Retinal OCT data processing

Patent metadata

Field	Value
Publication number	US-11625826-B2
Application number	US-202017062999-A
Country	US
Kind code	B2
Filing date	Oct 5, 2020
Priority date	Oct 7, 2019
Publication date	Apr 11, 2023
Grant date	Apr 11, 2023

Abstract

A method of method of processing optical coherence tomography, OCT, image data representing an OCT image of a retina of an eye, to generate mapping data which maps out a predetermined band of a plurality of distinct bands which extend across the OCT image and correspond to respective anatomical layers of the retina. The method comprises: receiving the OCT image data; processing A-scan data of the received OCT image data to generate data indicative of sequences of A-scan elements corresponding to the predetermined band of the plurality of distinct bands and having respective A-scan element values that vary in accordance with a predetermined pattern; and generating the mapping data by applying a line-finding algorithm to determine a line passing through the sequences of A-scan elements indicated by the generated data.

First claim

The invention claimed is:

1. A method of processing optical coherence tomography, OCT, image data representing an OCT image of a retina of an eye, to generate mapping data which maps out a predetermined band of a plurality of distinct bands which extend across the OCT image and correspond to respective anatomical layers of the retina, the method comprising:
receiving the OCT image data;
processing A-scan data of the received OCT image data to generate a resulting map indicative of sequences of A-scan elements corresponding to the predetermined band of the plurality of distinct bands and having respective A-scan element values that vary in accordance with a predetermined pattern, wherein the processing includes:
selecting a kernel based on the predetermined band, and
calculating values of one of a cross-correlation or a convolution of the A-scan elements with the kernel to generate the resulting map; and

generating the mapping data by applying a line-finding algorithm to the resulting map obtained in the calculating to determine a line passing through the sequences of A-scan elements indicated by the resulting map, based on values of coordinates of the resulting map, wherein the line has an associated total value which is higher than total values of other lines passing through the resulting map, the total value of each line being a sum of calculated values associated with coordinate positions through which the line passes.

2. The method according to claim 1 , wherein, in the selecting, the kernel is selected based on a variation of A-scan element values of the predetermined band compared to A-scan element values in immediately adjacent bands of the plurality of distinct bands, the kernel being configured to accentuate sequences of A-scan elements having respective A-scan element values that-vary in accordance with the predetermined pattern.

3. The method according to claim 2 , wherein the kernel is one of a sine Gaussian kernel and a sine-squared Gaussian kernel.

4. The method according to claim 2 , further comprising concatenating the A-scan data to form a one-dimensional array of concatenated A-scan data, wherein the calculating is performed by calculating values of one of a cross-correlation or a convolution between the one-dimensional array of concatenated A-scan data and the kernel to generate the resulting map.

5. The method according to claim 1 , wherein the A-scan data of the received OCT image data is processed to generate the resulting map indicative of the sequences of A-scan elements by using a feature enhancing algorithm to enhance A-scan elements belonging to respective sequences of A-scan elements whose A-scan element values vary in accordance with the predetermined pattern.

6. The method according to claim 1 , wherein the line-finding algorithm comprises a Viterbi algorithm.

7. The method according to claim 1 , wherein the predetermined one of the plurality of distinct bands corresponds to one of an inner segment/outer segment junction layer of the retina and a photoreceptor outer segment of the retina.

8. The method according to claim 1 , wherein the received OCT image data comprises one of OCT B-scan data and OCT C-scan data.

9. The method according to claim 1 , wherein the calculating is performed by calculating values of one of a cross-correlation or a convolution of a normalized version of the A-scan elements with a normalized version of the kernel to generate the resulting map.

10. The method according to claim 9 , further comprising normalizing the A-scan elements and the kernel.

11. An image processing apparatus configured to process optical coherence tomography, OCT, image data representing an OCT image of a retina of an eye, to generate mapping data which maps out a predetermined band of a plurality of distinct bands which extend across the OCT image and correspond to respective anatomical layers of the retina, the apparatus comprising:
receiver module configured to receive the OCT image data;
an A-scan processing module configured to process A-scan data of the received OCT image data to generate a resulting map indicative of sequences of A-scan elements corresponding to the predetermined band of the plurality of distinct bands and having respective A-scan element values that vary in accordance with a predetermined pattern, by:
selecting a kernel based on the predetermined band; and
calculating values of one of a cross-correlation or a convolution of the A-scan elements with the kernel to generate the resulting map; and

a mapping module configured to generate the mapping data by applying a line-finding algorithm to the resulting map obtained in the calculating to determine a line passing through the sequences of A-scan elements indicated by the resulting map, based on values of coordinates of the resulting map, wherein the line has an associated total value which is higher than total values of other lines passing through the resulting map, the total value of each line being a sum of calculated values associated with coordinate positions through which the line passes.

12. The image processing apparatus according to claim 11 , wherein in the selecting, the a kernel is selected based on a variation of A-scan element values of the predetermined band compared to A-scan element values in immediately adjacent bands of the plurality of distinct bands, the kernel being configured to accentuate sequences of A-scan elements having respective A-scan element values that vary in accordance with the predetermined pattern.

13. The image processing apparatus according to claim 12 , wherein the kernel is one of a sine-Gaussian kernel and a sine-squared Gaussian kernel.

14. The image processing apparatus according to claim 12 , wherein the A-scan processing module also is configured to concatenate the A-scan data to form a one-dimensional array of concatenated A-scan data, and performs the calculating by calculating values of one of a cross-correlation or a convolution between the one-dimensional array of concatenated A-scan data and the kernel to generate the resulting map.

15. The image processing apparatus according to claim 11 , wherein the A-scan processing module is configured to process the A-scan data of the received OCT image data to generate the resulting map indicative of the sequences of A-scan elements by using a feature enhancing algorithm to enhance A-scan elements belonging to respective sequences of A-scan elements whose A-scan element values vary in accordance with the predetermined pattern.

16. The image processing apparatus according to claim 11 , wherein the line-finding algorithm comprises a Viterbi algorithm.

17. The image processing apparatus according to claim 11 , wherein the predetermined one of the plurality of distinct bands corresponds to one of an inner segment/outer segment junction layer of the retina and a photoreceptor outer segment of the retina.

18. The image processing apparatus according to claim 11 , wherein the receiver module is configured to receive, as the OCT image data, at least one of OCT B-scan data comprising the A-scan data or OCT C-scan data comprising the A-scan data.

19. The image processing apparatus according to claim 11 , wherein the calculating is performed by calculating values of one of a cross-correlation or a convolution of a normalized version of the A-scan elements with a normalized version of the kernel to generate the resulting map.

20. The image processing apparatus method according to claim 19 , wherein the A-scan processing module is further config

Assignees

• Optos Plc

Inventors

• Wakeford Peter Robert

Classifications

• G06T7/12 Primary

  Edge-based segmentation · CPC title

• G06T2207/30041

  Eye; Retina; Ophthalmic · CPC title

• G16H30/20

  for handling medical images, e.g. DICOM, HL7 or PACS · CPC title

• G06T7/143

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

• A61B3/1225

  using coherent radiation · CPC title