System and method for detection of transitions in an image stream of the gastrointestinal tract

• US 9,324,145 B1
• Filed: 08/08/2014
• Issued: 04/26/2016
• Est. Priority Date: 08/08/2013
• Status: Active Grant

• Current Assignee: Given Imaging Limited (Medtronic Plc)
• Original Assignee: Given Imaging Limited (Medtronic Plc)
• Inventors: Peleg, Dori, Cherevatsky, Boris
• Primary Examiner(s): Shah, Utpal
• Application Number:
  US14/454,975
• Time in Patent Office: 627 Days
• Field of Search: None
• US Class Current: 1/1
• CPC Class Codes: A61B 1/000094 extracting biological structures
• CPC Class Codes: A61B 1/000096 using artificial intelligence
• CPC Class Codes: A61B 1/041 Capsule endoscopes for imaging
• CPC Class Codes: A61B 34/20 Surgical navigation systems; Devices for tracking or guiding surgical instruments, e.g. for frameless stereotaxis
• CPC Class Codes: A61B 5/066 Superposing sensor position on an image of the patient, e.g. obtained by ultrasound or x-ray imaging
• CPC Class Codes: G06F 18/2411 based on the proximity to a decision surface, e.g. support vector machines
• CPC Class Codes: G06T 2207/10016 Video; Image sequence
• CPC Class Codes: G06T 2207/30028 Colon; Small intestine
• CPC Class Codes: G06T 7/0012 Biomedical image inspection
• CPC Class Codes: G06T 7/42 using transform domain methods
• CPC Class Codes: G06V 10/764 using classification, e.g. of video objects
• CPC Class Codes: G06V 2201/03 Recognition of patterns in medical or anatomical images

Abstract

A system and method for detecting a transition in a stream of images of a gastrointestinal (GI) tract may include selecting images from an in-vivo image stream; calculating a segment score for each selected image indicating in which segment of the GI tract the image was captured; applying a smoothing function on the scores; detecting a global step in the smoothed segment score signal indicating a substantial change in a parameter calculated based on segment score signal values of the segment score signal values; detecting a local step indicating a substantial change in a parameter calculated based on segment score signal values of a predetermined interval of the of the segment score signal values; combining the local step and the global step; and determining a point of transition in the stream from one anatomical segment to another, the point of transition correlating to the combined step.

Claims

1. A computer-implemented method for detecting a transition in a stream of in-vivo images of a gastrointestinal (GI) tract, the method comprising:

receiving an in-vivo image stream from an in vivo imaging device;

selecting a subset of images from the image stream for analysis;

calculating a segment score for each selected image, the segment score indicating in which anatomic segment of the GI tract the image was captured;

applying a smoothing function on the segment scores of the selected images to obtain a smoothed segment score signal;

detecting a global step in the smoothed segment score signal, said global step indicating a substantial change in a parameter calculated based on the segment score signal values;

detecting a local step in the smoothed segment score signal, said local step indicating a substantial change in a parameter calculated based on segment score signal values of a predetermined interval of the selected images from the image stream;

combining the local step and the global step to obtain a combined step; and

determining a point of transition in the image stream from one anatomical segment to another, the point of transition correlating to the combined step.

2. The method of claim 1 comprising calculating a feature vector for each selected image, the feature vector based on:

a villi score, the villi score corresponding to an amount of villi texture found in a selected image;

a content score, the content score corresponding to the probability that the image depicts intestinal contents; and

a white score, the white score corresponding to pixels in the image which are substantially white.

3. The method of claim 1 comprising:

detecting a bubble sequence in the image stream, the bubble sequence corresponding to a sequence of images, wherein each image of the sequence contains at least a certain predetermined amount of bubbles; and

determining a corrected point of transition according to the detected bubble sequence.

4. The method of claim 3 comprising:

detecting a low segment score sequence, the low segment score sequence including a sequence of images which received a low segment score value relative to images in their vicinity;

detecting an intersection of the bubble sequence and the low segment score sequence; and

determining the middle of the bubble sequence as a corrected point of transition in the image stream.

5. The method of claim 1 wherein the transition in the image stream is from the small bowel to the colon.

6. The method of claim 1 comprising determining a noisiness level of the segment score signal, the noisiness level correlating to the reliability of the segment score signal for determining a point of transition in the image stream.

