BLOOD VESSEL IMAGE SEGMENTING METHOD AND APPARATUS USING PLURALITY OF PREDICTION RESULTS

Abstract

A blood vessel image segmentating method according to an embodiment may comprise the steps of: generating a plurality of candidate mask images regarding a target blood vessel by applying a plurality of blood vessel segmentation models on a blood vessel image; evaluating an error level for each of the generated plurality of candidate mask images; and generating a target blood vessel segmentation result from the candidate mask images, on the basis of the evaluated error level.

Description

Claims

1. A method, performed by a processor, of segmenting a blood vessel image, the method comprising: generating a plurality of candidate mask images regarding a target blood vessel by applying a plurality of blood vessel segmentation models to a blood vessel image;evaluating an error level for each of the generated plurality of candidate mask images; andgenerating a target blood vessel segmentation result from the candidate mask images based on the evaluated error level.

2. The method of claim 1, wherein the evaluating of the error level for each of the generated plurality of candidate mask images comprises, when at least one of pixels indicating a target blood vessel area in a corresponding candidate mask image is separated, evaluating the candidate mask image as an error.

3. The method of claim 1, wherein the evaluating of the error level for each of the generated plurality of candidate mask images comprises, when a number of pixels included in a blob other than a main blob in a corresponding candidate mask image is equal to or greater than a first threshold ratio compared to a number of pixels indicating a target blood vessel, evaluating the candidate mask image as an error.

4. The method of claim 1, wherein the evaluating of the error level for each of the generated plurality of candidate mask images comprises evaluating an error level of a corresponding candidate mask image based on a topology of an area indicating the target blood vessel in the corresponding candidate mask image.

5. The method of claim 4, wherein the evaluating of the error level of the candidate mask image based on the topology of the area indicating the target blood vessel comprises, based on a trend line calculated based on diameter information of the area indicating the target blood vessel in the candidate mask image, when there is an area having diameter information equal to or greater than a second threshold ratio from the trend line within the area indicating the target blood vessel, evaluating the candidate mask image as an error.

6. The method of claim 4, wherein the evaluating of the error level of the candidate mask image based on the topology of the area indicating the target blood vessel comprises, based on a trend line calculated based on brightness information of the area indicating the target blood vessel in the candidate mask image, when there is an area having a brightness difference equal to or greater than a third threshold ratio from the trend line within the area indicating the target blood vessel, evaluating the candidate mask image as an error.

7. The method of claim 1, wherein the evaluating of the error level for each of the generated plurality of candidate mask images comprises, when a length of a centerline of an area indicating a target blood vessel in a corresponding candidate mask image is equal to or less than a first threshold length, evaluating the candidate mask image as an error.

8. The method of claim 1, wherein the generating of the target blood vessel segmentation result from the candidate mask images based on the evaluated error level comprises generating a target blood vessel segmentation result based on candidate mask images obtained by excluding candidate mask images that are evaluated as errors from the plurality of candidate mask images.

9. The method of claim 8, wherein the generating of the target blood vessel segmentation result from the candidate mask images based on the evaluated error level comprises, when all of the plurality of candidate mask images are evaluated as errors, generating a target blood vessel segmentation result based on a candidate mask image having an error level equal to or less than a predetermined error level or a candidate mask image having a lowest error level.

10. The method of claim 1, wherein the evaluating of the error level for each of the generated plurality of candidate mask images comprises: calculating an error score based on connectivity between pixels indicating a target blood vessel in a corresponding candidate mask image, calculating an error score based on a blob indicating the target blood vessel, calculating an error score based on a topology of an area indicating the target blood vessel, and calculating an error score based on a length of a centerline of the area indicating the target blood vessel.

11. A non-transitory computer-readable storage medium storing instructions that, when executed by a processor, cause the processor to perform the method of claim 1.

12. An apparatus for segmenting a blood vessel image, the apparatus comprising: an image receiver configured to receive a blood vessel image; anda processor configured to generate a plurality of candidate mask images regarding a target blood vessel by applying a plurality of blood vessel segmentation models to the blood vessel image, evaluate an error level for each of the generated plurality of candidate mask images, and generate a target blood vessel segmentation result from the candidate mask images based on the evaluated error level.