Method and data processing unit for optimizing an image reconstruction algorithm

• US 10,614,597 B2
• Filed: 07/26/2017
• Issued: 04/07/2020
• Est. Priority Date: 08/12/2016
• Status: Active Grant

• Current Assignee: Siemens Healthineers AG (Siemens AG)
• Original Assignee: Siemens Healthcare GMBH (Siemens AG)
• Inventors: Hoelzer, Philipp, Ionasec, Razvan
• Primary Examiner(s): Bali, Vikkram
• Application Number:
  US15/659,707
• Publication Number:
  US 20170337713A1
• Time in Patent Office: 986 Days
• Field of Search: None
• CPC Class Codes: G06T 11/003 Reconstruction from projections, e.g. tomography
• CPC Class Codes: G06T 11/006 Inverse problem, transformation from projection-space into object-space, e.g. transform methods, back-projection, algebraic methods
• CPC Class Codes: G06T 2207/10081 Computed x-ray tomography [CT]
• CPC Class Codes: G06T 2207/20081 Training; Learning
• CPC Class Codes: G06T 2207/30101 Blood vessel; Artery; Vein; Vascular
• CPC Class Codes: G06T 7/0012 Biomedical image inspection
• CPC Class Codes: G06T 7/11 Region-based segmentation

Abstract

A method includes provisioning a set of training data sets, each of the training data sets respectively including an acquisition data set; generating a first medical image for each of the training data sets of at least one first subset of the set of training data sets using the image reconstruction algorithm based on a respective acquisition data set; determining an image processing result for each of the respective first medical images using an image processing algorithm based on the respective first medical image; determining image processing information for each of the respective first medical images relating to a quality of the respective image processing result based on the respective image processing result; and optimizing the image reconstruction algorithm based on a first machine learning algorithm, the at least one first subset of the set of training data sets and the image processing information for the respective first medical images.

Claims

1. A method for adjusting an image reconstruction algorithm, the method comprising:

provisioning a set of training data sets, each training data set among the set of training data sets including an acquisition data set;

generating a first medical image for each respective training data set among at least one first subset of the set of training data sets using the image reconstruction algorithm and based on a respective acquisition data set;

determining an image processing result for each respective first medical image using an image processing algorithm based on the respective first medical image;

determining, for each of the respective first medical images, image processing information based on the respective image processing result, the image processing information corresponding to a quality of the respective image processing result; and

adjusting the image reconstruction algorithm based on (i) a first machine learning algorithm, (ii) the at least one first subset of the set of training data sets, and (iii) the image processing information for the respective first medical images.

2. The method of claim 1, wherein the image processing result corresponds to medical information.

3. The method of claim 1, wherein the image processing information corresponds to at least one of a measure of an inaccuracy of the image processing result or a measure of a deviation of the image processing result from a respective reference image processing result.

4. The method of claim 1, wherein the image processing information is determined based on a Dice coefficient.

5. The method of claim 1, wherein the training data sets each include an examination parameter set selected from the group including at least one of an acquisition parameter set, an injection parameter set, a patient parameter set, or an external image parameter set.

6. The method of claim 5, wherein at least one of,

at least one of the image reconstruction algorithm or the image processing algorithm is adjusted based on the examination parameter set, or

at least one of during the adjustment of the image reconstruction algorithm or during the adjustment of the image processing algorithm, the method includes

determining an adjusted region for at least one examination parameter of the examination parameter set based on (i) the first machine learning algorithm, (ii) the set of training data sets, and (iii) the image processing information.

7. A non-transitory computer program product, directly loadable into a memory facility of a data processing system, the non-transitory computer program product including,

program sections for carrying out the method of claim 1 in response to the computer program being executed by the data processing system.

8. A non-transitory computer-readable medium on which program sections, readable-in and executable by a data processing system, are stored to carry out the method of claim 1 in response to the program sections being executed by the data processing system.

9. A method, comprising:

generating a medical image by using the image reconstruction algorithm adjusted in accordance with the method of claim 1.

10. The method of claim 1, wherein at least one of,

the image processing algorithm includes an algorithm to identify a structure in a patient based on the respective first medical image;

or

the image processing algorithm includes an algorithm for the determining a physiological parameter in a patient based on the respective first medical image.

11. The method of claim 1,

wherein each respective training data set of the set of training data sets further comprises a reference image processing result, and

wherein the image processing information associated with each of the respective first medical images is determined based on (i) the respective image processing result and (ii) the respective reference image processing result.

12. The method of claim 1, wherein at least one of,

the image processing algorithm includes an algorithm for segmenting a structure in the first medical image, or

each respective training data set of the set of training data sets further comprises a reference image processing result, and the reference image processing result corresponds to an assignment of image data points of the first medical image to a structure in the first medical image.

