at least one reference frame of the plurality of frames based on anatomical landmarks in the at least one reference frame comprises:

means for detecting a plurality of anatomic features in the at least one reference frame; and

means for calculating a thin-plate-spline transformation based on the plurality of detected anatomic features; and

means for mapping the physiological aortic valve model to the at least one reference frame using the thin-plate-spline transformation.

19. The apparatus of claim 18, wherein said means for detecting a plurality of anatomic features in the at least one reference frame comprises:

means for detecting candidates for a center of gravity of an anatomic feature at a low resolution in the at least one reference frame using a first trained detector;

means for detecting candidates for the center of gravity of the anatomic feature at a high resolution in the at least one reference frame using a second trained detector; and

means for determining a location of the anatomic feature in the at least one reference frame by calculating the mean of the candidates for the center of gravity of the anatomic feature at the high resolution.

20. The apparatus of claim 16, wherein said means for refining the initial estimate for the physiological aortic valve model in the at least one reference frame using boundary detection to generate a final estimate for the physiological aortic valve model in the at least one reference frame comprises:

means for detecting local boundaries in the at least one reference frame at boundary locations of the initial estimate of the physiological aortic valve model using a trained boundary detector;

means for evolving boundaries in the initial estimate of the physiological aortic valve model based on the detected local boundaries; and

means for fitting the physiological aortic valve model to the evolved boundaries.

21. The apparatus of claim 16, wherein said means for generating a dynamic aortic valve model by estimating the physiological aortic valve model for remaining frames of the plurality of frames based on the final estimate for the physiological aortic valve model in the at least one reference frame comprises:

means for determining an initial estimate for the physiological aortic valve model for each of the remaining frames in the plurality of frames based on the final estimate for the physiological aortic valve model in the at least one reference frame; and

means for refining the initial estimate for the physiological aortic valve model in each of the remaining frames in the plurality of frames using boundary detection to generate a final estimate for the physiological aortic valve model in each of the remaining frames.

22. The apparatus of claim 21, wherein the at least one reference frame comprises a first reference frame and a second reference frame representing end-diastole and end-systole phases of a cardiac cycle, respectively, and said means for determining an initial estimate for the physiological aortic valve model for each of the remaining frames in the plurality of frames based on the final estimate for the physiological aortic valve model in the at least one reference frame comprises:

means for determining the initial estimate for the physiological aortic valve model for each of the remaining frames in the plurality of frames as a linear combination of the final estimate for the physiological aortic valve model in the first and second reference frames.

23. The apparatus of claim 16, further comprising:

means for quantitatively evaluating the aortic valve using the dynamic aortic valve model.

24. A computer readable medium encoded with computer executable instructions for generating a dynamic aortic valve model, the computer executable instructions defining steps comprising:

receiving a 4D cardiac image sequence comprising a plurality of frames, each frame comprising 3D image data;

determining an initial estimate for a physiological aortic valve model in at least one reference frame of the plurality of frames based on anatomical landmarks in the at least one reference frame;

refining the initial estimate for the physiological aortic valve model in the at least one reference frame using boundary detection to generate a final estimate for the physiological aortic valve model in the at least one reference frame; and

generating a dynamic aortic valve model by estimating the physiological aortic valve model for remaining frames of the plurality of frames based on the final estimate for the physiological aortic valve model in the at least one reference frame.

25. The computer readable medium of claim 24, wherein the physiological aortic valve model represents an aortic root and leaflets of the aortic valve by non-uniform rational B-splines (NURBS), and applies topological and geometrical constraints to the NURBS to define a physiologically compliant model of the aortic valve.

26. The computer readable medium of claim 24, wherein the computer executable instructions defining the step of determining an initial estimate for a physiological aortic valve model in at least one reference frame of the plurality of frames based on anatomical landmarks in the at least one reference frame comprise computer executable instructions defining the steps of:

detecting a plurality of anatomic features in the at least one reference frame; and

calculating a thin-plate-spline transformation based on the plurality of detected anatomic features; and

mapping the physiological aortic valve model to the at least one reference frame using the thin-plate-spline transformation.

27. The computer readable medium of claim 26, wherein the computer executable instructions defining the step of detecting a plurality of anatomic features in the at least one reference frame comprise computer executable instructions for defining the following steps for each of the plurality of anatomic features:

detecting candidates for a center of gravity of the anatomic feature at a low resolution in the at least one reference frame using a first trained detector;

detecting candidates for the center of gravity of the anatomic feature at a high resolution in the at least one reference frame using a second trained detector; and

determining a location of the anatomic feature in the at least one reference frame by calculating the mean of the candidates for the center of gravity of the anatomic feature at the high resolution.

28. The computer readable medium of claim 24, wherein the computer executable instructions defining the step of refining the initial estimate for the physiological aortic valve model in the at least one reference frame using boundary detection to generate a final estimate for the physiological aortic valve model in the at least one reference frame comprise computer executable instructions defining the steps of:

detecting local boundaries in the at least one reference frame at boundary locations of the initial estimate of the physiological aortic valve model using a trained boundary detector;

evolving boundaries in the initial estimate of the physiological aortic valve model based on the detected local boundaries; and

fitting the physiological aortic valve model to the evolved boundaries.

29. The computer readable medium of claim 24, wherein the computer executable instructions defining the step of generating a dynamic aortic valve model by estimating the physiological aortic valve model for remaining frames of the plurality of frames based on the final estimate for the physiological aortic valve model in the at least one reference frame comprise computer executable instructions defining the steps of:

determining an initial estimate for the physiological aortic valve model for each of the remaining frames in the plurality of frames based on the final estimate for the physiological aortic valve model in the at least one reference frame; and

refining the initial estimate for the physiological aortic valve model in each of the remaining frames in the plurality of frames using boundary detection to generate a final estimate for the physiological aortic valve model in each of the remaining frames.

30. The computer readable medium of claim 29, wherein the at least one reference frame comprises a first reference frame and a second reference frame representing end-diastole and end-systole phases of a cardiac cycle, respectively, and the computer executable instructions defining the step of determining an initial estimate for the physiological aortic valve model for each of the remaining frames in the plurality of frames based on the final estimate for the physiological aortic valve model in the at least one reference frame comprise computer executable instructions defining the step of:

determining the initial estimate for the physiological aortic valve model for each of the remaining frames in the plurality of frames as a linear combination of the final estimate for the physiological aortic valve model in the first and second reference frames.

31. The computer readable medium of claim 24, further comprising computer executable instructions defining the step of:

quantitatively evaluating the aortic valve using the dynamic aortic valve model.