Computation of a geometric parameter of a cardiac chamber from a cardiac tomography data set

• US 20050201598A1
• Filed: 02/08/2005
• Published: 09/15/2005
• Est. Priority Date: 02/09/2004
• Status: Active Grant

• Current Assignee: Institut De Cardiologie De Montrã©Al
• Original Assignee: Institut De Cardiologie De Montrã©Al
• Inventors: Finnerty, Vincent, Harel, Francois
• Granted Patent: US 7,430,309 B2
• Time in Patent Office: Days
• US Class Current: 382/128
• CPC Class Codes: G06T 2207/30048 Heart; Cardiac
• CPC Class Codes: G06T 7/0012 Biomedical image inspection
• CPC Class Codes: G06T 7/62 of area, perimeter, diameter or volume
• CPC Class Codes: Y10S 378/901 Computer tomography program or processor

Abstract

A method for extracting a geometric parameter from a cardiac tomography data set. The method includes providing a cardiac tomography data set and a region of interest, computing an approximative model of a surface of interest, computing a plurality of normal vectors associated with a respective surface coordinate from the approximative model of the surface of interest, computing a plurality of biopsy data sets, each biopsy data set being oriented along a respective normal vector from the plurality of normal vectors, computing an exact model of the surface of interest from the plurality of biopsy data sets, and computing the geometric parameter on a basis of the exact model of the surface of interest.

Claims

1. ) A method for extracting a geometric parameter from a cardiac tomography data set, said method comprising the steps of:

providing a cardiac tomography data set including a plurality of voxels, each voxel including a respective tri-dimensional position vector and a respective intensity value;

providing a region of interest including a plurality of voxels selected from the cardiac tomography data set, the region of interest including at least in part voxels from the cardiac tomography data set representative of a cardiac chamber of interest, the cardiac chamber of interest having a surface of interest;

computing an approximative model of the surface of interest on a basis of the region of interest using a non-parametric method, the approximative model of the surface of interest including a plurality of surface coordinates tri-dimensional vectors;

computing a plurality of normal vectors, each normal vector being associated with a respective surface coordinate from the approximative model of the surface of interest;

computing a plurality of biopsy data sets, each biopsy data set including a plurality of data points, each biopsy data set being representative of a plurality of voxels oriented in a direction defined by a respective normal vector selected from the plurality of normal vectors and including the surface coordinate associated with the respective normal vector;

computing an exact model of at least a portion of the surface of interest on a basis of the plurality of biopsy data sets, the exact model of the portion of the surface of interest including a plurality of surface coordinates tri-dimensional vectors, the exact model of the portion of the surface of interest being computed by expanding locally the approximative model of the surface of interest in directions defined by the plurality of normal vectors; and

computing the geometric parameter on a basis of at least one surface coordinate from the plurality of surface coordinates included in the exact model of the portion of the surface of interest.

2. ) A method as defined in claim 1, wherein said step of providing a cardiac tomography data set includes performing a coordinate system standardisation wherein the tri-dimensional position vector of each voxel is mapped to a standardised coordinate system.

3. ) A method as defined in claim 2, wherein the coordinate system standardisation is performed at least on a basis of a detection of the plane separating the atriums from the ventricles in the cardiac tomography data set.

4. ) A method as defined in claim 3, wherein in the standardised coordinate system, the plane separating the atriums from the ventricles includes an origin of the standardised coordinate system.

5. ) A method as defined in claim 1, wherein said step of providing a region of interest including a plurality of voxels selected from the cardiac tomography data set includes segmenting the cardiac tomography data set using a watershed method.

6. ) A method as defined in claim 1, wherein the approximative model of the surface of interest includes an isosurface computed from the cardiac tomography data set.

7. ) A method as defined in claim 6, wherein the isosurface substantially delimits a subset of the cardiac tomography data set including voxels having an intensity above a predetermined threshold and located within the cardiac chamber of interest.

8. ) A method as defined in claim 7, wherein the predetermined threshold is a predetermined fraction of the maximal intensity of all the voxels contained within the region of interest.

9. ) A method as defined in claim 8, wherein the predetermined fraction of the maximal intensity is from about 0.2 of the maximal intensity to about 0.5 of the maximal intensity.

