Robust segmentation of retinal pigment epithelium layer

• US 20160133013A1
• Filed: 11/04/2015
• Published: 05/12/2016
• Est. Priority Date: 11/06/2014
• Status: Active Grant

• Current Assignee: Canon Kabushiki Kaisha (Canon Inc.)
• Original Assignee: Canon Kabushiki Kaisha (Canon Inc.)
• Inventors: DOCHERTY, ANDREW, PAN, RUIMIN
• Granted Patent: US 10,460,166 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: A61B 3/0025 characterised by electronic signal processing, e.g. eye models
• CPC Class Codes: A61B 3/102 for optical coherence tomography [OCT]
• CPC Class Codes: G06T 2207/10101 Optical tomography; Optical coherence tomography [OCT]
• CPC Class Codes: G06T 2207/30041 Eye; Retina; Ophthalmic
• CPC Class Codes: G06T 7/12 Edge-based segmentation
• CPC Class Codes: G06T 7/149 involving deformable models, e.g. active contour models
• CPC Class Codes: G06V 10/755 Deformable models or variational models, e.g. snakes or active contours
• CPC Class Codes: G06V 40/193 Preprocessing; Feature extraction
• CPC Class Codes: G06V 40/197 Matching; Classification

Abstract

A method determining a retinal pigment epithelium identifies a plurality of regions using captured image data of the eye and fits a curve into at least some of the regions. A curve score is determined associated with the fitted curve using at least a distance between the fitted curve and at least some of the regions in which a contribution of the regions to the curve score is biased towards (asymmetric) regions below the fitted curve. These steps are repeated whereupon one of the fitted curves is selected, using the corresponding associated curve score, for classifying some of the regions as forming at least a part of a retinal pigment epithelium.

Claims

1. A computer-implemented method for determining a retinal pigment epithelium (RPE), the method comprising:

(a) identifying a plurality of regions using captured image data of the eye;

(b) fitting a curve into at least some of the regions;

(c) determining a curve score associated with the fitted curve using at least a distance between the fitted curve and at least some of the regions, wherein a contribution of the regions to the curve score is biased towards regions below the fitted curve;

(d) repeating steps (b) and (c) at least once; and

(e) selecting one of the fitted curves, using the corresponding associated curve score, for classifying some of the regions as forming at least a part of a retinal pigment epithelium.

4. A method according to claim 1, wherein the captured image data comprises at least one B-scan slice of retina, the B-scan slice comprises intensity image data and polarisation data of an area of an eye.

5. A method according to claim 4, further comprising using the intensity image data and polarisation data to produce degree of polarisation uniformity (DOPU) data.

6. A method according to claim 5, wherein the identifying a plurality of regions comprises determining a plurality of connected components of the DOPU data, so that each depolarising region in the DOPU data is described as at least one connected component.

7. A method according to claim 1, wherein the depolarising regions below the curve are weighted higher than the depolarising regions above the curve in determining the curve score.

8. The method according to claim 1, further comprising:

selecting depolarising regions within a predetermined distance to the selected curve;

fitting a smooth RPE curve to the selected depolarising regions;

classifying the selected depolarising regions as forming at least one of RPE tissues, choroid tissues, and other depolarising particles.

9. A method according to claim 8, wherein a depolarising region is classified as being part of the RPE if the depolarising region substantially falls within the predetermined distance to the fitted RPE curve.

10. A method according to claim 9, wherein a depolarising region is classified as choroid if the depolarising region is below the fitted RPE curve at a distance exceeding the predetermined distance.

11. A method according to claim 8, wherein a depolarising region is classified as belonging to other depolarising particles if the depolarising region is above the fitted RPE curve at a distance exceeding the predetermined distance.

12. A method according to claim 8, wherein depolarising regions are classified using corresponding shape descriptors.

2. A computer implemented method for determining a retinal pigment epithelium, the method comprising:

identifying a plurality of regions using captured image data of the eye;

fitting a plurality of curves to at least some of the identified regions, each fitted curve defining a first plurality of depolarising regions a predetermined distance above the fitted curve and a second plurality of depolarising regions the predetermined distance below the fitted curve; and

selecting one of fitted curves by biasing in favour of the second plurality of depolarising regions, the selected fitted curve being used to classify some of the depolarising regions as forming at least a part of a retinal pigment epithelium (RPE).

13. A method according to claim 2, wherein the captured image data comprises at least one B-scan slice of retina, the B-scan slice comprises intensity image data and polarisation data of an area of an eye, the method further comprising using the intensity image data and polarisation data to produce degree of polarisation uniformity (DOPU) data, wherein the identifying a plurality of regions comprises determining a plurality of connected components of the DOPU data, so that each depolarising region in the DOPU data is described as at least one connected component.

3. A computer-implemented method for determining a retinal pigment epithelium, the method comprising:

identifying a plurality of depolarising regions using captured image data;

obtaining a disease related arrangement of depolarising regions relative to the retinal pigment epithelium;

identifying a curve running through at least some of the identified depolarising regions based on the obtained disease related arrangement; and

classifying some of the depolarising regions as forming at least a part of a retinal pigment epithelium using the identified curve.

