Self-aware image segmentation methods and systems

Patent metadata

Field	Value
Publication number	US-10565711-B2
Application number	US-201615574296-A
Country	US
Kind code	B2
Filing date	Apr 29, 2016
Priority date	May 18, 2015
Publication date	Feb 18, 2020
Grant date	Feb 18, 2020

Abstract

The following relates generally to image segmentation. In one aspect, an image is received and preprocessed. The image may then be classified as segmentable if it is ready for segmentation; if not, it may be classified as not segmentable. Multiple, parallel segmentation processes may be performed on the image. The result of each segmentation process may be marked as a potential success (PS) or a potential failure (PF). The results of the individual segmentation processes may be evaluated in stages. An overall failure may be declared if a percentage of the segmentation processes marked as PF reaches a predetermined threshold.

First claim

The invention claimed is:

1. An apparatus for segmenting a medical image, comprising:
a memory that stores instructions; and
a processor that executes the instructions, wherein, when executed by the processor, the instructions cause the processor to:
perform multiple, parallel segmentation processes employing different segmentation process initializations on an input image to generate a plurality of segmentation results, wherein the different segmentation process initializations are generated by random perturbations of a baseline segmentation process initialization;
mark each segmentation result of the multiple, parallel segmentation processes as a potential success (PS) or potential failure (PF); and
combine the segmentation results marked as PS to produce an output segmentation result for the input image.

2. The apparatus according to claim 1 , wherein:
the plurality of segmentation results include both intermediate segmentation results and a final segmentation result for each segmentation process of the multiple, parallel segmentation processes; and
only the final segmentation results marked as PS are combined to produce the output segmentation result for the input image.

3. The apparatus according to claim 2 , wherein the instructions further cause the processor to:
declare an overall failure if a percentage of the multiple, parallel segmentation processes having an intermediate segmentation result marked as PF reaches a predetermined threshold.

4. The apparatus according to claim 1 , wherein the multiple, parallel segmentation processes are iterative segmentation processes, the plurality of segmentation results include both intermediate segmentation results produced by non-terminal iterations of the segmentation processes and a final segmentation result produced by each segmentation process, and the instructions further cause the processor to:
at each iteration of the iterative segmentation processes, adjust a measurement criteria used in marking each segmentation result of the multiple, parallel segmentation processes as a PS or PF.

5. The apparatus according to claim 1 , wherein the instructions cause the processor to mark each segmentation result of the multiple, parallel segmentation processes as a potential success (PS) or potential failure (PF), the instructions further cause the processor to:
identify a largest group of mutually similar segmentation results, wherein: segmentation results belonging to the largest group of mutually similar segmentation results are marked as PS; and
segmentation results not belonging to the largest group of mutually similar segmentation results are marked as PF.

6. The apparatus according to claim 1 , wherein the instructions further cause the processor to: generate an uncertainty or confidence interval for the output segmentation result based on a statistical variation of the segmentation results marked as PS.

7. The apparatus according to claim 1 , wherein each segmentation result is marked with a probability value P PS of being a PS and with a probability value P PF of being a PF, where for each segmentation result P PS is in a range [0,1], P PF is in a range [0,1], and P PS +P PF =1.

8. The apparatus according to claim 7 , wherein prior to the performing multiple, parallel segmentation processes on the input image, the instructions further cause the processor to:
preprocess the input image; and
classify, with a binary classifier, the input image as a segmentable or not segmentable, wherein the multiple, parallel segmentation processes are performed only if the input image is classified as segmentable.

9. The apparatus of claim 8 , wherein the preprocessing comprises performing at least one of the following on the input image:
smoothing;
contrast enhancement;
edge detection; or
non-rigid deformation.

10. The apparatus according to claim 8 , wherein the instructions further cause the processor to perform a training phase in which the binary classifier is trained by receiving multiple training images wherein each training image of the multiple training images is labeled as segmentable or not segmentable.

11. A medical system for segmenting a medical image, the medical system comprising:
an input for receiving an input image;
image quality (IQ) binary classifier configured to determine in the input image is segmentable;
a memory that stores instructions; and
a processor that executes the instructions, wherein, when executed by the processor, the instructions cause the processor to:
perform multiple, parallel segmentation processes employing different segmentation process initializations on an input image to generate a plurality of segmentation results, wherein the different segmentation process initializations are generated by random perturbations of a baseline segmentation process initialization;
mark each segmentation result of the multiple, parallel segmentation processes as a potential success (PS) or potential failure (PF); and
combine the segmentation results marked as PS to produce an output segmentation result for the input image.

12. The medical system according to claim 11 , wherein:
the plurality of segmentation results include both intermediate segmentation results and a final segmentation result for each segmentation process of the multiple, parallel segmentation processes; and
only the final segmentation results marked as PS are combined to produce the output segmentation result for the input image.

13. The medical system according to claim 12 , wherein the instructions further cause the processor to:
declare an overall failure if a percentage of the multiple, parallel segmentation processes having an intermediate segmentation result marked as PF reaches a predetermined threshold.

14. The medical system according to claim 11 , wherein the multiple, parallel segmentation processes are iterative segmentation processes, the plurality of segmentation results include both intermediate segmentation results produced by non-terminal iterations of the segmentation processes and a final segmentation result produced by each segmentation process, and the instructions further cause the processor to:
at each iteration of the iterative segmentation processes, adjust a measurement criteria used in marking each segmentation result of the multiple, parallel segmentation processes as a PS or PF.

15. The medical system according to claim 11 , wherein the instructions cause the processor to mark each segmentation result of the multiple, parallel segmentation processes as a potential success (PS) or potential failure (PF), the instructions further cause the processor to:
identify a largest group of mutually similar segmentation results, wherein: segmentation results belonging to the largest group of mutually similar segmentation results are marked as PS; and
segmentation results not belonging to the largest group of mutually similar segmentation results are marked as PF.

16. The medical system according to claim 11 , wherein the instructions further cause the processor to:
generate an uncertainty or confidence interval for the output segmentation result based on a statistical variation of the segmentation results marked as PS.

17. The medical system according to claim 11 , wherein each segmentation result is marked with a probability value P PS of being a PS and with a probability value P PF of being a PF, where for each segmentation result P PS is in a range [0,1], P PF is in a range [0,1], and P PS +P PF =1.

18. The medical system according to claim 17 , wherein prior

Assignees

• Koninklijke Philips Nv

Inventors

• Yan Pingkun
• Hall Christopher Stephen
• Lu Kongkuo

Classifications

• G06T7/10 Primary

  Segmentation; Edge detection (motion-based segmentation G06T7/215) · CPC title

• G06T2207/30004

  Biomedical image processing · CPC title

• G06T2207/20081

  Training; Learning · CPC title

• G06T7/143

  involving probabilistic approaches, e.g. Markov random field [MRF] modelling · CPC title

• G06T7/0012

  Biomedical image inspection · CPC title