10. ) A method as defined in claim 9, wherein the predetermined fraction of the maximal intensity is about 0.4 of the maximal intensity.

11. ) A method as defined in claim 1, wherein said step of computing an exact model of the surface of interest on a basis of the plurality of biopsy data sets includes applying a gradient method to each biopsy data set from the plurality of biopsy data sets.

12. ) A method as defined in claim 1, wherein:

said step of computing a plurality of biopsy data sets includes computing a laplacian of the cardiac tomography data set; and

the exact model of the portion of the surface of interest including a plurality of surface coordinates tri-dimensional vectors at which the laplacian of the cardiac tomography data set is substantially equal to zero.

13. ) A method as defined in claim 1, wherein said step of computing a plurality of biopsy data sets includes computing a multiresolution data set from said cardiac tomography data set.

14. ) A method as defined in claim 13, wherein said step of computing a plurality of biopsy data sets includes computing a multiresolution laplacian data set from said cardiac tomography data set.

15. ) A method as defined in claim 14, wherein each of the biopsy data set is representative the variance of the multiresolution laplacian data set at a plurality of voxels oriented in a direction defined by a respective normal vector selected from the plurality of normal vectors and including the surface coordinate associated with the respective normal vector.

16. ) A method as defined in claim 15, wherein the exact model of the portion of the surface of interest including a plurality of surface coordinates tri-dimensional vectors at which the variance of the multiresolution laplacian data set is substantially minimal.

17. ) A method as defined in claim 1, wherein the portion of the surface of interest includes a surface enclosing completely a volume of interest.

18. ) A method as defined in claim 17, wherein said step of computing the geometric parameter includes computing an ejection fraction for the cardiac chamber of interest.

19. ) A method as defined in claim 17, wherein:

the cardiac tomography data set includes a plurality of voxels, each voxel including a respective tri-dimensional position, a respective time stamp and a respective intensity value;

said steps of providing a region of interest including a plurality of voxels selected from the cardiac tomography data set;

computing an approximative model of the surface of interest on a basis of the region of interest using a non-parametric method;

computing a plurality of normal vectors;

computing a plurality of biopsy data sets; and

computing an exact model of the surface of interest on a basis of the plurality of biopsy data sets are performed for a plurality of simultaneously acquired sets of voxels, each of the simultaneously acquired sets of voxels including all the voxels from the cardiac tomography data set having a substantially identical time stamp; and

said step of computing an exact model of the surface of interest produces a plurality of simultaneous models of the surface of interest, each simultaneous model of the surface of interest being an exact model of the surface of interest computed from a respective simultaneously acquired sets of voxels.

20. ) A method as defined in claim 19, wherein said step of computing the geometric parameter includes performing a frequency-domain analysis of the plurality of simultaneous models of the surface of interest.

21. ) A method as defined in claim 20, wherein the geometric parameter includes an amplitude of a Fourier transform.

22. ) A method as defined in claim 20, wherein the geometric parameter includes a phase of a Fourier transform.

23. ) A method as defined in claim 20, wherein the geometric parameter includes a phase difference between the Fourier transforms of corresponding surface portions from two different cardiac chambers.

24. ) A method as defined in claim 23, wherein the geometric parameter includes a phase difference between the Fourier transforms of the coordinates of two different locations on the surface of interest.

25. ) A computer readable storage medium containing a program element for execution by a computing device for extracting a geometric parameter from a cardiac tomography data set, said program element comprising:

An input module operative for receiving;

i) a cardiac tomography data set including a plurality of voxels, each voxel including a respective tri-dimensional position vector and a respective intensity value; and

ii) a region of interest including a plurality of voxels selected from the cardiac tomography data set, the region of interest including at least in part voxels from the cardiac tomography data set representative of a cardiac chamber of interest, the cardiac chamber of interest having a surface of interest;

a preprocessing module operative for;

i) computing an approximative model of the surface of interest on a basis of the region of interest using a non-parametric method, the approximative model of the surface of interest including a plurality of surface coordinates tri-dimensional vectors;

ii) computing a plurality of normal vectors, each normal vector being associated with a respective surface coordinate from the approximative model of the surface of interest;

iii) computing a plurality of biopsy data sets, each biopsy data set including a plurality of data points, each biopsy data set being representative of a plurality of voxels oriented in a direction defined by a respective normal vector selected from the plurality of normal vectors and including the surface coordinate associated with the respective normal vector; and

iv) computing an exact model of the surface of interest on a basis of the plurality of biopsy data sets, the exact model of the surface of interest including a plurality of surface coordinates tri-dimensional vectors, the exact model of the surface of interest being computed by expanding locally the approximative model of the surface of interest in directions defined by the plurality of normal vectors; and

