1. An Interactive Virtual Endoscopy Tool With Automated Path Generation Delphine Nain, MIT AI Laboratory. Thesis Advisor : W. Eric. L Grimson, MIT AI Laboratory.

2. Presentation Overview Background and Motivation Interactive System Central Path Planning Algorithm Synchronized Virtual Endoscopy Conclusion

3. Medical Motivation Cancer is the 2 nd cause of death in the US 43 % of people have a risk to be diagnosed with cancer –Out of those 88 % are cancer in inner organ How can “see” inside the body to screen and cure?

4. Conventional Endoscopy advantages: –minimally invasive –high resolution –interactivity disadvantages: –can be painful and uncomfortable –limited exploration

5. Conventional Medical Imaging

6. Conventional Visualization advantages: –non invasive –information on tissue shape through and beyond walls of organ disadvantages: –mentally align contiguous slides –lower resolution than video

7. Segmentation: Volume

8. 3D Reconstruction : Model

9. 3D Visualization

10. Virtual Endoscopy Combines strengths of previous alternatives on patient-specific dataset –Spatial exploration –Cross-correlation with original volume Compact and Intuitive way to explore huge amount of information

11. Virtual Endoscopy: advantages clinical studies: –planning and post-operation: generates views that are not observable in actual endoscopic examinations –color coding algorithms give supplemental information

12. Virtual Colonoscopy

13. System Requirements Combination of Interactivity and Automation is key Cross Reference between 3D models and grayscale volumes

14. Presentation Overview Background and Motivation Interactive System Central Path Planning Algorithm Synchronized Virtual Endoscopy Conclusion

15. Display

16. Navigation Interface

17. Cross Reference Provided by Arjan Welmers

18. Path: Update

19. Applications: Middle Ear Thomas Rodt Soenke Bartling

20. Applications: Cardiovascular Provided by Bonglin Chung

21. Presentation Overview Background and Motivation Interactive System Central Path Planning Algorithm Synchronized Virtual Endoscopy Conclusion

22. Automated Path Planning Goal: provide a “create path” button that produces a centerline inside a 3D model of any topology

23. Input

24. Output

25. Step 1: Produce a Labelmap

26. Step 2: Produce a distance map

27. Step 3: Create a Graph Create a Graph description of the Distance Map Nodes are voxels inside the model Edge weight are 1/(distance) 2 from the wall of the organ

28. Step 4: Run modified Dijkstra Dijkstra algorithm is a single source shortest path algorithm We use a binary heap An optimization: keep an evolving front, stop when reach the end node

29. Step 5: Results Running Time: ~7s

30. Step 5: Results Running Time: ~3s

31. Presentation Overview Background and Motivation Interactive System Central Path Planning Algorithm Synchronized Virtual Endoscopy Conclusion

32. Synchronized Virtual Colonoscopy

33. Dynamic Programming

34. Results

35. Conclusion Combination of Automation and Interactivity is key Cross Reference is important Synchronized Fly-Throughs is novel contribution Publication: D. Nain, S. Haker, E. Grimson, R. Kikinis “An Interactive Virtual Endoscopy Tool”, IMIVA workshop, MICCAI 2001.

36. Acknowledgements Ron Kikinis Steve Haker Lauren O’Donnell David Gering Carl-Fredrik Westin Peter Everett Sandy Wells Eric Cosman Polina Golland Soenke Bartling John Fisher Mike Halle Ferenc Jolesz

37. Thank You! Steve Haker, Hoon Ji, Connie Sehnert

38. Correspondance T is transformation matrix (translation or rotation along local axis) VC = To uniquely determine the coordinates of the virtual camera: coordinates of camera : VC new = VC old * T coordinates of the focal point: FP new = VC new * T

39. Cross Reference Provided by Arjan Welmers

40. 3D Visualization

41. Synchronized Virtual Endoscopy