System and method for image segmentation

1. An image processing method implemented on at least one machine each of which has at least one processor and one storage, the method comprising:

acquiring image data, the image data including a plurality of ribs;

determining a rib region containing at least a portion of the plurality of ribs;

selecting, based on the rib region, at least one rib of the plurality of ribs as a target rib;

generating, based on an artificial intelligence algorithm, at least one rib-probability-map relating to the target rib;

determining, based on the image data, a starting point of the target rib, the starting point indicating a starting position for tracking the target rib;

tracking, based on the starting point and the at least one rib-probability-map, at least one portion of the target rib, wherein the at least one portion of the target rib is determined based on at least one rib model including:

determining, based on the at least one rib-probability-map, a predicted rib segment;

matching the predicted rib segment with the at least one rib model; and

in response to a determination that the predicted rib segment does not match with the at least one rib model, terminating tracking the at least one portion of the target rib; or

in response to a determination that the predicted rib segment matches with the at least one rib model, designating the predicted rib segment as a matched rib segment of the target rib; and

obtaining a segmented rib by segmenting the at least one portion of the target rib.

2. The method of claim 1 , wherein the selecting, based on the rib region; at least one rib of the plurality of ribs as a target rib comprises:

determining a seed point for the at least one rib of the plurality of ribs;

performing pre-segmentation based on the image data and the seed point to obtain a preliminary rib; and

designating, based on a determination that the preliminary rib is adhesive to a vertebra, the preliminary rib as the target rib for further segmentation, or

designating, based on a determination that the preliminary rib is not adhesive to a vertebra, the preliminary rib as the segmented rib.

3. The method of claim 1 ; wherein the determining a starting point of the target rib comprises:

determining a histogram based on a plurality of image layers of the target rib in a coronal plane; and

designating, based on the histogram, a characteristic point of the target rib as the starting point.

4. The method of claim 3 , wherein the determining a histogram comprises:

superimposing a plurality of rib pixels or voxels of the plurality of image layers along an anterior-posterior direction to obtain a diagram, each element at a position of the diagram representing a total number of pixels or voxels that are located at a corresponding position in one or more of the plurality of image layers and belong to a portion of the plurality of rib pixels or voxels, wherein each pixel or voxel of the portion of the plurality of rib pixels or voxels has a gray value larger than a first threshold; and

superimposing a plurality of elements of the diagram along a superior-inferior direction to obtain the histogram, each element of the histogram representing a sum of elements belonging to a portion of the plurality of elements, wherein all of the portion of the plurality of elements have a same position in a left-right direction.

5. The method of claim 3 , wherein the characteristic point is determined based on a position in the histogram, wherein a point at the position has a minimum value in the histogram.

6. The method of claim 1 , wherein the generating at least one rib-probability-map relating to the target rib comprises:

generating, based on a classifier, the at least one rib-probability-map, wherein the classifier is trained based on the artificial intelligence algorithm and a plurality of images relating to at least one sample rib.

7. The method of claim 1 , wherein the determining, based on the at least one rib-probability-map, a predicted rib segment comprises:

determining, based on the image data, a trace direction range; and

determining, based on the trace direction range and the at least one rib-probability-map, the predicted rib segment.

8. The method of claim 7 , wherein the determining the predicted rib segment comprises:

determining, within the trace direction range, at least one portion of the at least one rib-probability-map;

determining, based on the at least one portion of the at least one rib-probability-map, a trace direction; and

predicting, based on the trace direction, the predicted rib segment.

9. The method of claim 1 , further comprising:

in response to a determination that the predicted rib segment does not match with the at least one rib model,

performing, based on a plurality of matched rib segments, model reconstruction to obtain a reconstructed model; and

extracting, based on the plurality of matched rib segments, the at least one portion of the target rib.

10. The method of claim 1 , further comprising:

tracking, based on the matched rib segment of the target rib and the at least one rib-probability-map, a next rib segment of the target rib.

