Techniques for 3-D Elastic Spatial Registration of Multiple Modes of Measuring a Body

• US 20080265166A1
• Filed: 08/28/2006
• Published: 10/30/2008
• Est. Priority Date: 08/30/2005
• Status: Active Grant

• Current Assignee: Cleveland Clinic Foundation, IGI Medical Technologies, LLC
• Original Assignee: University of Maryland Baltimore County (University of Maryland)
• Inventors: Shekhar, Raj, Walimbe, Vivek
• Granted Patent: US 7,948,503 B2
• Time in Patent Office: Days
• US Class Current: 250/363.03
• CPC Class Codes: G01T 1/1611 using both transmission and emission sources sequentially SPECT imaging G01T1/1642; PET imaging G01T1/2985; detecting prohibited goods, e.g. weapons, explosives, hazardous substances, contraband or smuggled objects G01V5/20
• CPC Class Codes: G01T 1/1647 Processing of scintigraphic data not related to a particular imaging system G01T1/2992

Abstract

Techniques for registration of multiple measurement modes of a body include receiving first and second data from different modes. Each includes measured values with coordinate values. For two mechanically aligned modes, any nonrigid registration is performed. For some modes, the nonrigid registration includes a coarse transformation and multiple fine scale transformations. The coarse transformation maximizes a coarse similarity measure. The second data is subdivided into contiguous subregions. Fine transformations are determined between the subregions and corresponding portions of the first data to maximize a fine similarity measure. Subdividing and determining fine transformations repeats until stop conditions are satisfied. Transformations between the last divided subregions are interpolated. Any of the fine similarity measure, a search region, interpolation method, sub-division location, and the use of rigid or non-rigid fine transformations are adaptive to properties of the first or second data so that the registration is automatic without human intervention.

Claims

1. A method for automatic registration of multiple measurement modes of a target body, comprising the steps of:

receiving first data comprising a first plurality of measured values each associated with a plurality of coordinate values for three spatial dimensions based on a first measurement device for measuring a target body;

receiving second data comprising a second plurality of measured values each associated with a plurality of coordinate values for three spatial dimensions based on a different second measurement device for measuring the target body, wherein the first measurement device and the second measurement device are mounted on the same combined apparatus and are mechanically aligned;

determining a non-rigid transform between the first data and the second data without human intervention based on the first data and the second data, wherein the non-rigid transform determines changes in the plurality of coordinate values for each of the plurality of measured values of the second data to increase a measure of similarity between a measured value of the first data and a nearby measured value of the second data, which changes are not accounted for by a constant value for a three dimensional translation and a constant value for a three dimensional rotation over the plurality of measured values of the second data; and

transforming the plurality of coordinate values for each measured value of the second data based on the non-rigid transform.

2. A method as recited in claim 1, wherein the first measurement device is an X-ray computer tomography (CT) device and the second measurement device is a positron emission tomography (PET) measurement device on a hybrid CT-PET apparatus.

3. A method as recited in claim 1, said step of determining the non-rigid transform further comprising the steps of:

determining a coarse scale transformation to maximize a coarse scale similarity measure between the first data and the second data;

sub-dividing the second data into a plurality of sub-regions each comprising a spatially contiguous subset of the second data; and

determining a plurality of fine scale transformations to maximize a fine scale similarity measure between the plurality of sub-regions and a plurality of corresponding portions of the first data.

4. A method as recited in claim 3, wherein both the coarse scale transformation and the fine scale transformations are rigid-body translational and rotational transformations.

5. A method as recited in claim 3, wherein both the coarse scale similarity measure and the fine scale similarity measure are based on normalized mutual information.

6. A method as recited in claim 3, further comprising:

repeating said steps of sub-dividing the second data and determining a plurality of fine scale transformations until conditions for further sub-dividing the second data are no longer satisfied, wherein fine scale transformations for the last-formed sub-regions are each associated with a canonical position within a corresponding last-formed sub-region; and

interpolating fine scale transformations to positions between canonical positions.

7. A method as recited in claim 3, wherein:

the fine scale transformations include rotational transformations; and

said step of interpolating fine scale transformations further comprises expressing rotational transformations as quaternion values, and interpolating the quaternion values to positions between canonical positions.

