System and method for automatic detection of in vivo polyps in video sequences

• US 9,342,881 B1
• Filed: 11/26/2014
• Issued: 05/17/2016
• Est. Priority Date: 12/31/2013
• Status: Active Grant

• Current Assignee: Given Imaging Limited (Medtronic Plc)
• Original Assignee: Given Imaging Limited (Medtronic Plc)
• Inventors: Peleg, Dori
• Primary Examiner(s): Bitar, Nancy
• Application Number:
  US14/554,934
• Time in Patent Office: 538 Days
• Field of Search: None
• US Class Current: 1/1
• CPC Class Codes: A61B 1/00016 using wireless means
• CPC Class Codes: A61B 1/041 Capsule endoscopes for imaging
• CPC Class Codes: G06T 2207/10068 Endoscopic image
• CPC Class Codes: G06T 2207/30032 Colon polyp
• CPC Class Codes: G06T 7/0012 Biomedical image inspection
• CPC Class Codes: G06T 7/13 Edge detection
• CPC Class Codes: G06T 7/64 of convexity or concavity
• CPC Class Codes: G06V 2201/032 of protuberances, polyps nodules, etc.

Abstract

A system and method for identifying an in-vivo image frame with one or more polyps. Center points of candidate imaged structures may be determined, each center point defining a respective circle. Polyp edge scores may be computed for a plurality of points around each center point, each polyp edge score measuring the coincidence of a respective candidate imaged structure and its respective circle. For each center point, a plurality of edge points of the respective candidate imaged structure may be detected based on the polyp edge score and a candidate imaged structure may be identified based on the detected edge points. For each candidate imaged structure, a candidate polyp may be identified for which extrapolating the detected edge points forms an edge in a shape approximating an ellipse. Image frames with candidate polyps contributing to an above threshold overall image score are determined and marked to include a polyp.

Claims

1. A method of identifying one or more candidate polyps in an in-vivo image frame, the method comprising:

receiving an image frame of a body lumen captured in vivo;

determining one or more center points of candidate imaged structures, each center point defining a respective circle;

computing polyp edge scores for a plurality of points around each center point, each polyp edge score measuring the coincidence of a respective candidate imaged structure and its respective circle;

for each center point;

detecting a plurality of edge points of the respective candidate imaged structure based on the polyp edge score;

identifying a candidate imaged structure based on the detected edge points, and

for each candidate imaged structure, identifying a candidate polyp for which extrapolating the detected edge points forms an edge in a shape approximating an ellipse.

2. The method of claim 1 comprising:

generating one or more feature scores for one or more respective features for each of one or more candidate polyps in the image frame;

combining the feature scores for each of the candidate polyps in the image to generate an overall image score; and

identifying that the image frame includes at least one verified candidate polyp if the overall image score satisfies a predetermined threshold.

3. The method of claim 2, wherein at least one of the feature scores is computed to be a function of the amount of the detected candidate polyp edge that coincides with the approximated ellipse.

4. The method of claim 2 comprising indicating a probability that the image frame includes a polyp in the image based on the overall image score.

5. The method of claim 4 comprising measuring the amount of the detected candidate polyp edge that coincides with the approximated ellipse by sampling the candidate polyp edge using a non-uniform array of lines.

6. The method of claim 1, wherein the polyp edge score is computed to be a function of the product of the gradient of the image and the gradient of the circle centered at the center point.

7. The method of claim 1 comprising:

for each candidate imaged structure;

computing a polyp edge score for points of the candidate imaged structure located along a ray extending radially outward from the center point; and

detecting at most one edge point along each of a plurality of rays, wherein the detected edge point is the point along each ray that has the greatest difference between the greatest local maximum polyp edge score so far and a local minimum of a following neighboring point.

8. The method of claim 1, wherein the ray has a predetermined length approximating an upper bound of a radius of a typical large polyp.

9. The method of claim 1 comprising removing outlier edge points having radii that substantially differs from the radii of the majority of the other edge points.

10. A system for identifying one or more candidate polyps in an in-vivo image frame, the system comprising:

a memory to store an image frame of a body lumen captured in vivo by an in vivo imaging device; and

a processor to determine one or more center points of candidate imaged structures, each center point defining a respective circle, and compute polyp edge scores for a plurality of points around each center point, each polyp edge score measuring the coincidence of a respective candidate imaged structure and its respective circle, wherein for each center point the processor is to detect a plurality of edge points of the respective candidate imaged structure based on the polyp edge score and identify a candidate imaged structure based on the detected edge points, and wherein for each candidate imaged structure the processor is to identify a candidate polyp for which extrapolating the detected edge points forms an edge in a shape approximating an ellipse.

11. The system of claim 10, wherein the processor is to generate one or more feature scores for one or more respective features for each of one or more candidate polyps in the image frame, combine the feature scores for each of the candidate polyps in the image to generate an overall image score, and identify that the image frame includes at least one verified candidate polyp if the overall image score satisfies a predetermined threshold.

12. The system of claim 11, wherein the processor is to compute at least one of the feature scores to be a function of the amount of the detected candidate polyp edge that coincides with the approximated ellipse.

13. The system of claim 11, wherein the processor is to determine a probability that the image frame includes a polyp in the image based on the overall image score.

14. The system of claim 13, wherein the processor is to measure the amount of the detected candidate polyp edge that coincides with the approximated ellipse by sampling the candidate polyp edge using a non-uniform array of lines.

15. The system of claim 10, wherein the processor is to compute the polyp edge score to be a function of the product of the gradient of the image and the gradient of the circle centered at the center point.

16. The system of claim 10, wherein for each candidate imaged structure the processor is to compute a polyp edge score for points of the candidate imaged structure located along a ray extending radially outward from the center point, and detect at most one edge point along each of a plurality of rays, wherein the detected edge point is the point along each ray that has the greatest difference between the greatest local maximum polyp edge score so far and a local minimum of a following neighboring point.

17. The system of claim 10, wherein the ray has a predetermined length approximating an upper bound of a radius of a typical large polyp.

18. The system of claim 10, wherein the processor is to remove outlier edge points having radii that substantially differs from the radii of the majority of the other edge points.

19. The system of claim 11 further comprising a display to display the image frame identified to include at least one verified candidate polyp.

20. The system of claim 19, wherein the display further displays location, size, shape or color information associated with one or more candidate polyps in the displayed image frame.