13. The method of claim 1, wherein each respective training data set of the set of training data sets further comprises a reference image processing result and the respective reference image processing result includes a map with which image data points of the respective first medical image are assignable to a structure in the respective first medical image.

14. The method of claim 1, wherein the image reconstruction algorithm is adjusted based on a first item of accumulated image processing information, the first item of accumulated image processing information being determined based on the image processing information for the respective first medical images.

15. The method of claim 1, wherein at least one of,

the image reconstruction algorithm is adjusted by determining an at least one image reconstruction parameter set of the image reconstruction algorithm;

or

the image processing algorithm is adjusted by determining at least one image reconstruction parameter set of the image processing algorithm.

16. A method for adjusting an image reconstruction algorithm and an image processing algorithm, the method comprising:

provisioning a set of training data sets, each respective training data set among the training data sets including an acquisition data set;

generating a plurality of first medical images for training data sets among at least one first subset of the set of training data sets using the image reconstruction algorithm and based on acquisition data sets included in the training data sets among the at least one first subset;

determining a first image processing result for each respective first medical image among the plurality of first medical images using an image processing algorithm based on the respective first medical image;

determining first image processing information for each respective first medical image based on the respective first image processing result, the first image processing information corresponding to a quality of the respective first image processing result;

adjusting the image reconstruction algorithm based on (i) first machine learning algorithm, (ii) the at least one first subset of the set of training data sets, and (iii) the first image processing information;

generating a plurality of second medical images for training data sets among at least one second subset of the set of training data sets using the adjusted image reconstruction algorithm and based on acquisition data sets included in the training data sets among the at least one second subset;

determining a second image processing result for each respective second medical image using the image processing algorithm based on the respective second medical image;

determining second image processing information for each respective second medical image based on the respective second image processing result, the second image processing information corresponding to a quality of the respective second image processing result; and

adjusting the image processing algorithm based on (i) a second machine learning algorithm, (ii) the plurality of second medical images, and (iii) the second image processing information.

17. The method of claim 16,

wherein at least two iterations of the method are performed,

wherein, in a first iteration of the at least two iterations, all steps of the method are performed,

wherein, with each further iteration of the at least two iterations, the image reconstruction algorithm is further adjusted proceeding from the image reconstruction algorithm adjusted in a preceding iteration based on the image processing algorithm adjusted in the preceding iteration, and

wherein the image processing algorithm is further adjusted proceeding from the image processing algorithm adjusted in the preceding iteration based on the image reconstruction algorithm that was previously further adjusted in the further iteration.

18. The method of claim 17, wherein at least one of the first image processing result or the second image processing result corresponds to medical information.

19. The method of claim 17, wherein at least one of,

the image processing algorithm includes an algorithm for identifying a structure in a patient based on at least one medical image among the plurality of first medical images or the plurality of second medical images, or

the image processing algorithm includes an algorithm for determining a physiological parameter in the patient based on at least one medical image among the plurality of first medical images or the plurality of second medical images.

20. The method of claim 17,

wherein each respective training data set of the set of training data sets further includes a reference image processing result,

wherein at least one of (i) the first image processing information is determined based on at least the respective first image processing result and a corresponding first reference image result, or (ii) the second image processing information is determined based on the respective second image processing result and a corresponding second reference image processing result.

21. The method of claim 17, wherein at least one of,

the image processing algorithm includes an algorithm for segmenting a structure in at least one of the first medical image, the medical image being one of a first medical image among the plurality of first medical images or a second medical image among the plurality of second medical images, or

each training data set among the set of training data sets further includes a reference image processing result, and the reference image processing result relates to an assignment of image data points of the a medical image to a structure in the medical image, the medical image being one of a first medical image among the plurality of first medical images or a second medical image among the plurality of second medical images.

22. The method of claim 17, wherein each respective training data set of the set of training data sets further comprises a reference image processing result, wherein the reference image processing result includes,

a map with which image data points of a corresponding first medical image are assignable to a structure in the corresponding medical image, the corresponding medical image being a first medical image among the plurality of first medical images or a second medical image among the plurality of second medical images.

23. The method of claim 16, wherein at least one of

the image processing algorithm includes an algorithm for identifying a structure in a patient based on at least one medical image among the plurality of first medical images or the plurality of second medical images;

or

the image processing algorithm includes an algorithm for determining a physiological parameter in the patient based on at least one medical image among the plurality of first medical images or the plurality of second medical images.

24. The method of claim 16,

wherein each respective training data set of the set of training data sets further includes a reference image processing result, and

wherein at least one of (i) the first image processing information is determined based on at least the respective first image processing result and a corresponding first reference image processing result, or (ii) the second image processing information is determined based on the respective second image processing result and a corresponding second reference image processing result.

