General framework for image segmentation using ordered spatial dependency
View Patent ↗A general framework to enhance performance of automatic segmentation of a plurality of structures in medical imaging applications incorporates inter-structure spatial dependencies in to existing segmentation algorithms. Ranking the structures according to their dependencies allows a hierarchical approach to automatically segmenting multiple structures that improves each individual segmentation and provides automatic initializations.
1. A method for segmenting a plurality of target structures in three-dimensional image data comprising:
(a) providing pre-segmented training images of structures representing the target structures;
(b) modeling, using a computer, inter-structure structure ordered spatial dependency for each of the target structures from the pre-segmented training images by placing all pre-segmented training images in a common reference coordinate system using the known neighboring structures as reference anchor structures;
(c) integrating the ordered spatial dependency with a segmentation algorithm by registering anchor structures from the three-dimensional image data to the reference anchor structures used in step (b);
(d) determining an optimal sequence for segmentation of the plurality of target structures; and
(e) sequentially segmenting each of the plurality of target structures according to the optimal sequence thus restricting the plausible segmentation space for each subsequent target structure in the sequence to produce enhanced segmentations of the target structures.
2. The method of claim 1 , wherein the step (b) comprises estimating a warping between each instance of the reference anchor structure to the common reference coordinate system.
3. The method of claim 1 , wherein the step (d) comprises:
(f) identifying a group of possible segmentation sequences for the plurality of target structures;
(g) segmenting all target structures for each possible segmentation sequences using the ordered spatial dependency of the target structures;
(h) measuring the overall quality of the segmentation of the target structures in each of the sequences by comparing the results with the manual segmentation according to a similarity measure; and
(i) evaluating the similarity measures for each sequence to determine that sequence with the best similarity measure as the optimal sequence for segmentation.
4. A program storage device readable by a computer, tangibly embodying a program of instructions executable by the computer to perform method steps for segmenting a plurality of target structures in three-dimensional image data, the method steps comprising:
(a) providing pre-segmented training images of structures representing the target structures;
(b) modeling inter-structure ordered spatial dependency for each of the target structures from the pre-segmented training images by placing all pre-segmented training images in a common reference coordinate system using the known neighboring structures as reference anchor structures;
(c) integrating the ordered spatial dependency with a segmentation algorithm by registering anchor structures from the three-dimensional image data to the reference anchor structures used in step (b);
(d) determining an optimal sequence for segmentation of the plurality of target structures; and
(e) sequentially segmenting each of the plurality of target structures according to the optimal sequence thus restricting the plausible segmentation space for each subsequent target structure in the order to produce enhanced segmentations of the target structures.
5. The device of claim 4 , wherein the step (b) comprises estimating a warping between each instance of the reference anchor structure to the common reference coordinate system.
6. The device of claim 4 , wherein the step (d) comprises:
(f) identifying a group of possible segmentation sequences for the plurality of target structures;
(g) segmenting all target structures for each possible segmentation sequences using the ordered spatial dependency of the target structures;
(h) measuring the overall quality of the segmentation of the target structures in each of the sequences by comparing the results with the manual segmentation according to a similarity measure; and
(i) evaluating the similarity measures for each sequence to determine that sequence with the best similarity measure as the optimal sequence for segmentation.
7. A system for segmenting a plurality of target structures in three-dimensional image data comprising:
means for storing a program of executable instructions to perform method steps for segmenting a plurality of target structures in three-dimensional image data, the method steps comprising:
(a) providing pre-segmented training images of structures representing the target structures;
(b) modeling inter-structure ordered spatial dependency for each of the target structures from the pre-segmented training images by placing all pre-segmented training images in a common reference coordinate system using the known neighboring structures as reference anchor structures;
(c) integrating the ordered spatial dependency with a segmentation algorithm by registering anchor structures from the three-dimensional image data to the reference anchor structures used in step (b);
(d) determining an optimal sequence for segmentation of the plurality of target structures; and
(e) sequentially segmenting each of the plurality of target structures according to the optimal sequence thus restricting the plausible segmentation space for each subsequent target structure in the order to produce enhanced segmentations of the target structures;
means for receiving said three-dimensional image data; and
means for executing said program of executable instructions whereby segmenting said plurality of structures in said three-dimensional image data.
8. The system of claim 7 , wherein the step (b) comprises estimating a warping between each instance of the reference anchor structure to the common reference coordinate system.
9. The system of claim 7 , wherein the step (d) comprises:
(f) identifying a group of possible segmentation sequences for the plurality of target structures;
(g) segmenting all target structures for each possible segmentation sequences using the ordered spatial dependency of the target structures;
(h) measuring the overall quality of the segmentation of the target structures in each of the sequences by comparing the results with the manual segmentation according to a similarity measure; and
(i) evaluating the similarity measures for each sequence to determine that sequence with the best similarity measure as the optimal sequence for segmentation.