7. The method of claim 6 comprising, if the noisiness level of the segment score signal is above a predetermined threshold:

calculating a tissue area score for each selected image to obtain a tissue area score signal, the tissue area score per image indicating an amount of tissue which is captured in an image;

determining a step in the tissue area score signal; and

determining the transition in the image stream according to the detected step in the tissue area score signal.

8. The method of claim 1 comprising refining the detected point of transition of the image stream by selecting a new subset of images from the image stream, the new subset selected from the vicinity of the detected point of transition.

9. A system for detecting a transition in a stream of in-vivo images of a gastrointestinal (GI) tract, the system comprising:

a storage unit to store an in-vivo image stream received from an in vivo imaging device;

a processor to;

select a subset of images from the image stream for analysis;

calculate a segment score for each selected image, the segment score indicating in which anatomic segment of the GI tract the image was captured;

apply a smoothing function to the segment scores of the selected images to obtain a smoothed segment score signal;

detect a global step in the smoothed segment score signal, said global step indicating an abrupt change in a parameter calculated based on the segment score signal values;

detect a local step in the smoothed segment score signal, said local step indicating an abrupt change in a parameter calculated based on segment score signal values of a predetermined interval of the selected images from the image stream;

combine the local step and the global step to obtain a combined step; and

determine a transition in the image stream from one anatomical segment to another, the point of transition correlating to the combined step.

10. The system of claim 9, wherein the processor is to calculate a feature vector for each selected image, the feature vector based on:

a villi score, the villi score corresponding to an amount of villi texture found in a selected image;

a content score, the content score corresponding to the probability that the pixel depicts intestinal contents; and

a white score, the white score indicating pixels values which are substantially white.

11. The system of claim 9, wherein the processor is to detect a bubble sequence in the image stream, the bubble sequence corresponding to a sequence of images, wherein each image of the sequence contains at least a certain predetermined amount of bubbles;

and to correct the point of transition according to the detected bubble sequence.

12. The system of claim 11, wherein the processor is to:

detect a low segment score sequence, detect an intersection of the bubble sequence and the low segment score sequence; and

determine the middle of the bubble sequence as a corrected point of transition in the image stream.

13. The system of claim 9, wherein the transition in the image stream is from the small bowel to the colon.

14. The system of claim 9, wherein the processor is to determine a noisiness level of the segment score signal.

15. The system of claim 9, wherein if the noisiness level of the segment score signal is above a predetermined threshold, the processor is to:

calculate a tissue area score for each selected image to obtain a tissue area score signal, the tissue area score per image indicating an amount of tissue which is captured in an image;

determine a step in the tissue area score signal; and

determine the transition in the image stream according to the detected step in the tissue area score signal.

16. The system of claim 9, wherein the processor is to refine the detected point of transition of the image stream by selecting a new subset of images from the image stream, the new subset selected from the vicinity of the detected point of transition.

17. The system of claim 9, wherein the abrupt change in the parameter calculated based on the segment score signal values is a maximal detected change.

18. A method for detecting a transition in a stream of in-vivo images of a gastrointestinal (GI) tract, the method comprising:

selecting images from an in-vivo image stream;

calculating a rating for each selected image, the rating indicating a probability that the image was captured in a predetermined anatomic segment of the GI tract;

applying a smoothing function on the ratings of the selected images to obtain a rating function;

detecting a global increment in the smoothed segment score signal indicating a maximal change in a parameter calculated based on the segment score signal values;

detecting a local increment in the smoothed segment score signal indicating a maximal change in a parameter calculated based on a predetermined window of the of the segment score signal values;

determining a point of transition in the image stream from one anatomical segment based on a combination of the local increment and the global increment.

19. The method of claim 18 comprising calculating a feature vector for each selected image, the feature vector based on:

a villi score, the villi score corresponding to an amount of villi texture found in a selected image;

a content score, the content score corresponding to the probability that the image depicts intestinal contents; and

a white score, the white score corresponding to pixels in the image which are substantially white.

20. The method of claim 19 comprising:

detecting a bubble sequence in the image stream, the bubble sequence corresponding to a sequence of images, wherein each image of the sequence contains at least a certain predetermined amount of bubbles; and

determining a corrected point of transition according to the detected bubble sequence.