Systems, methods, and devices for detecting anatomical features
A method of detecting a vertebral endplate rim according to embodiments of the present disclosure includes receiving, at a processor, a binary mask of a vertebra; applying a Laplacian filter to the binary mask to yield a processed image; identifying, in the processed image, a set of edges having a sharpness that exceeds a predetermined threshold; deformably registering a template to a plurality of points corresponding to at least some of the edges in the set of edges; excluding some of the plurality of points based on at least one predetermined criterion; and generating additional points by interpolating among a remaining subset of the plurality of points to yield a set of points defining an endplate rim of the vertebra.
1. A system for identifying vertebral endplate rims, comprising:
a communication interface;
at least one processor; and
a memory storing instructions for execution by the at least one processor that, when executed, cause the at least one processor to:
receive a binary mask of a 3D image of a vertebra;
identify a set of edges of the vertebra based on a determined change of gradient direction;
categorize each edge of the set of edges as a plausible edge or an implausible edge;
reject the implausible edges;
select a plurality of points based on the plausible edges; and
generate additional points by interpolating between adjacent ones of at least some of the plurality of points to fill gaps between the at least some of the plurality of points.
2. The system of claim 1 , wherein the memory stores additional instructions for execution by the at least one processor that, when executed, cause the at least one processor to output coordinates of a vertebral endplate rim.
3. The system of claim 1 , wherein the memory stores additional instructions for execution by the at least one processor that, when executed, cause the at least one processor to fit a template to the set of edges.
4. The system of claim 3 , wherein categorizing each edge of the set of edges comprises comparing a location of each of a plurality of fitted points on the template to at least two adjacent fitted points on the template.
5. The system of claim 4 , wherein categorizing each edge of the set of edges comprises categorizing any edge that extends more than a predetermined distance from the template as implausible.
6. The system of claim 1 , wherein identifying the set of edges comprises applying a Laplacian filter to the binary mask.
7. The system of claim 1 , wherein identifying the set of edges comprises identifying changes of gradient direction that exceed a predetermined magnitude.
8. The system of claim 1 , wherein the implausible edges are those with a length in a superior-inferior direction that is greater by at least a predetermined amount than each of a length in an anterior-posterior direction and a length in a lateral direction.
9. The system of claim 1 , wherein the additional points are generated by interpolating at regular intervals.
10. The system of claim 1 , wherein the memory stores additional instructions for execution by the at least one processor that, when executed, cause the at least one processor to:
exclude some of the plurality of points based on at least one predetermined criterion.
11. A method of detecting a vertebral endplate rim, comprising:
receiving, at a processor, a binary mask of a vertebra;
applying a Laplacian filter to the binary mask to yield a processed image;
identifying, in the processed image, a set of edges having a sharpness that exceeds a predetermined threshold;
deformably registering a template to a plurality of points corresponding to at least some of the edges in the set of edges;
excluding some of the plurality of points based on at least one predetermined criterion; and
generating additional points by interpolating among a remaining subset of the plurality of points to yield a set of points defining an endplate rim of the vertebra.
12. The method of claim 11 , wherein the at least one predetermined criterion comprises whether a point from the plurality of points is beyond a predetermined distance from a subset of the plurality of points.
13. The method of claim 11 , wherein the at least one predetermined criterion comprises whether a point from the plurality of points defines a contour that exceeds a predetermined sharpness.
14. The method of claim 11 , wherein the at least one predetermined criterion comprises whether a point from the plurality of points is closer to a plane fitted to the template than an adjacent point.
15. The method of claim 11 , wherein each of the at least some of the edges in the set of edges satisfy an edge criterion.
16. The method of claim 15 , wherein the edge criterion corresponds to a dimension of an edge in a superior-inferior direction relative to each of a dimension of the edge in an anterior-posterior direction and a dimension of the edge in a lateral direction.
17. An image processing device, comprising:
a communication interface;
at least one processor; and
a memory storing instructions for execution by the at least one processor that, when executed, cause the at least one processor to:
apply a Laplacian filter to a binary mask of a vertebra to yield a processed image, the binary mask received via the communication interface;
identify a set of plausible edges using the processed image;
register a template to a plurality of points, each of the plurality of points located on at least one plausible edge of the set of plausible edges;
apply a predetermined criterion to each point of the plurality of points to yield a rejected set comprising points that do not satisfy the predetermined criterion and an accepted set comprising points that do satisfy the predetermined criterion; and
generate at least one new point by interpolating between at least two points of the accepted set.
18. The image processing device of claim 17 , wherein the memory stores additional instructions for execution by the at least one processor that, when executed, further cause the at least one processor to determine coordinates of an endplate rim of the vertebra based on at least the accepted set.
19. The image processing device of claim 17 , wherein the at least one processor does not use a machine learning algorithm in connection with any of the instructions.
20. The image processing device of claim 17 , wherein each of the instructions is executed by the at least one processor automatically.
21. The image processing device of claim 17 , wherein the memory stores additional instructions for execution by the at least one processor that, when executed, further cause the at least one processor to fit a plane to the registered template, and further wherein the predetermined criterion is based on the fitted plane.