Method and system for medical image registration

1 . A method, implemented in a processing unit, for registering a first image and a second image, wherein said first image comprise data significant for x-ray attenuation from a first medical imaging study and said second image comprise data significant for x-ray attenuation from at least a second medical imaging study, the method comprises

obtaining a first tissue class first subimage associated with said first image and comprising information of a first tissue class in said first image,

obtaining a first tissue class second subimage associated with said second image and comprising information of said first tissue class in said second image, wherein the first tissue class is bone tissue;

a first registration of said first image to said second image; and

a second registration of said first image to said second image,

wherein said first registration comprises

optimization of a first weighted cost function, said first weighted cost function comprising the sum of

a correlation of the first image to the second image with a first weight factor of said first registration; and

a correlation of said first tissue class first subimage to said first tissue class second subimage with a second weight factor of said first registration, said first and second weight factors of said first registration are nonzero;

wherein a first set of deformation parameters from said first registration associated with a transformation of said first image to said second image are obtained from said optimization, and

wherein said second registration comprises

correlation of at least part of said first image to at least part of said second image based on said first set of deformation parameters, wherein a second set of deformation parameters associated with a transformation of said first image to said second image are obtained from said correlation.

2 . The method according to claim 1 , wherein said first registration provides a first set of constraint parameters defining a set of image voxels in said first image associated with said first tissue class and said second registration takes said first set of constraint parameters into account by said first set of constraint parameters defining a set of voxels in the first image having a predefined limitation of registering when performing said second registration.

3 . The method according to claim 1 , wherein the method comprises

obtaining a second tissue class first subimage associated with said first image and comprising information of a second tissue class in said first image,

obtaining a second tissue class second subimage associated with said second image and comprising information of said second tissue class in said second image, wherein the second tissue class is selected from a group comprising lean soft tissue and fat tissue;

wherein the correlation of at least part of said first image to at least part of said second image in said second registration comprises

optimization of a second weighted cost function, said second weighted cost function comprising the sum of

a correlation of the first image to the second image with a first weight factor of the second registration and

a correlation of the second tissue class first subimage to said second tissue class second subimage with a second weight factor of the second registration, said first and second weight factors of the second registration are nonzero,

said second registration provides a second set of constraint parameters defining a set of image voxels in said first image associated with said second tissue class,

wherein said second set of deformation parameters are obtained from said optimization.

4 . The method according to claim 3 , wherein the method further comprises

obtaining a third tissue class first subimage of said second registration associated with said first image and comprising information of a third tissue class in said first image;

obtaining a third tissue class second subimage of said second registration associated with said first image and comprising information of a third tissue class in said second image, wherein the second tissue class is lean soft tissue and the third tissue class is fat tissue;

wherein the second weighted cost function comprises the sum of

a correlation of the first image to the second image with a first weight factor of said second registration

a correlation of said second tissue class first subimage to said second tissue class second subimage with a second weight factor of the second registration and

a correlation of said third tissue class first subimage to said third tissue class second subimage with a third weight factor of the second registration, said first, second and third weight factors of said second registration are nonzero,

said second registration further provides constraint parameters to the second set of constraint parameters defining a set of image voxels in said first image associated with said third tissue class.

5 . The method according to claim 1 , wherein the method comprises

obtaining a fourth tissue class first subimage associated with said first image and comprising information of composed tissues in said first image;

obtaining a fourth tissue second subimage associated with said second image and comprising information of composed tissues in said second image;

wherein said composed tissues are defined as tissues enclosed by subcutaneous fat in said first and second image respectively,

obtaining a fifth tissue class first subimage associated with said first image and comprising information of fat tissue in said first image;

obtaining a fifth tissue class second subimage associated with said second image and comprising information of fat tissue in said second image;

an intermediate registration of said first image to said second image based on said first set of deformation parameters, wherein said intermediate registration comprises

optimization of a third weighted cost function comprising the sum of

a correlation of the first image to the second image with a first weight factor of said intermediate registration,

a correlation of the fourth tissue class first subimage to the fourth tissue class second subimage with a second weight factor of said intermediate registration, and

a correlation of the fifth tissue class first subimage to the fifth tissue class second subimage with a third weight factor of said intermediate registration, said first, second and third weight factors of the intermediate registration are nonzero,

wherein an intermediate set of deformation parameters associated with a transformation of said first image to said second image are obtained from said optimization and said second registration is based on said intermediate set of deformation parameters.

6 . The method according to claim 5 , wherein

said first registration provides a first set of constraint parameters defining a set of image voxels in said first image associated with said first tissue class and said second registration takes said first set of constraint parameters into account by said first set of constraint parameters defining a set of voxels in the first image having a predefined limitation of registering when performing said second registration;

said intermediate registration takes the first set of constraint parameters into account by said first set of constraint parameters defining at least one set of voxels in the first image having a predefined limitation of registering when performing said intermediate registration

said intermediate registration provides a third set of constraint parameters defining at least one set of image voxels in said first image associated with said composed tissues and

said second registration further takes said third set of constraint parameters into account by said third set of constraint parameters defining at least one set of voxels in the first image having a predefined limitation of registering when performing said second registration.

