Shape modeling and detection of catheter

• US 8,374,415 B2
• Filed: 04/05/2011
• Issued: 02/12/2013
• Est. Priority Date: 05/19/2010
• Status: Active Grant

• Current Assignee: Siemens Healthineers AG (Siemens AG)
• Original Assignee: Siemens Corp. (Siemens AG)
• Inventors: Yi, Sheng, Chen, Terrence, Wang, Peng, Comaniciu, Dorin
• Primary Examiner(s): AZARIAN, SEYED H
• Application Number:
  US13/079,846
• Publication Number:
  US 20110288404A1
• Time in Patent Office: 679 Days
• Field of Search: 382/100, 382/103, 382106-107, 382128-134, 382/162, 382/168, 382/173, 382/181, 382/203, 382/224, 382/232, 382/254, 382/274, 382/276, 382/305, 382/312, 378/4, 378/21, 378/98.12, 378/42, 128/899, 600/424
• US Class Current: 382/131
• CPC Class Codes: A61B 6/12 Arrangements for detecting or locating foreign bodies
• CPC Class Codes: A61B 6/487 involving fluoroscopy
• CPC Class Codes: A61B 6/5211 involving processing of medical diagnostic data
• CPC Class Codes: G16H 50/20 for computer-aided diagnosis, e.g. based on medical expert systems

Abstract

A method and system for detecting and modeling a catheter in a fluoroscopic image is disclosed. Catheter tip candidates and catheter body candidates are detected in the fluoroscopic image. One of a plurality of trained shape models is fitted to the catheter tip candidates and the catheter body candidates in order to model a shape of the catheter in the fluoroscopic image.

Claims

1. A method for detecting and modeling a catheter in a fluoroscopic image, comprising:

detecting catheter tip candidates in the fluoroscopic image;

detecting catheter body candidates in the fluoroscopic image; and

fitting one of a plurality of trained shape models to the catheter tip candidates and the catheter body candidates by calculating a probability of each of the plurality of trained shape models, the probability representing the likelihood of the respective trained shape model matching the catheter tip candidates and the catheter body candidates.

2. The method of claim 1, wherein the plurality of trained shape models are grouped into a plurality of clusters and fitting one of a plurality of trained shape models to the catheter tip candidates and the catheter body candidates comprises:

selecting one of the clusters that best matches the catheter tip candidates and the catheter body candidates; and

selecting one of the trained shape models of the selected cluster that best matches the catheter tip candidates and the catheter body candidates.

3. The method of claim 2, further comprising:

determining a general planar curve corresponding to the selected one of the plurality of trained shape models,

wherein the trained shape models and general planar curve are trained based on annotated training data.

4. The method of claim 1, further comprising:

fitting the trained shape model in a learned subspace detected using principle component analysis to reduce parameters associated with the trained shape models.

5. The method of claim 1, wherein detecting catheter tip candidates is performed by a first classifier trained using a Probabilistic Boosting Tree and detecting catheter body candidates is performed by a second classifier trained using a Probabilistic Boosting Tree.

6. The method of claim 1, wherein detecting catheter body candidates is based upon positions of the catheter tip candidates within the fluoroscopic image.

7. The method of claim 1, wherein the fluoroscopic image is one of a plurality of fluoroscopic images in a fluoroscopic image sequence.

8. A system for detecting and modeling a catheter in a fluoroscopic image, comprising:

means for detecting catheter tip candidates in the fluoroscopic image;

means for detecting catheter body candidates in the fluoroscopic image; and

means for fitting one of a plurality of trained shape models to the catheter tip candidates and the catheter body candidates by calculating a probability of each of the plurality of trained shape models, the probability representing the likelihood of the respective trained shape model matching the catheter tip candidates and the catheter body candidates.

9. The system of claim 8, wherein the plurality of trained shape models are grouped into a plurality of clusters and means for fitting one of a plurality of trained shape models to the catheter tip candidates and the catheter body candidates comprises:

means for selecting one of the clusters that best matches the catheter tip candidates and the catheter body candidates; and

means for selecting one of the trained shape models of the selected cluster that best matches the catheter tip candidates and the catheter body candidates.

10. The system of claim 9, further comprising:

means for determining a general planar curve corresponding to the selected one of the plurality of trained shape models,

means for wherein the trained shape models and general planar curve are trained based on annotated training data.

11. The system of claim 8, further comprising:

means for fitting the trained shape model in a learned subspace detected using principle component analysis to reduce parameters associated with the trained shape models.

12. The system of claim 8, wherein means for detecting catheter tip candidates comprises a first classifier trained using a Probabilistic Boosting Tree and means for detecting catheter body candidates comprises a second classifier trained using a Probabilistic Boosting Tree.

13. The system of claim 8, wherein means for detecting catheter body candidates uses positions of the catheter tip candidates within the fluoroscopic image.

14. The system of claim 8, wherein the fluoroscopic image is one of a plurality of fluoroscopic images in a fluoroscopic image sequence.

15. A computer readable medium encoded with computer program instructions for detecting and modeling a catheter in a fluoroscopic image, the computer program instructions defining the steps comprising:

detecting catheter tip candidates in the fluoroscopic image;

detecting catheter body candidates in the fluoroscopic image; and

fitting one of a plurality of trained shape models to the catheter tip candidates and the catheter body candidates by calculating a probability of each of the plurality of trained shape models, the probability representing the likelihood of the respective trained shape model matching the catheter tip candidates and the catheter body candidates.

16. The computer readable medium of claim 15, wherein the plurality of trained shape models are grouped into a plurality of clusters and the instructions defining the step of fitting one of a plurality of trained shape models to the catheter tip candidates and the catheter body candidates comprises the steps of:

selecting one of the clusters that best matches the catheter tip candidates and the catheter body candidates; and

selecting one of the trained shape models of the selected cluster that best matches the catheter tip candidates and the catheter body candidates.

17. The computer readable medium of claim 16, wherein the instructions further define the step comprising:

determining a general planar curve corresponding to the selected one of the plurality of trained shape models,

wherein the trained shape models and general planar curve are trained based on annotated training data.

18. The computer readable medium of claim 15, wherein the instructions further define the step comprising:

fitting the trained shape model in a learned subspace detected using principle component analysis to reduce parameters associated with the trained shape models.

19. The computer readable medium of claim 15, wherein detecting catheter tip candidates is performed by a first classifier trained using a Probabilistic Boosting Tree and the instructions defining the step of detecting catheter body candidates is performed by a second classifier trained using a Probabilistic Boosting Tree.

20. The computer readable medium of claim 15, wherein detecting catheter body candidates is based upon positions of the catheter tip candidates within the fluoroscopic image.

21. The computer readable medium of claim 15, wherein the fluoroscopic image is one of a plurality of fluoroscopic images in a fluoroscopic image sequence.