Image quality driven non-rigid image registration

Patent metadata

Field	Value
Publication number	US-10290102-B2
Application number	US-201314405007-A
Country	US
Kind code	B2
Filing date	Jun 18, 2013
Priority date	Jun 27, 2012
Publication date	May 14, 2019
Grant date	May 14, 2019

Abstract

An image registration apparatus ( 118 ) includes an image quality driven image registration determiner ( 202 ) that determines an image quality driven image registration for a set of images to register based on a non-rigid registration ( 204 ), which includes an optimization of an image similarity term and a regularization term, and a registration steering factor, and a registration component ( 206 ) that registers the set of images using the image quality driven image registration. A method determining an image quality driven image registration for a set of images to register based on a non-rigid registration, which includes an optimization of an image similarity term and a regularization term, and a registration steering factor, and registering the set of images using the fidelity driven image registration, generating a set of registered images.

First claim

The invention claimed is:

1. A computing system, comprising:
an image quality determiner, implemented by a computer processor, that determines deformation vector fields for a set of images registered with a set of input images by a non-rigid registration algorithm, generates volume curves based on the deformation vector fields, fits motion models to the generated volume curves, and determines an error between the fitted models and the volume curves, wherein the non-rigid registration algorithm includes an optimization of an image similarity term and a regularization term, and the error is an image quality metric;
an image quality driven image registration determiner, implemented by the computer processor, that determines an image quality driven image registration algorithm for a subsequent registration of the set of input images based on the set of images registered with the non-rigid registration algorithm and a registration steering factor, which is based on the image quality metric; and
a registration component, implemented by the computer processor, that performs the subsequent registration of the set of input images by registering the set of input images using the determined image quality driven image registration algorithm.

2. The computing system of claim 1 , further comprising:
a registration steering factor determiner, implemented by a computer processor, that determines the registration steering factor based on a known imaging system image quality, which varies across at least one image of the set of images, which are generated by an imaging system.

3. The computing system of claim 2 , wherein the registration steering factor down weights the image similarity term for a voxel or voxel region of an image depending on the known imaging system image quality of the voxel or voxel region.

4. The computing system of claim 1 , wherein the registration component registers the set of images using the non-rigid registration, and performs the subsequent registration based on a result of the registration of the set of images.

5. The computing system of claim 1 , wherein the error is greater for a voxel region of an image with lower image quality and lower for a voxel region of the image with a higher image quality.

6. The computing system of claim 1 , further comprising:
a decision logic, implemented by the computer processor, that invokes the subsequent registration in response to the image quality metric satisfying a predetermined decision threshold.

7. The computing system of claim 1 , wherein the registration steering factor steers the image quality driven image registration towards the regularization term based on a degree of an image quality of at least one of the registered images, wherein the lower the image quality, the more the registration is steered towards the regularization term and the less the image similarity term contributes to the registration.

8. A method, comprising:
determining, with a computing system, an image quality metric by determining deformation vector fields for a set of images registered with a set of input images by a non-rigid registration algorithm, generating volume curves based on the deformation vector fields, fitting motion models to the generated volume curves, and determining an error between the fitted models and the volume curve, wherein the non-rigid registration algorithm includes an optimization of an image similarity and a regularization term, and the error is the image quality metric;
determining, with the computing system, an image quality driven image registration algorithm for subsequent registration of the set of input images based on the non-rigid registration algorithm and a registration steering factor that is based on the image quality metric; and
performing, with the computing system, the subsequent registration of the set of input images by registering the set of input images using the determined image quality driven image registration algorithm, generating a set of registered images.

9. The method of claim 8 , wherein an image quality indicated by the image quality metric varies across at least one image of the set of images.

10. The method of claim 8 , further comprising:
registering the set of images using the non-rigid registration to determine the deformation vector fields.

11. The method of claim 10 , further comprising:
determining the registration steering factor based on the image quality metric.

12. The method of claim 8 , wherein the registration steering factor steers the image quality driven image registration towards the regularization term based on a degree of an image quality of the registered images, wherein the lower the image quality, the more the registration is steered towards the regularization term and the less the image similarity term contributes to the registration.

13. The method of claim 8 , further comprising:
receiving, via an interactive user interface, an input that determines the registration steering factor.

14. The method of claim 8 , further comprising:
visually displaying, via an interactive user interface, the set of registered images.

15. The method of claim 14 , wherein the registration steering factor locally decreases an image influence with respect to an increase of a regularization influence and increases, with respect to the decrease to an image influence, the regularization influence in an energy term optimized during the determining an image quality driven image registration algorithm in regions of lower image quality.

Assignees

• Koninklijke Philips Nv

Inventors

• Lorenz Cristian
• Kabus Sven
• Klinder Tobias
• Von Berg Jens

Classifications

• G06T7/0014 Primary

  using an image reference approach · CPC title

• G06T7/20

  Analysis of motion (motion estimation for coding, decoding, compressing or decompressing digital video signals H04N19/43, H04N19/51) · CPC title

• G06T7/344

  involving models · CPC title

• G06T17/10

  Constructive solid geometry [CSG] using solid primitives, e.g. cylinders, cubes · CPC title

• G06T3/0068

  Physics · mapped topic