7 . The method according to claim 1 , wherein the method further comprises

obtaining a sixth tissue class first subimage associated with said first image and comprising information of a sixth tissue class in said first image

obtaining a sixth tissue class second subimage associated with said second image and comprising information of said sixth tissue class in said second image, wherein said sixth tissue class is subcutaneous fat;

a third registration of said first image to said second image, wherein said third registration comprises

optimization of a fourth weighted cost function of said third registration comprising the sum of

a correlation of the first image to the second image with a first weight factor of said third registration; and

a correlation of said sixth tissue class first subimage to said sixth tissue class second subimage with a second weight factor of said third registration, said first and second weight factors of said third registration are nonzero;

wherein a third set of deformation parameters associated with a transformation of said first image to said second image are obtained from said optimization.

8 . The method according to claim 7 , wherein

said first registration provides a first set of constraint parameters defining a set of image voxels in said first image associated with said first tissue class and said second registration takes said first set of constraint parameters into account by said first set of constraint parameters defining a set of voxels in the first image having a predefined limitation of registering when performing said second registration;

wherein the method comprises

obtaining a second tissue class first subimage associated with said first image and comprising information of a second tissue class in said first image,

obtaining a second tissue class second subimage associated with said second image and comprising information of said second tissue class in said second image, wherein the second tissue class is selected from a group comprising lean soft tissue and fat tissue;

wherein the correlation of at least part of said first image to at least part of said second image in said second registration comprises

optimization of a second weighted cost function, said second weighted cost function comprising the sum of

a correlation of the first image to the second image with a first weight factor of the second registration and

a correlation of the second tissue class first subimage to said second tissue class second subimage with a second weight factor of the second registration, said first and second weight factors of the second registration are nonzero,

said second registration provides a second set of constraint parameters defining a set of image voxels in said first image associated with said second tissue class,

wherein said second set of deformation parameters are obtained from said optimization;

said third registration takes at least said first set of constraint parameters into account, by said first set of constraint parameters defining a set of voxels in the first image having a predefined limitation of registering when performing said third registration,

optionally, said third registration takes at least said first and second sets of constraint parameters into account when said first and second sets of constraint parameters are respectively present, by said first and second set of constraint parameters defining a respective set of voxels in the first image having a predefined limitation of registering when performing said third registration,

optionally, said third registration takes at least said first, second and third set of constraint parameters into account when said first, second and third sets of constraint parameters are respectively present, by said first, second and third set of constraint parameters defining a respective set of voxels in the first image having a predefined limitation of registering when performing said third registration.

9 . The method according to claim 1 , wherein each step of obtaining a respective tissue class subimage associated with said first image comprises

identifying based on the data in said first image a first set of first image voxels associated with the respective tissue class in said first image and wherein the first set of first image voxels represents the respective tissue class based on the data in said first image, and

creating a subimage associated with said first image from said identified first set of first image voxels in said first image,

which each step of obtaining a respective tissue class subimage associated with said second image comprises

identifying based on the data in said second image a second set of second image voxels associated with the respective tissue class in said second image and the second set of second image voxels represents the respective tissue class based on the data in said second image, and

creating a subimage associated with said second image from said identified second set of second image voxels in said second image.

10 . The method according to claim 1 , wherein each respective correlation of the first image to the second image is selected from a group comprising normalized cross correlation, sum of squared differences and mutual information and wherein each correlation of each first subimage to each respective second subimage is selected from a group comprising normalized cross correlation, sum of squared differences and mutual information.

11 . The method according to claim 1 , wherein the method comprises:

affine registration of said first tissue class first subimage to said first tissue class second subimage, wherein an initial set of deformation parameters associated with a transformation of said first image to said second image are obtained from said affine registration,

wherein said first registration is based on said initial set of deformation parameters and

optionally, said method further comprises image cropping of said first tissue class first subimage and said first tissue class second subimage according to a predetermined cropping region.

12 . The method according to claim 1 , wherein the method comprises initial image adjustment of a first pre-image of said first image and of a second pre-image of said second image, wherein said initial image adjustment outputs said first and second image, respectively, and comprises at least one of

intensity scaling comprising linear or non-linear intensity scaling of said first and second pre-image, respectively and

image filtering comprising applying at least one of a mean filter, a median filter and a Sobel filter to said first and second pre-image, respectively.

13 . The method according to claim 1 , comprising deforming said first image based on any of said sets of deformation parameters.

14 . The method according to claim 1 , the method comprising

obtaining a third image captured in a Positron Emission Tomography, PET, study, wherein said third image has a known spatial and temporal relationship to said first image;

deforming said third image based on any of said sets of deformation parameters.

15 . A system for processing image information comprising a processing unit, wherein said processing unit is configured to perform the method according to claim 1 .