8. A method for adaptive non-rigid registration of multiple measurement modes of a target body, comprising the steps of:

receiving first data comprising a first plurality of measured values each associated with a plurality of coordinate values for spatial or temporal dimensions based on a first measurement mode of a target body;

receiving second data comprising a second plurality of measured values each associated with a plurality of coordinate values for spatial or temporal dimensions based on a different second measurement mode of the target body;

determining a coarse scale transformation to maximize a coarse scale similarity measure between the first data and the second data;

sub-dividing the second data into a plurality of sub-regions each comprising a spatially or temporally contiguous subset of the second data;

determining an adaptive bin size for a mutual histogram based on a sample distribution for each sub-region of the plurality of sub-regions;

determining a fine scale similarity measure between the plurality of sub-regions and a plurality of corresponding portions of the first data based on the mutual histogram with the adaptive bin size;

determining a plurality of fine scale transformations to maximize the fine scale similarity measure; and

transforming the plurality of coordinate values for each measured value of the second data based on the plurality of fine scale transformations.

9. A method as recited in claim 8, wherein both the coarse scale transformation and the fine scale transformations are rigid-body translational and rotational transformations.

10. A method as recited in claim 8, wherein both the coarse scale similarity measure and the fine scale similarity measure are based on normalized mutual information.

11. A method as recited in claim 8, wherein the second measurement mode uses the same sensing system as the first measurement mode but at a different time.

12. A method as recited in claim 8, wherein the second measurement mode uses positron emission tomography (PET) and the first measurement mode uses X-ray computer tomography (CT).

13. A method as recited in claim 8, further comprising:

repeating said steps of sub-dividing the second data and determining a plurality of fine scale transformations until conditions for further sub-dividing the second data are no longer satisfied, wherein fine scale transformations for the last-formed sub-regions are each associated with a canonical position within a corresponding last-formed sub-region; and

interpolating fine scale transformations to positions between canonical positions.

14. A method as recited in claim 13, wherein:

the fine scale transformations include rotational transformations; and

said step of interpolating fine scale transformations further comprises expressing rotational transformations as quaternion values, and interpolating the quaternion values to positions between canonical positions.

15. A method for adaptive non-rigid registration of multiple measurement modes of a target body, comprising the steps of:

receiving first data comprising a first plurality of measured values each associated with a plurality of coordinate values for spatial or temporal dimensions based on a first measurement mode of a target body;

receiving second data comprising a second plurality of measured values each associated with a plurality of coordinate values for spatial or temporal dimensions based on a different second measurement mode of the target body;

determining a coarse scale transformation to maximize a coarse scale similarity measure between the first data and the second data;

sub-dividing the second data into a plurality of sub-regions each comprising a spatially or temporally contiguous subset of the second data;

determining a plurality of fine scale transformations to maximize within the search region a fine scale similarity measure between the plurality of sub-regions and a plurality of corresponding portions of the first data, wherein no fine scale transformation causes an effective displacement greater than a particular threshold; and

transforming the plurality of coordinate values for each measured value of the second data based on the plurality of fine scale transformations.

16. A method as recited in claim 15, wherein the particular threshold is 25 percent of the smallest sub-region of the plurality of subregions.

17. A method as recited in claim 15, wherein both the coarse scale transformation and the fine scale transformations are rigid-body translational and rotational transformations.

18. A method as recited in claim 15, wherein both the coarse scale similarity measure and the fine scale similarity measure are based on normalized mutual information.

19. A method as recited in claim 15, wherein the second measurement mode uses the same sensing system as the first measurement mode but at a different time.

20. A method as recited in claim 15, wherein the second measurement mode uses positron emission tomography (PET) and first measurement mode uses X-ray computer tomography (CT).

21. A method as recited in claim 15, further comprising:

repeating said steps of sub-dividing the second data and determining a plurality of fine scale transformations until conditions for further sub-dividing the second data are no longer satisfied, wherein fine scale transformations for the last-formed sub-regions are each associated with a canonical position within a corresponding last-formed sub-region; and

interpolating fine scale transformations to positions between canonical positions.

22. A method as recited in claim 21, wherein:

the fine scale transformations include rotational transformations; and

said step of interpolating fine scale transformations further comprises expressing rotational transformations as quaternion values, and interpolating the quaternion values to positions between canonical positions.

23. A method for adaptive non-rigid registration of multiple measurement modes of a target body, comprising the steps of:

receiving first data comprising a first plurality of measured values each associated with a plurality of coordinate values for spatial or temporal dimensions based on a first measurement mode of a target body;

receiving second data comprising a second plurality of measured values each associated with a plurality of coordinate values for spatial or temporal dimensions based on a different second measurement mode of the target body;

determining a linear transformation to maximize a coarse scale similarity measure between the first data and the second data;

sub-dividing the second data into a plurality of sub-regions each comprising a spatially or temporally contiguous subset of the second data;

determining a plurality of non-rigid transformations to maximize a fine scale similarity measure between the plurality of sub-regions and a plurality of corresponding portions of the first data; and

transforming the plurality of coordinate values for each measured value of the second data based on the plurality of non-rigid transformations.

