Image Registration Method and Apparatus for Medical Imaging Based on Multiple Masks

• US 20080049994A1
• Filed: 08/05/2005
• Published: 02/28/2008
• Est. Priority Date: 08/09/2004
• Status: Active Grant

• Current Assignee: Bracco Suisse SA
• Original Assignee: Bracco International B.V. (Bracco SPA)
• Inventors: Ruth, Campos, Thiran, Jean-Philippe, Rognin, Nicolas, Messager, Tristan, Arditi, Marcel
• Granted Patent: US 7,912,259 B2
• Time in Patent Office: Days
• US Class Current: 382/128
• CPC Class Codes: G06T 2207/30048 Heart; Cardiac
• CPC Class Codes: G06T 7/35 using statistical methods

Abstract

An image registration method for use in medical imaging is proposed. The method includes the steps of: providing a sequence of images each one including a digital representation of a body-part under analysis, selecting a reference image within the sequence, the remaining images of the sequence defining moving images, and re-aligning at least one portion of a moving image with respect to the reference image. The step of re-aligning includes: defining a delimitation mask identifying a region on the reference image with which the at least one portion of the moving image has to be re-aligned, and a feature mask identifying a further region on the reference image within which the re-alignment is calculated, determining an optimized transformation for compensating a displacement of the moving image with respect to the reference image by optimizing a similarity measure, and transforming the at least one portion of the moving image according to the optimized transformation. The similarity measure is calculated alternatively between: a) a first computation region identified on the reference image by a computation mask and a second computation region identified by the computation mask on the moving image transformed according to a proposed transformation, the computation mask being determined by the intersection between the feature mask and the delimitation mask transformed according to the proposed transformation, or b) a first computation region identified by a computation mask on the reference image transformed according to a proposed transformation and a second computation region identified by the computation mask on the moving image, the computation mask being determined by the intersection between the feature mask transformed according to the proposed transformation and the delimitation mask, or c) a first computation region identified on the reference image by a computation mask transformed according to the reverse of a proposed transformation and a second computation region identified by the computation mask on the moving image, the computation mask being determined by the intersection between the feature mask transformed according to the proposed transformation and the delimitation mask.

Claims

1. An image registration method for use in medical imaging, the method including the steps of:

providing a sequence of images each one including a digital representation of a body-part under analysis,

selecting a reference image within the sequence, the remaining images of the sequence defining moving images, and

re-aligning at least one portion of a moving image with respect to the reference image,

wherein the step of re-aligning includes;

a) defining a delimitation mask identifying a region on the reference image with which the at least one portion of the moving image has to be re-aligned, and a feature mask identifying a further region on the reference image within which the re-alignment is calculated,

b) determining an optimized transformation for compensating a displacement of the moving image with respect to the reference image by optimizing a similarity measure between;

b1) a first computation region identified on the reference image by a computation mask and a second computation region identified by the computation mask on the moving image transformed according to a proposed transformation, the computation mask being determined by the intersection between the delimitation mask transformed according to the proposed transformation and the feature mask, or

b2) a first computation region identified by a computation mask on the reference image transformed according to a proposed transformation and a second computation region identified by the computation mask on the moving image, the computation mask being determined by the intersection between the feature mask transformed according to the proposed transformation and the delimitation mask, or

b3) a first computation region identified on the reference image by a computation mask transformed according to the reverse of a proposed transformation and a second computation region identified by the computation mask on the moving image, the computation mask being determined by the intersection between the feature mask transformed according to the proposed transformation and the delimitation mask, and

c) transforming the at least one portion of the moving image according to the optimized transformation.

2. The method according to claim 1, wherein the step c) of transforming includes:

c1) applying the optimized transformation to the at least one portion of the moving image when the optimized transformation is determined according to step b1), or

c2) applying the reverse of the optimized transformation to the at least one portion of the moving image when the optimized transformation is determined according to steps b2) or b3).

3. The method according to claim 1, wherein the step of determining the optimized transformation includes the iteration of calculating the similarity measure corresponding to the proposed transformation until the similarity measure or a change thereof reaches a threshold value.

4. The method according to claim 1, further including the iteration of the step of re-aligning for each further moving image of the sequence.

5. The method according to claim 1, wherein the feature mask is defined inside the delimitation mask.

6. The method according to claim 1, wherein the feature mask has a size larger than 50% of a size of the delimitation mask.

7. The method according to claim 1, wherein the similarity measure is a mutual information measure.

8. The method according to claim 1, further including the steps of:

estimating a spatial resolution along each dimension of the reference image,

calculating a sub-sampling factor for each dimension according to the spatial resolution, and

sub-sampling at least part of each image according to the sub-sampling factors.

9. The method according to claim 8, wherein the step of estimating the spatial resolution includes:

determining an estimation region on the reference image having a rectangular shape and being included in the smallest rectangle surrounding the feature mask, the spatial resolution being estimated in the estimation region.

10. The method according to claim 4, wherein the sequence includes at least one sub-sequence each one ordered from the reference image to a corresponding boundary image of the sequence, the step of determining the optimized transformation further including, for each next moving image being not adjacent to the reference image:

initializing the proposed transformation for the next moving image according to the optimized transformation for at least one previous moving image in the corresponding sub-sequence.

11. The method according to claim 10, wherein the at least one previous moving image consists of a plurality of previous moving images, the step of initializing the proposed transformation for the next moving image including:

estimating the proposed transformation for the next moving image through a predictive algorithm based on the optimized transformations for the previous moving images.

12. The method according to claim 10, further including the steps of:

skipping a number of moving images following a current moving image in the corresponding sub-sequence for defining the next moving image, said number being determined according to a gradient of the optimized transformation for the current moving image, and

interpolating the optimized transformation for each skipped moving image between the optimized transformation for the current moving image and the optimized transformation for the next moving image.

13. The method according to claim 10, wherein the reference image differs from the boundary images of the sequence, the at least one sub-sequence consisting of a sub-sequence ordered from the reference image to a first image of the sequence and a further sub-sequence ordered from the reference image to a last image of the sequence.

14. The method according to claim 1, wherein each transformation is a rigid transformation.

15. The method according to claim 1, wherein each image includes a plurality of visualizing elements each one representing a corresponding basic area of the body-part, for each moving image the method further including the step of:

discarding each visualizing element of the moving image or of the moving image transformed according to the optimized transformation being outside the delimitation mask.

16. The method according to claim 15, further including the steps of:

reducing each image by discarding each visualizing element being discarded in at least one of the moving images, and

determining a result of the analysis according to the reduced images.

17. The method according to claim 16, further including the steps of:

identifying a most intense image and a least intense image in the sequence,

for each basic area calculating a difference between the corresponding visualizing element in the most intense image and the corresponding visualizing element in the least intense image, and

discarding the visualizing elements for the basic area in all the images if the corresponding difference is lower than a threshold value.

18. The method according to claim 1, wherein each image includes a plurality of frames each one being representative of a corresponding slice of the body-part, the delimitation mask and the feature mask being defined on at least two of the frames of the reference image.

19. The method according to claim 1, wherein each image is representative of an ultrasound response of the body-part.

20. The method according to claim 1, wherein the analysis consists of the assessment of a perfusion of a contrast agent in the body-part.

21. A computer program for performing the method of claim 1 when the computer program is executed on a data processing system.

22. A computer program product including a computer-usable medium embodying a computer program, the computer program when executed on a data processing system causing the system to perform the method according to claim 1.

23. An image registration apparatus including means for performing the steps of the method according to claim 1.

24. A medical imaging system including the registration apparatus according to claim 23 and means for acquiring the sequence of images.