Learning-based spine vertebra localization and segmentation in 3D CT
Described herein is a novel method and system for segmentation of the spine using 3D volumetric data. In embodiments, a method includes an extracting step, localization step, and segmentation step. The extracting step comprises detecting the spine centerline and the spine canal centerline. The localization step comprises localizing the vertebra and intervertebral disc centers. Background and foreground constraints are created for each vertebra digit. Segmentation is performed for each vertebra digit and based on the hard constraints.
1. A method for identifying an anatomic structure of a patient, wherein the anatomic structure comprises a plurality of interconnected components, the method comprising:
receiving a 3D image data set of the patient;
detecting a characteristic feature of each of the interconnected components of the anatomic structure from the 3D image data set of the patient wherein the detecting is based on a prediction map, wherein the plurality of interconnected components of the anatomic structure comprises a plurality of ribs;
building case-specific foreground constraints based on the characteristic feature for each rib;
building case-specific background constraints based on components adjacent to each rib;
automatically segmenting the anatomic structure based on the detecting step and the building steps; and
displaying the segmented anatomic structure.
2. The method of claim 1 wherein the characteristic feature is a centerline of the anatomic structure.
3. The method of claim 2 wherein the detecting is performed based on optimizing prediction values of the prediction map.
4. The method of claim 3 wherein the optimizing prediction values of the prediction map is based on a shortest path algorithm.
5. The method of claim 2 further comprising localizing a center of the anatomical structure, and wherein the center is required to be on the centerline.
6. The method of claim 5 further comprising computing an augmented constrained region encompassing the anatomic structure, and based on the detecting step.
7. The method of claim 5 wherein the localizing step is estimated by a probabilistic inference algorithm.
8. The method of claim 5 further comprising constructing hard constraints for each rib, and wherein the automatically segmenting step is based on the constructing step.
9. The method of claim 1 wherein the prediction map is based on a machine learning algorithm.
10. The method of claim 1 wherein the anatomic structure is rigid.
11. The method of claim 1 further comprising identifying the anatomical structure and wherein the identifying is performed by a computer and based on an annotated exemplary anatomical structure.
12. The method of claim 1 further comprising use of the segmented anatomical structure for diagnosing, excising or ablating a tumor.
13. The method of claim 1 , wherein the building the background constraints comprises creating upper and lower geometric shapes that exclude the segmented anatomic structure.
14. A method for treatment planning comprising segmenting each of the ribs as described in claim 1 .
15. A system for segmenting an anatomical structure of a patient, the anatomic structure comprising a plurality of ribs, the system comprising:
a memory unit for storing 3D image data of the patient;
a programmed processor operable to:
detect a characteristic feature of each of the ribs and compute an augmented constrained region encompassing the ribs, wherein the characteristic feature comprises a rib centerline, and wherein the detecting is based on a prediction map;
localize a rib center, and wherein the center is required to be on the centerline;
build case-specific foreground constraints based on the characteristic feature for each rib;
build case-specific background constraints based on components adjacent to each rib;
segment the ribs based on the detected characteristic feature, localized center, case-specific constraints, and the augmented constrained region; and
a display in communication with the processor and for showing the segmented anatomical structure.
16. The system of claim 15 wherein the programmed processor is operable to construct hard constraints for the rib based on the rib centerline and localized center.
17. The system of claim 16 wherein the hard constraints comprise case-specific foreground and background constraints for the rib based on the rib centerline, center, and adjacent interconnected components.
18. The system of claim 15 wherein the prediction map is based on a machine learning algorithm.