Systems and methods for automatic vertebra edge detection, segmentation and identification in 3D imaging

1. A system for automated segmentation of anatomical structures in image data comprising:

a digital image input for receiving at least one digital image of an anatomical structure and;

a storage device, coupled to the image input for storing the at least one digital image; and

a processor coupled to the storage device and configured to:

receive the at least one digital image;

analyze the at least one digital image and automatically locate at least one anatomical landmark;

select a template from a database of pre-defined templates based upon the at least one anatomical landmark;

segment the at least one digital image by applying a coarse segmentation and a fine segmentation to obtain an optimized segmented template,

wherein applying the coarse segmentation comprises, for each of a plurality of sub-regions of the anatomical structure, applying a corresponding sub-region of the selected template to the anatomical structure, calculating a first edge response, deforming the sub-region of the selected template, calculating a deformation edge response, selecting the sub-region or deformed sub-region of the template having a highest edge response, and combining the highest edge-response sub-region from each of the plurality of sub-regions into a single coarse-segmented template,

wherein applying the fine segmentation comprises, for each of a plurality of patches of the coarse-segmented template, calculating an edge-response of the patch relative to the anatomical structure, deforming the patch, calculating an edge-response of the deformed patch, selecting the patch or deformed patch having a highest edge-response, and combining the highest edge-response patches into a single optimized segmented template.

2. The system of claim 1 , wherein the at least one digital image comprises at least one three-dimensional (3D) image, and the template comprises a 3D surface template.

3. The system of claim 1 , wherein receiving the at least one digital image further comprises acquiring the at least one digital image using an imager selected from: computer tomography (CT), positron emission tomography (PET), single photon emission computed tomography (SPECT), magnetic resonance (MR), and ultrasound.

4. The system of claim 1 , wherein the processor being configured to select a template further comprises the processor being configured to:

select a template from a database of pre-defined templates based upon the at least one anatomical landmark;

perform a plurality of rigid transformation on the selected template;

calculate a fit response for each transformed template; and

select a transformed template having a highest fit response.

5. The system of claim 1 , wherein deforming the template further comprises:

performing a plurality of translations of the template along each of the x, y, and z axes;

performing a plurality of re-scalings of the template along each of the x, y, and z axes; and

performing a plurality of rotations of the template along each of the x, y, and z axes.

6. The system of claim 1 , wherein, subsequent to combining the highest edge-response sub-region from each of the plurality of sub-regions, the processor is further configured to perform a smoothing operation to smooth the single coarse-segmented template.

7. The system of claim 1 , wherein deforming the patch further comprises:

calculating a mean normal for the patch; and

deforming the patch along the mean normal.

8. The system of claim 1 , wherein, subsequent to combining the highest edge-response patches, the processor is further configured perform a smoothing operation to smooth the single optimized segmented template.

9. A method for automated segmentation of anatomical structures in image data, comprising:

acquiring at least one medical image of an anatomical structure;

automatically locating at least one anatomical landmark in the at least one medical image;

selecting a template from a database of pre-defined templates based at least in part on the at least one anatomical landmark;

segmenting the at least one medical image by applying a coarse segmentation and a fine segmentation to obtain an optimized segmented template,

wherein applying the coarse segmentation comprises, for each of a plurality of sub-regions of the anatomical structure, applying a corresponding sub-region of the selected template to the image of the anatomical structure, calculating a first edge-response, deforming the sub-region of the selected template, calculating a deformation edge-response, selecting the sub-region or deformed sub-region of the template having a highest edge-response, and combining the highest edge response sub-region from each of the plurality of sub-regions into a single coarse-segmented template,

wherein applying a fine segmentation comprises, for each of a plurality of patches of the coarse-segmented template, calculating an edge-response of the patch relative to the anatomical structure, deforming the patch, calculating an edge-response of the deformed patch, selecting the patch or deformed patch having a highest edge-response, and combining the highest edge-response patches into a single optimized segmented template.

10. The method of claim 9 , wherein the at least one digital image comprises at least one three-dimensional (3D) image.

11. The method of claim 9 , wherein acquiring the at least one digital image comprises acquiring the at least one digital image using an imaging technique selected from: computer tomography (CT), positron emission tomography (PET), single photon emission computed tomography (SPECT), magnetic resonance (MR), and ultrasound.

12. The method of claim 9 , wherein selecting a template further comprises:

selecting a template from a database of pre-defined templates based at least in part on the at least one anatomical landmark;

performing a plurality of rigid transformation on the selected template;

calculating a fit response for each transformed template; and

selecting a transformed template having a highest fit response.

13. The method of claim 9 , wherein deforming the template further comprises:

performing a plurality of translations of the template along each of the x, y, and z axes;

performing a plurality of re-scalings of the template along each of the x, y, and z axes; and

performing a plurality of rotations of the template along each of the x, y, and z axes.

14. The method of claim 9 , further comprising, subsequent to combining the highest edge-response sub-region from each of the plurality of sub-regions, performing a smoothing operation to smooth the single coarse-segmented template.

15. The method of claim 9 , wherein deforming the patch further comprises:

calculating a mean normal for the patch; and

deforming the patch along the mean normal.

16. The method of claim 9 , further comprising, subsequent to combining the highest edge-response patches, performing a smoothing operation to smooth the single optimized segmented template.