Image quality score using a deep generative machine-learning model

• US 10,043,088 B2
• Filed: 05/26/2017
• Issued: 08/07/2018
• Est. Priority Date: 06/23/2016
• Status: Active Grant

• Current Assignee: Siemens Healthineers AG (Siemens AG)
• Original Assignee: Siemens Healthcare GMBH (Siemens AG)
• Inventors: Odry, Benjamin L., Mailhe, Boris, Cetingul, Hasan Ertan, Chen, Xiao, Nadar, Mariappan S.
• Primary Examiner(s): BHATNAGAR, ANAND P
• Application Number:
  US15/606,069
• Publication Number:
  US 20170372155A1
• Time in Patent Office: 438 Days
• CPC Class Codes: G06F 18/2413 based on distances to training or reference patterns
• CPC Class Codes: G06F 18/24143 Distances to neighbourhood prototypes, e.g. restricted Coulomb energy networks [RCEN]
• CPC Class Codes: G06F 18/2415 based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
• CPC Class Codes: G06F 18/28 Determining representative reference patterns, e.g. by averaging or distorting; Generating dictionaries
• CPC Class Codes: G06V 10/82 using neural networks
• CPC Class Codes: G06V 10/98 Detection or correction of errors, e.g. by rescanning the pattern or by human intervention; Evaluation of the quality of the acquired patterns
• CPC Class Codes: G06V 2201/03 Recognition of patterns in medical or anatomical images
• CPC Class Codes: G06V 30/1914 Determining representative reference patterns, e.g. averaging or distorting patterns; Generating dictionaries, e.g. user dictionaries
• CPC Class Codes: G06V 30/19173 Classification techniques
• CPC Class Codes: G16H 30/40 for processing medical images, e.g. editing
• CPC Class Codes: G16H 40/63 for local operation
• CPC Class Codes: G16H 50/20 for computer-aided diagnosis, e.g. based on medical expert systems

Abstract

For image quality scoring of an image from a medical scanner, a generative model of an expected good quality image may be created using deep machine-learning. The deviation of an input image from the generative model is used as an input feature vector for a discriminative model. The discriminative model may also operate on another input feature vector derived from the input image. Based on these input feature vectors, the discriminative model outputs an image quality score.

Claims

1. A method for image quality scoring of an image from a medical scanner, the method comprising:

generating, by the medical scanner, the image representing a patient, the image having a level of artifacts due to the generating by the medical scanner;

determining, by a machine, a probability map of artifacts as a function of location for the image with a deep generative machine-learnt model;

assigning, by the machine, a quality score for the image with application of the probability map to a discriminative machine-learnt classifier; and

transmitting the quality score for the image of the patient.

2. The method of claim 1 wherein generating comprises generating computed tomography, magnetic resonance, ultrasound, positron emission tomography, or single photon emission computed tomography image.

3. The method of claim 1 wherein generating comprises generating a two-dimensional representation of pixels or a three-dimensional set of voxels as the image.

4. The method of claim 1 wherein generating comprises generating with noise artifact, blur artifact, shading artifact, under-sampling artifact, or combinations thereof.

5. The method of claim 1 wherein determining comprises determining with the deep generative machine-learnt model learnt with only training images having a quality above a threshold.

6. The method of claim 1 wherein determining comprises determining the probability map as a function of log-likelihoods of the locations of the image matching the deep generative machine-learnt model.

7. The method of claim 1 wherein determining comprises determining the probability map as a deviation from the deep generative machine-learnt modeled normal image.

8. The method of claim 1 wherein assigning the quality score comprises assigning with the discriminative machine-learnt classifier comprising a deep neural network.

9. The method of claim 1 wherein assigning comprises assigning the quality score as a function of severity and extent of the artifacts.

10. The method of claim 1 further comprising identifying a type of the artifacts, the assigning and identifying being performed with the discriminative machine-learnt classifier being a multi-task classifier, and wherein transmitting comprises transmitting the quality score and the type of the artifacts.

11. The method of claim 1 wherein assigning comprises assigning with the application of the probability map and application of the image to the discriminative machine-learnt classifier, a first set of features used by the discriminative machine-learnt classifier derived from the probability map and a second set of features used by the discriminative machine-learnt classifier derived from the image.

12. The method of claim 1 wherein the discriminative machine-learnt classifier, the deep generative machine-learnt model, or both are responsive to segmentation of the image.

13. The method of claim 1 wherein transmitting comprises transmitting the quality score to a display with the image.

14. The method of claim 1 further comprising scanning the patient again with the medical scanner in response to the quality score.

15. A method for training a machine to determine an image quality score, the method comprising:

training, by the machine, a deep generative model using a piecewise-differentiable function, the deep generative model trained to output a spatial distribution of probability in response to an input image; and

training, by the machine, a discriminative classifier, the discriminative classifier trained to output a score of image quality as a function of input of the spatial distribution of probability.

16. The method of claim 15 wherein training the deep generative model comprises training using images as training data, all the images having a threshold level of the image quality, the output being a probability of matching.

17. The method of claim 15 wherein training the discriminative classifier comprises training with deep learning where the probability input is a deviation of the input image from the deep generative model.

18. The method of claim 15 wherein training the discriminative classifier comprises training with the input of the spatial distribution of probability and deep learnt features extracted from the input image.

19. A method for image quality scoring of an image from a medical scanner, the method comprising:

generating, by the medical scanner, the image representing a patient, the image having a level of artifacts due to the generating by the medical scanner;

determining, by a machine, a probability map of artifacts as a function of location for the image with a deep generative machine-learnt model;

assigning, by the machine, a quality score for the image with application of the probability map to a discriminative machine-learnt classifier, the probability map comprising a first input vector and features of the image comprising a second input vector; and

transmitting the quality score for the image of the patient.

20. The method of claim 19 wherein the discriminative machine-learnt classifier comprises a deep learnt classifier, the second input vector learned training images and the first input vector learned from training probability maps.