14. A method according to claim 3, wherein the captured image data comprises at least one B-scan slice of retina, the B-scan slice comprises intensity image data and polarisation data of an area of an eye, the method further comprising using the intensity image data and polarisation data to produce degree of polarisation uniformity (DOPU) data, wherein the identifying a plurality of regions comprises determining a plurality of connected components of the DOPU data, so that each depolarising region in the DOPU data is described as at least one connected component.

15. A method according to claim 3, wherein the disease related arrangement is modelled using representative image data of an area of an eye having a particular disease based on distribution of depolarising tissues in the representative image data and strength of optical attenuation and scattering in the depolarising tissues.

16. A non-transitory computer readable storage medium having a program recorded thereon, the program being executable by computerised apparatus to determine a retinal pigment epithelium, the program comprising:

(a) code for identifying a plurality of regions using captured image data of the eye;

(b) code for fitting a curve into at least some of the regions;

(c) code for determining a curve score associated with the fitted curve using at least a distance between the fitted curve and at least some of the regions, wherein a contribution of the regions to the curve score is biased towards regions below the fitted curve;

(d) code for repeating execution of the code (b) and (c) at least once; and

(e) code for selecting one of the fitted curves, using the corresponding associated curve score, for classifying some of the regions as forming at least a part of a retinal pigment epithelium.

17. A non-transitory computer readable storage medium having a program recorded thereon, the program being executable by computerised apparatus to determine a retinal pigment epithelium, the program comprising:

code for identifying a plurality of regions using captured image data of the eye;

code for fitting a plurality of curves to at least some of the identified regions, each fitted curve defining a first plurality of depolarising regions a predetermined distance above the fitted curve and a second plurality of depolarising regions the predetermined distance below the fitted curve;

code for selecting one of fitted curves by biasing in favour of the second plurality of depolarising regions, the selected fitted curve being used to classify some of the depolarising regions as forming at least a part of a retinal pigment epithelium.

18. A non-transitory computer readable storage medium having a program recorded thereon, the program being executable by computerised apparatus to determine a retinal pigment epithelium, the program comprising:

code for identifying a plurality of depolarising regions using captured image data;

code for obtaining a disease related arrangement of depolarising regions relative to the retinal pigment epithelium;

code for identifying a curve running through at least some of the identified depolarising regions based on the obtained disease related arrangement; and

code for classifying some of the depolarising regions as forming at least a part of a retinal pigment epithelium using the identified curve.

20. A system according to claim 18, the captured image data comprises at least one B-scan slice of retina, the B-scan slice comprises intensity image data and polarisation data of an area of an eye, and the program further comprises code for using the intensity image data and polarisation data to produce degree of polarisation uniformity (DOPU) data, wherein the identifying a plurality of regions comprises determining a plurality of connected components of the DOPU data, so that each depolarising region in the DOPU data is described as at least one connected component.

21. A system according to claim 18, wherein the depolarising regions below the curve are weighted higher than the depolarising regions above the curve in determining the curve score.

19. A system for determining retinal pigment epithelium (RPE), the system comprising:

a source of optical coherence tomography (OCT) images representing captured image data of an eye;

a computer system having a processor coupled to a memory, the memory storing a program for execution by the processor to determine the RPE, the program comprising;

code for identifying a plurality of regions using the captured image data of the eye;

code for one of;

(A) (A)(i) fitting a curve into at least some of the regions;

(A)(ii) determining a curve score associated with the fitted curve using at least a distance between the fitted curve and at least some of the regions, wherein a contribution of the regions to the curve score is biased towards regions below the fitted curve;

(A)(iii) repeating steps (A)(i) and (A)(ii) at least once; and

(A)(iv) selecting one of the fitted curves, using the corresponding associated curve score, for classifying some of the regions as forming at least a part of a retinal pigment epithelium (RPE);

(B) (B)(i) fitting a plurality of curves to at least some of the identified regions, each fitted curve defining a first plurality of depolarising regions a predetermined distance above the fitted curve and a second plurality of depolarising regions the predetermined distance below the fitted curve; and

(B)(ii) selecting one of fitted curves by biasing in favour of the second plurality of depolarising regions, the selected fitted curve being used to classify some of the depolarising regions as forming at least a part of a retinal pigment epithelium (RPE); and

(C) (C)(i) obtaining a disease related arrangement of depolarising regions relative to the retinal pigment epithelium;

(C)(ii) identifying a curve running through at least some of the identified depolarising regions based on the obtained disease related arrangement; and

(C)(iii) classifying some of the depolarising regions as forming at least a part of a retinal pigment epithelium using the identified curve.

22. A computer-implemented method for determining a retinal pigment epithelium, the method comprising:

receiving captured image data of the eye, the captured image data comprising multiple points, each point being characterised by a degree of polarisation uniformity and an intensity value;

extracting, for each point, a plurality of features based on at least a degree of polarisation uniformity and intensity values associated with a corresponding region containing said point; and

determining a likelihood score that a point of the captured image data belongs to the retinal pigment epithelium based on the extracted plurality of features to determine the retinal pigment epithelium in the captured image data.