24. A method as recited in claim 23, wherein the non-rigid transformations include shear and at least one of compression and expansion.

25. A method as recited in claim 23, wherein both the coarse scale similarity measure and the fine scale similarity measure are based on normalized mutual information.

26. A method as recited in claim 23, wherein the second measurement mode uses the same sensing system as the first measurement mode but at a different time.

27. A method as recited in claim 23, wherein the second measurement mode uses positron emission tomography (PET) and the first measurement mode uses X-ray computer tomography (CT).

28. A method as recited in claim 23, further comprising:

repeating said steps of sub-dividing the second data and determining a plurality of non-rigid transformations until conditions for further sub-dividing the second data are no longer satisfied, wherein non-rigid transformations for the last-formed sub-regions are each associated with a canonical position within a corresponding last-formed sub-region; and

interpolating fine scale transformations to positions between canonical positions.

29. A method for adaptive non-rigid registration of multiple measurement modes of a target body, comprising the steps of:

receiving first data comprising a first plurality of measured values each associated with a plurality of coordinate values for spatial or temporal dimensions based on a first measurement mode of a target body;

receiving second data comprising a second plurality of measured values each associated with a plurality of coordinate values for spatial or temporal dimensions based on a different second measurement mode of the target body, wherein coordinate values for each dimension fall in a range of values;

determining a coarse scale transformation to maximize a coarse scale similarity measure between the first data and the second data;

sub-dividing the second data into a plurality of sub-regions each comprising a spatially or temporally contiguous subset of the second data by dividing at least one of the spatial or temporal dimensions at a division position substantively different from a midpoint of the range of values for that dimension;

determining a plurality of fine scale transformations to maximize a fine scale similarity measure between the plurality of sub-regions and a plurality of corresponding portions of the first data; and

transforming the plurality of coordinate values for each measured value of the second data based on the plurality of fine scale transformations.

30. A method as recited in claim 29, wherein the division position is related to an automatically determined anatomical boundary.

31. A method as recited in claim 29, wherein both the coarse scale transformation and the fine scale transformations are rigid-body translational and rotational transformations.

32. A method as recited in claim 29, wherein both the coarse scale similarity measure and the fine scale similarity measure are based on normalized mutual information.

33. A method as recited in claim 29, wherein the second measurement mode uses the same sensing system as the first measurement mode but at a different time.

34. A method as recited in claim 29, wherein the second measurement mode uses positron emission tomography (PET) and the first measurement mode uses X-ray computer tomography (CT).

35. A method as recited in claim 29, further comprising:

repeating said steps of sub-dividing the second data by dividing at least one of the spatial or temporal dimensions at a division position substantively different from a midpoint of a range of values in that dimension for the sub-region and determining a plurality of fine scale transformations until conditions for further sub-dividing the second data are no longer satisfied, wherein fine scale transformations for the last-formed sub-regions are each associated with a canonical position within a corresponding last-formed sub-region; and

interpolating fine scale transformations to positions between canonical positions.

36. A method as recited in claim 35, wherein:

the fine scale transformations include rotational transformations; and

said step of interpolating fine scale transformations further comprises expressing rotational transformations as quaternion values, and interpolating the quaternion values to positions between canonical positions.

37. A computer-readable medium carrying one or more sequences of instructions for adaptive non-rigid registration of multiple measurement modes of a target body, wherein execution of the one or more sequences of instructions by one or more processors causes the one or more processors to perform the steps of:

receiving first data comprising a first plurality of measured values each associated with a plurality of coordinate values for three spatial dimensions based on a first measurement device for measuring a target body;

receiving second data comprising a second plurality of measured values each associated with a plurality of coordinate values for three spatial dimensions based on a different second measurement device for measuring the target body, wherein the first measurement device and the second measurement device are mounted on the same combined apparatus and are mechanically aligned;

determining a non-rigid transform between the first data and the second data without human intervention based on the first data and the second data, wherein the non-rigid transform determines changes in the plurality of coordinate values for each of the plurality of measured values of the second data to increase a measure of similarity between a measured value of the first data and a nearby measured value of the second data, which changes are not accounted for by a constant value for a three dimensional translation and a constant value for a three dimensional rotation over the plurality of measured values of the second data; and

transforming the plurality of coordinate values for each measured value of the second data based on the non-rigid transform.

