Automatic determination of regularization factor for iterative image reconstruction with regularization and/or image de-noising

Patent metadata

Field	Value
Publication number	US-9076191-B2
Application number	US-201214364715-A
Country	US
Kind code	B2
Filing date	Dec 4, 2012
Priority date	Dec 13, 2011
Publication date	Jul 7, 2015
Grant date	Jul 7, 2015

Abstract

A processing component ( 122 ) processes images based on an iterative reconstruction algorithm with regularization and/or de-noising algorithm. The processing component includes a set point determiner ( 224 ) that determines a quality set point ( 216 ) between predetermined lower and upper quality bounds ( 226 ) based on a quality variable ( 228 ) indicative of an image quality of interest. The processing component further includes a comparator ( 214 ) that compares, each processing iteration, a quality metric of a current generated image with the quality set point and generates a difference value indicative of a difference between the quality metric and the quality set point. The processing component further includes a regularization factor updater ( 220 ) that generates an updated regularization factor for a next processing iteration based on a current value ( 222 ) of the regularization factor and at least the quality metric in response to the difference value indicating that the quality metric is outside of a predetermined range about the quality set point.

First claim

The invention claimed is:

1. A processing component that processes images based on an iterative reconstruction algorithm with regularization and/or de-noising algorithm, the processing component comprising:
a set point determiner that determines a quality set point between predetermined lower and upper quality bounds based on a quality variable indicative of an image quality of interest;
a comparator that compares, each processing iteration, a quality metric of a current generated image with the quality set point and generates a difference value indicative of a difference between the quality metric and the quality set point; and
a regularization factor updater that generates an updated regularization factor for a next processing iteration based on a current value of the regularization factor and at least the quality metric in response to the difference value indicating that the quality metric is outside of a predetermined range about the quality set point.

2. The processing component of claim 1 , wherein the quality variable defines a relative target level between the lower and upper bounds.

3. The processing component of claim 2 , wherein the relative target level corresponds to a target image noise or artifact level reduction of interest.

4. The processing component of claim 1 , wherein the upper bounds corresponds to a quality metric of the image before regularizing the image and the lower bounds corresponds to a quality metric of the image after fully regularizing the image.

5. The processing component of claim 1 , wherein the lower bound is approximated without fully regularizing the image.

6. The processing component of claim 1 , wherein the lower and the upper bounds are normalized by the upper bound.

7. The processing component of claim 1 , wherein the regularization factor updater generates the updated regularization factor by one or more of multiplying the current value of the regularization factor by an update term, which is a function of the quality metric or by adding to the current regularization factor an update term, which is a function of the quality metric.

8. The processing component of claim 1 , wherein the regularization factor updater generates the updated regularization factor based on proportional-integral-derivative control.

9. The processing component of claim 1 , wherein the quality metric of an iteration is indicative of an image noise level of the generated image of that iteration.

10. The processing component of claim 1 , wherein the quality metric of the iteration is indicative of a standard deviation of an image noise level of the generated image of that iteration.

11. The processing component of claim 1 , wherein the quality metric of an iteration is indicative of an artifact level of the generated image of that iteration.

12. The processing component of claim 1 , further comprising:
a de-noiser that de-noises the image each iteration based on the updated regularization factor for the corresponding iteration.

13. The processing component of claim 1 , further comprising:
an image updater that updates the image each iteration based on the updated regularization factor for the corresponding iteration.

14. The processing component of claim 1 , wherein the regularization factor updater generates the updated regularization factor based on a predetermined behavior of the quality metric.

15. A method, comprising:
processing an initial image during a first processing iteration using an algorithm, which employs regularization, and an initial regularization factor, and generating a first image;
determining a first quality metric of the first image;
comparing the first quality metric of the first image with a quality set point and determining a difference value there between; and
generating an updated regularization factor for a subsequent processing iteration of the image in response to the first quality metric being outside of a predetermined range about the quality set point,

wherein the updated regularization factor is generated based on the initial regularization factor and the first quality metric, and the updated regularization factor is used to process the first image during the subsequent processing iteration and generate a subsequent image.

16. The method of claim 15 , wherein the quality set point is based on a user defined target percentage between a lower quality bounds and an upper quality bounds.

17. The method of claim 16 , wherein the lower bounds corresponds to a quality value of the image before regularization and the upper bounds corresponds to a quality value of the image after full regularization.

18. The method of claim 16 , wherein the quality metric each iteration is indicative of an image noise level or an artifact level of the generated image each respective processing iteration.

19. The method of claim 15 , further comprising:
generating an updated regularization factor each processing iteration until the first quality metric is within the predetermined range about the quality set point.

20. A processing component, comprising:
a component that processes an image using an iterative algorithm employing regularization and generates an image, wherein the iterative algorithm at least includes a regularization term multiplied by a regularization factor;
an analyzer that generates a quality metric after each processing iteration for a respective generated image; and
a regularization factor determiner that automatically generates an updated regularization factor for a next processing iteration by the component based on a quality metric of a generated image of a current processing iteration and a predetermined quality set point for the generated image.

Assignees

• Koninkl Philips Nv

Inventors

• Bergner Frank
• Brendel Bernhard Johannes
• Koehler Thomas

Classifications

• G06T12/20 Primary

  Inverse problem, i.e. transformations from projection space into object space · CPC title

• G06T2211/424

  Iterative · CPC title

• G06T2207/20004

  Adaptive image processing · CPC title

• G06T11/006

  Physics · mapped topic

• G06T5/00 Primary

  Image enhancement or restoration · CPC title