A geometry module operative for computing and outputting the geometric parameter on a basis of at least one surface coordinate from the plurality of surface coordinates included in the exact model of the surface of interest.

26. ) A cardiac tomograph, said cardiac tomograph comprising:

a tomography data acquisition system for acquiring a cardiac tomography data set including a plurality of voxels, each voxel including a respective tri-dimensional position vector and a respective intensity value;

a storage medium connected to said tomography data acquisition system; and

a computing device connected to said tomography data acquisition system and to said storage medium;

wherein said computer readable storage medium contains a program element for execution by said computing device, said program element being adapted to extract a geometric parameter from the cardiac tomography data set, said program element including;

i) An input module operative for receiving;

(1) the cardiac tomography data set; and

(2) a region of interest including a plurality of voxels selected from the cardiac tomography data set, the region of interest including at least in part voxels from the cardiac tomography data set representative of a cardiac chamber of interest, the cardiac chamber of interest having a surface of interest;

ii) a preprocessing module operative for;

(1) computing an approximative model of the surface of interest on a basis of the region of interest using a non-parametric method, the approximative model of the surface of interest including a plurality of surface coordinates tri-dimensional vectors;

(2) computing a plurality of normal vectors, each normal vector being associated with a respective surface coordinate from the approximative model of the surface of interest;

(3) computing a plurality of biopsy data sets, each biopsy data set including a plurality of data points, each biopsy data set being representative of a plurality of voxels oriented in a direction defined by a respective normal vector selected from the plurality of normal vectors and including the surface coordinate associated with the respective normal vector; and

(4) computing an exact model of the surface of interest on a basis of the plurality of biopsy data sets, the exact model of the surface of interest including a plurality of surface coordinates tri-dimensional vectors, the exact model of the surface of interest being computed by expanding locally the approximative model of the surface of interest in directions defined by the plurality of normal vectors; and

iii) A geometry module operative for computing and outputting the geometric parameter on a basis of at least one surface coordinate from the plurality of surface coordinates included in the exact model of the surface of interest.

27. ) A method for performing a segmentation of a multi-dimensional data set, said method comprising the steps of:

providing a multi-dimensional data set including a plurality of voxels, each voxel including a respective multi-dimensional position vector and a respective intensity value;

providing a region of interest including a plurality of voxels selected from the multi-dimensional data set, the region of interest defining a surface of interest;

computing an approximative model of the surface of interest on a basis of the region of interest using a non-parametric method, the approximative model of the surface of interest including a plurality of surface coordinates multi-dimensional vectors;

computing a plurality of normal vectors, each normal vector being associated with a respective surface coordinate from the approximative model of the surface of interest;

computing a plurality of biopsy data sets, each biopsy data set including a plurality of data points, each biopsy data set being representative of a plurality of voxels oriented in a direction defined by a respective normal vector selected from the plurality of normal vectors and including the surface coordinate associated with the respective normal vector, wherein computing a plurality of biopsy data sets includes computing a multiresolution laplacian data set from the multi-dimensional data set, each of the biopsy data set being representative of the variance of the multiresolution laplacian data set at a plurality of voxels oriented in a direction defined by a respective normal vector selected from the plurality of normal vectors and including the surface coordinate associated with the respective normal vector; and

computing an exact model of at least a portion of the surface of interest on a basis of the plurality of biopsy data sets, the exact model of the portion of the surface of interest including a plurality of surface coordinates multi-dimensional vectors, the exact model of the portion of the surface of interest being computed by expanding locally the approximative model of the surface of interest in directions defined by the plurality of normal vectors, the exact model of the portion of the surface of interest including a plurality of surface coordinates multi-dimensional vectors at which the variance of the multiresolution laplacian data set is substantially minimal.