38. A computer-readable medium carrying one or more sequences of instructions for adaptive non-rigid registration of multiple measurement modes of a target body, wherein execution of the one or more sequences of instructions by one or more processors causes the one or more processors to perform the steps of:

receiving first data comprising a first plurality of measured values each associated with a plurality of coordinate values for spatial or temporal dimensions based on a first measurement mode of a target body;

receiving second data comprising a second plurality of measured values each associated with a plurality of coordinate values for spatial or temporal dimensions based on a different second measurement mode of the target body;

determining a coarse scale transformation to maximize a coarse scale similarity measure between the first data and the second data;

sub-dividing the second data into a plurality of sub-regions each comprising a spatially or temporally contiguous subset of the second data;

determining an adaptive bin size for a mutual histogram based on a sample distribution for each sub-region of the plurality of sub-regions;

determining a fine scale similarity measure between the plurality of sub-regions and a plurality of corresponding portions of the first data based on the mutual histogram with the adaptive bin size;

determining a plurality of fine scale transformations to maximize the fine scale similarity measure; and

transforming the plurality of coordinate values for each measured value of the second data based on the plurality of fine scale transformations.

39. A computer-readable medium carrying one or more sequences of instructions for adaptive non-rigid registration of multiple measurement modes of a target body, wherein execution of the one or more sequences of instructions by one or more processors causes the one or more processors to perform the steps of:

receiving first data comprising a first plurality of measured values each associated with a plurality of coordinate values for spatial or temporal dimensions based on a first measurement mode of a target body;

receiving second data comprising a second plurality of measured values each associated with a plurality of coordinate values for spatial or temporal dimensions based on a different second measurement mode of the target body;

determining a coarse scale transformation to maximize a coarse scale similarity measure between the first data and the second data;

sub-dividing the second data into a plurality of sub-regions each comprising a spatially or temporally contiguous subset of the second data;

determining a plurality of fine scale transformations to maximize a fine scale similarity measure between the plurality of sub-regions and a plurality of corresponding portions of the first data, wherein no fine scale transformation causes an effective displacement greater than a particular threshold; and

transforming the plurality of coordinate values for each measured value of the second data based on the plurality of fine scale transformations.

40. A computer-readable medium carrying one or more sequences of instructions for adaptive non-rigid registration of multiple measurement modes of a target body, wherein execution of the one or more sequences of instructions by one or more processors causes the one or more processors to perform the steps of:

receiving first data comprising a first plurality of measured values each associated with a plurality of coordinate values for spatial or temporal dimensions based on a first measurement mode of a target body;

receiving second data comprising a second plurality of measured values each associated with a plurality of coordinate values for spatial or temporal dimensions based on a different second measurement mode of the target body;

determining a linear transformation to maximize a coarse scale similarity measure between the first data and the second data;

sub-dividing the second data into a plurality of sub-regions each comprising a spatially or temporally contiguous subset of the second data;

determining a plurality of non-rigid transformations to maximize a fine scale similarity measure between the plurality of sub-regions and a plurality of corresponding portions of the first data; and

transforming the plurality of coordinate values for each measured value of the second data based on the plurality of non-rigid transformations.

41. A computer-readable medium carrying one or more sequences of instructions for adaptive non-rigid registration of multiple measurement modes of a target body, wherein execution of the one or more sequences of instructions by one or more processors causes the one or more processors to perform the steps of:

receiving first data comprising a first plurality of measured values each associated with a plurality of coordinate values for spatial or temporal dimensions based on a first measurement mode of a target body;

receiving second data comprising a second plurality of measured values each associated with a plurality of coordinate values for spatial or temporal dimensions based on a different second measurement mode of the target body, wherein coordinate values for each dimension fall in a range of values;

determining a coarse scale transformation to maximize a coarse scale similarity measure between the first data and the second data;

sub-dividing the second data into a plurality of sub-regions each comprising a spatially or temporally contiguous subset of the second data by dividing at least one of the spatial or temporal dimensions at a division position substantively different from a midpoint of the range of values for that dimension;

determining a plurality of fine scale transformations to maximize a fine scale similarity measure between the plurality of sub-regions and a plurality of corresponding portions of the first data; and

transforming the plurality of coordinate values for each measured value of the second data based on the plurality of fine scale transformations.