11. The method of claim 1 , wherein the target rib has a first end and a second end, wherein the first end of the target rib is spaced from a vertebra by a first distance, and the second end of the target rib is spaced from the vertebra by a second distance, and the first distance is larger than the second distance.

12. The method of claim 11 , wherein the determining a starting point of the target rib comprises:

designating a point of the target rib closer to the second end than to the first end of the target rib as the starting point.

13. The method of claim 11 , wherein the tracking at least one portion of the target rib comprises:

tracking the at least one portion of the target rib from the starting point to the second end of the target rib.

14. The method of claim 11 , wherein the obtaining a segmented rib by segmenting the at least one portion of the target rib comprises:

segmenting a first portion of the target rib using a first segmentation algorithm, wherein the first portion includes a region between the starting point and the first end of the target rib; and

combining the first portion of the target rib and the segmented rib to obtain the target rib.

15. The method of claim 14 , wherein the first segmentation algorithm is a region growing algorithm.

16. The method of claim 1 , further comprising:

labelling the segmented rib.

17. A system comprising:

at least one processor, and

a storage configured to store instructions, the instructions, when executed by the at least one processor, causing the system to effectuate a method comprising:

acquiring image data, the image data including a plurality of ribs;

determining a rib region containing at least a portion of the plurality of ribs;

selecting, based on the rib region, at least one rib of the plurality of ribs as a target rib;

generating; based on an artificial intelligence algorithm, at least one rib-probability-map relating to the target rib;

determining, based on the image data, a starting point of the target rib, the starting point indicating a starting position for tracking the target rib;

tracking, based on the starting point and the at least one rib-probability-map, at least one portion of the target rib, wherein the at least one portion of the target rib is determined based on at least one rib model including:

determining, based on the at least one rib-probability-map, a predicted rib segment;

matching the predicted rib segment with the at least one rib model; and

in response to a determination that the predicted rib segment does not match with the at least one rib model, terminating tracking the at least one portion of the target rib; or

in response to a determination that the predicted rib segment matches with the at least one rib model, designating the predicted rib segment as a matched rib segment of the target rib; and

obtaining a segmented rib by segmenting the at least one portion of the target rib.

18. The system of claim 17 , wherein the selecting, based on the rib region, at least one rib of the plurality of ribs as a target rib comprises:

determining a seed point for the at least one rib of the plurality of ribs;

performing pre-segmentation based on the image data and the seed point to obtain a preliminary rib; and

designating, based on a determination that the preliminary rib is adhesive to a vertebra, the preliminary rib as the target rib for further segmentation, or

designating, based on a determination that the preliminary rib is not adhesive to a vertebra, the preliminary rib as the segmented rib.

19. The system of claim 17 , wherein the determining a starting point of the target rib comprises:

determining a histogram based on a plurality of image layers of the target rib in a coronal plane; and

designating, based on the histogram, a characteristic point of the target rib as the starting point.

20. A non-transitory computer readable medium storing instructions, the instructions, when executed by at least one processor, causing the at least one processor to implement a method comprising:

acquiring image data, the image data including a plurality of ribs;

determining a rib region containing at least a portion of the plurality of ribs;

selecting, based on the rib region, at least one rib of the plurality of ribs as a target rib;

generating, based on an artificial intelligence algorithm, at least one rib-probability-map relating to the target rib;

determining, based on the image data, a starting point of the target rib, the starting point indicating a starting position for tracking the target rib;

tracking, based on the starting point and the at least one rib-probability-map, at least one portion of the target rib, wherein the at least one portion of the target rib is determined based on at least one rib model including:

determining, based on the at least one rib-probability-map, a predicted rib segment;

matching the predicted rib segment with the at least one rib model; and

in response to a determination that the predicted rib segment does not match with the at least one rib model, terminating tracking the at least one portion of the target rib; or

in response to a determination that the predicted rib segment matches with the at least one rib model, designating the predicted rib segment as a matched rib segment of the target rib; and

obtaining a segmented rib by segmenting the at least one portion of the target rib.