System and method for learning relative distance in a shape space using image based features

• US 7,603,000 B2
• Filed: 08/16/2006
• Issued: 10/13/2009
• Est. Priority Date: 08/31/2005
• Status: Expired due to Fees

• Current Assignee: Siemens Medical Solutions USA Incorporated (Siemens AG)
• Original Assignee: Siemens Medical Solutions USA Incorporated (Siemens AG)
• Inventors: Zhou, Xiang, Georgescu, Bogdan, Zhou, Shaohua Kevin, Zheng, Yefeng, Comaniciu, Dorin
• Primary Examiner(s): Kassa; Yosef
• Application Number:
  US11/464,851
• Publication Number:
  US 20070046696A1
• Time in Patent Office: 1,154 Days
• Field of Search: 382/282, 382/286, 382/291, 382/296, 382/305
• US Class Current: 382/305
• CPC Class Codes: G06F 18/214 Generating training patterns; Bootstrap methods, e.g. bagging or boosting
• CPC Class Codes: G06F 18/28 Determining representative reference patterns, e.g. by averaging or distorting; Generating dictionaries
• CPC Class Codes: G06V 10/754 involving a deformation of the sample pattern or of the reference pattern; Elastic matching
• CPC Class Codes: G06V 10/772 Determining representative reference patterns, e.g. averaging or distorting patterns; Generating dictionaries
• CPC Class Codes: G06V 10/774 Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting

Abstract

A system and method for identifying a shape of an anatomical structure in an input image is disclosed. An input image is received and warped using a set of warping templates resulting in a set of warped images. An integral image is calculated for each warped image. Selected features are extracted based on the integral image. A boosted feature score is calculated for the combined selected features for each warped image. The warped images are ranked based on the boosted feature scores. A predetermined number of warped images are selected that have the largest feature scores. Each selected warped image is associated with its corresponding warping template. The corresponding warping templates are associated with stored shape models. The shape of the input image is identified based on the weighted average of the shapes models.

Claims

1. A method for populating a database with a set of images of an anatomical structure, the database being used to identify a shape of an anatomical structure in an input image, the method comprising the steps of:

receiving a set of images of anatomical structures, each image being annotated with a plurality of control points that identify a contour of the anatomical structure,

identifying a reference shape for the image set;

calculating a warping template for each image in the image set that warps that image into approximately the reference shape;

creating a matrix of images by applying each warping template calculated for each image in the image set to all of the images in the image set;

computing a shape difference for each image in the matrix by computing a distance measurement between each warped shape and the reference shape;

ranking the images based on the distance measurement;

identifying a set of features associated with the images;

computing feature scores for each image in the matrix;

using a ranking boosting process to selecting those combination of features that result in feature scores that produce a rank that is consistent with the rank associated with the warped image; and

storing the annotated contours of the anatomical structure, warping templates and the selected feature combination in the database.

2. The method of claim 1 wherein the step of computing feature scores for each image in the matrix further comprises the step of:

assigning a weight to each feature.

3. The method of claim 1 wherein the reference shape is the mean shape of all of the anatomical structures in the image set.

4. The method of claim 1 wherein the features are Haar features.

5. The method of claim 1 wherein the anatomical structure is a left ventricle and the shape is the contour of the left ventricle.

6. The method of claim 1 wherein the images are ultrasound images.

7. A method for identifying a shape of an anatomical structure in an input image, the method comprising the steps of:

receiving an input image;

warping the input image using a set of warping templates resulting in a set of warped images;

calculating an integral image for each warped image;

extracting selected features based on the integral image;

calculating a boosted feature score for the combined selected features for each warped image;

ranking the warped images based on the boosted feature scores;

selecting a predetermined number of warped images that have the largest feature scores;

associating each selected warped image with its corresponding warping template;

associating the corresponding warping templates with stored shape models ; and

identifying the shape of the input image based on the weighted average of the shapes models.

8. The method of claim 7 wherein the features are Haar features.

9. The method of claim 7 wherein the anatomical structure is a left ventricle and the shape is the contour of the left ventricle.

10. The method of claim 7 wherein the images are ultrasound images.

11. A system for non-rigid shape detection of an anatomical structure in an image, the system comprising;

a database that stores a set of images of anatomical structures, each image being annotated to indicate a contour of the shape of the anatomical structure, the database also storing selected features and combination weights;

a processor that receives an input image of an anatomical structure, the processor performing the following steps;

warping the input image using a warping template resulting in a set of warped versions of the input image;

calculating an integral image for each warped image;

extracting selected features based on the integral image;

calculating a boosted feature score for the combined selected features for each input warped image;

ranking the warped images based on the boosted feature scores;

selecting a predetermined number of warped images that have the largest feature scores;

associating each selected warped image with its corresponding warping template,

associating the corresponding warping templates with stored shape models;

identifying the shape of the input image based on the weighted average; and

a display to display the identified shape.

12. The system of claim 11 wherein the features are Haar features.

13. The system of claim 11 wherein the anatomical structure is a left ventricle and the shape is the contour of the left ventricle.

14. The system of claim 11 wherein the images are ultrasound images.