Systems and methods for segmenting medical images based on anatomical landmark-based features

1. A computer-implemented method for segmenting medical images, the method comprising:

receiving an image acquired by an imaging device;

identifying a plurality of landmark points within the image, each landmark point corresponding to a known anatomical landmark;

selecting an image point in the image;

determining, by a processor, at least one feature for the image point relative to the plurality of landmark points, wherein the at least one feature comprises a geometric measurement determined from a shape formed by the image point and the plurality of landmark points; and

associating, by the processor, the image point with an anatomical structure by using a classification model based on the at least one determined feature.

2. The method of claim 1 , wherein the geometric measurement includes an area or a volume determined from the shape formed by the image point and the plurality of landmark points.

3. The method of claim 1 , wherein the at least one determined feature comprises a derived value based on at least one geometric measurement determined from a shape formed by the image point and the plurality of landmark points.

4. The method of claim 1 , wherein each landmark point is associated with a different anatomical structure of the image.

5. The method of claim 1 , further comprising generating a contour of the anatomical structure using the classification model.

6. The method of claim 1 , wherein each landmark point is associated with a different anatomical structure of a training image.

7. A computer-implemented method for training a classification model used for segmenting medical images acquired by an imaging device, the method comprising:

receiving a training image, the training image including a plurality of image points, each image point being assigned an anatomical structure;

identifying a plurality of landmark points in the training image, each landmark point corresponding to a known anatomical landmark;

determining one or more features for each image point in the training image relative to the landmark points, wherein at least one of the determined one or more features comprises a geometric measurement determined from a shape formed by the image point and the plurality of landmark points; and

using the plurality of image points as training data to train the classification model, wherein the classification model associates an unclassified image point with an anatomical structure based on the one or more determined features for the unclassified image point.

8. The method of claim 7 , wherein the plurality of image points are randomly selected within the training image.

9. The method of claim 7 , wherein the geometric measurement includes an area or a volume determined from the shape formed by the image point and the plurality of landmark points.

10. The method of claim 7 , wherein the at least one determined feature comprises a derived value based on at least one geometric measurement determined from a shape formed by the image point and the plurality of landmark points.

11. A non-transitory computer-readable medium storing computer-executable instructions that, when executed by a computer, cause the computer to perform a method for segmenting medical images, the method comprising:

receiving an image acquired by an imaging device;

identifying a plurality of landmark points within the image, each landmark point corresponding to a known anatomical landmark;

selecting an image point in the image;

determining at least one feature for the image point relative to the plurality of landmark points, wherein the at least one feature comprises a geometric measurement determined from a shape formed by the image point and the plurality of landmark points; and

associating the image point with an anatomical structure by using a classification model based on the at least one determined feature.

12. The computer-readable medium of claim 11 , wherein each landmark point is associated with a different anatomical structure of the image.

13. A non-transitory computer-readable medium storing computer-executable instructions that, when executed by a computer, cause the computer to perform a method for training a classification model used for segmenting medical images acquired by an imaging device, the method including:

receiving a training image, the training image including a plurality of image points, each image point being assigned an anatomical structure;

identifying a plurality of landmark points in the training image, each landmark point corresponding to a known anatomical landmark;

determining one or more features for each image point in the training image relative to the landmark points, wherein at least one of the determined the one or more features comprises a geometric measurement determined from a shape formed by the image point and the plurality of landmark points; and

using the plurality of image points as training data to train the classification model, wherein the classification model associates an unclassified image point with an anatomical structure based on the one or more determined features for the unclassified image point.

14. The computer-readable medium of claim 13 , wherein the at least one determined feature comprises a derived value of the geometric measurement.

15. The computer-readable medium of claim 13 , wherein each landmark point is associated with a different anatomical structure of the training image.

16. A system for segmenting medical images acquired by an imaging device using a classification model, the system comprising:

a database configured to store a plurality of images acquired by the imaging device; and

an image processor connected to the database to access the plurality of images, image processor configure to:

receive an image from the database;

identify a plurality of landmark points within the image, each landmark point corresponding to a known anatomical landmark;

select an image point in the image;

determine at least one feature for the image point relative to the plurality of landmark points, wherein the at least one feature comprises a geometric measurement determined from a shape formed by the image point and the plurality of landmark points; and

associate the image point with an anatomical structure by using the classification model, based on the at least one determined feature.

17. A system for training a classification model used for segmenting medical images acquired by an imaging device, the system comprising:

a database configured store a plurality of training images, each training image including a plurality of image points and each image point is assigned an anatomical structure; and

an image processor connected to the database to access the plurality of training images, the image processor configured to:

receive a training image from the database, the training image including a plurality of image points and each image point within the training image being assigned an anatomical structure;

identify a plurality of landmark points in the training image, each landmark point corresponding to a known anatomical landmark;

determine one or more features for each image point in the training image relative to the plurality of landmark points, wherein at least one of the determined one or more features comprises a geometric measurement determined from a shape formed by the image point and the plurality of landmark points; and

use the plurality of image points as training data to train the classification model, wherein the classification model associates an unclassified image point with an anatomical structure based on the one or more determined features for the unclassified image point.