25. The method of claim 16, wherein at least one of,

the image processing algorithm includes an algorithm for segmenting a structure in a medical image, the medical image being one of a first medical image among the plurality of first medical images or a second medical image among the plurality of second medical images, or

each training data set among the set of training data sets further includes a reference image processing result, and the reference image processing result relates to an assignment of image data points of a medical image to a structure in the medical image, the medical image being one of a first medical image among the plurality of first medical images or a second medical image among the plurality of second medical images.

26. The method of claim 16, wherein a training data set among the set of training data sets further includes a reference image processing result, and the reference image processing result includes,

a map with which image data points of a corresponding medical image are assignable to a structure in the corresponding medical image, the corresponding medical image being a first medical image among the plurality of first medical images or a second medical image among the plurality of second medical images.

27. The method of claim 16, wherein at least one of,

the image reconstruction algorithm is adjusted based on a first item of accumulated image processing information, the first item of accumulated image processing information being determined based on the image processing information for the respective first medical images, or

the image processing algorithm is adjusted based on a second item of accumulated image processing information, the second item of accumulated image processing information being determined based on the image processing information for the respective second medical images.

28. The method of claim 16, wherein at least one of,

the image reconstruction algorithm is adjusted by determining at least one image reconstruction parameter set of the image reconstruction algorithm;

or

the image processing algorithm is adjusted by determining at least one image reconstruction parameter set of the image processing algorithm.

29. A method, comprising:

determining an image processing result by using the image reconstruction algorithm and an image processing algorithm that are adjusted in accordance with the method of claim 16, the determining based on an acquisition data set, wherein

the acquisition data set is provided,

a medical image is generated using the image reconstruction algorithm, the generating the medical image being based on the acquisition data set, and

the image processing result is determined using the image processing algorithm, the image processing result being based on the generated medical image.

30. The method of claim 16, wherein at least one of the first image processing result or the second image processing result corresponds to medical information.

31. The method of claim 16, wherein the first or second image processing information corresponds to at least one of a measure of an inaccuracy of a corresponding one of the first or second image processing results or a measure of a deviation of the corresponding one of the first or second image processing result from a respective reference image processing result.

32. The method of claim 16, wherein at least one of the first or second image processing information is determined based on a Dice coefficient.

33. The method of claim 16, wherein the training data sets each include an examination parameter set selected from the group including at least one of an acquisition parameter set, an injection parameter set, a patient parameter set, or an external image parameter set.

34. The method of claim 33, wherein at least one of,

at least one of the image reconstruction algorithm or the image processing algorithm is adjusted based on the examination parameter set, or

at least one of during the adjustment of the image reconstruction algorithm or during the adjustment of the image processing algorithm, an adjusted region is determined for at least one examination parameter of the examination parameter set based on (i) the first machine learning algorithm, (ii) the set of training data sets, and (iii) the first image processing information.

35. A non-transitory computer program product, directly loadable into a memory facility of a data processing system, including program sections for carrying out the method of claim 16 in response to the computer program being executed by the data processing system.

36. A non-transitory computer-readable medium on which program sections, readable-in and executable by a data processing system, are stored to carry out the method of claim 16 response to the program sections being executed by the data processing system.

37. A method, comprising:

generating a medical image based on an acquisition data set, using the image reconstruction algorithm adjusted in accordance with the method of claim 16.

38. A processing device comprising:

a memory configured to store computer readable instructions; and

at least one processor configured to execute the computer readable instructions to cause the processing device to

provide a set of training data sets, each training data set among the set of training data sets including an acquisition data set,

generate a first medical image for each respective training data set among at least one first subset of the set of training data sets, using an image reconstruction algorithm and based on the respective acquisition data set,

determine a first image processing result for each respective first medical image using an image processing algorithm based on the respective first medical image

determine image processing information for each respective first medical images quality of the respective image processing result based on the respective first image processing result, the first image processing information corresponding to a quality of the respective first image processing result, and

adjust the image reconstruction algorithm based on (i) a first machine learning algorithm, (ii) the at least one first subset of the set of training data sets, and (iii) the first image processing information.

39. The processing device of claim 38, wherein the memory includes computer readable instructions that, when executed by the processor, cause the processing device to,

generate a plurality of second medical images for training data sets among at least one second subset of the set of training data sets using the adjusted image reconstruction algorithm and based on acquisition data sets included in the training data sets among the at least one second subset,

determine a second image processing result for each respective second medical image using an image processing algorithm based on the respective second medical image,

determine second image processing information for each respective second medical images based on the respective second image processing result, the second image processing information corresponding to a quality of the respective second image processing result;

and

adjust the image processing algorithm based on (i) a second machine learning algorithm, (ii) the plurality of second medical images, and (iii) the second image processing information.

40. A medical imaging device comprising:

the processing device of claim 39.

41. A medical imaging device comprising:

the processing device of claim 38.