Moti Freiman, Noah Broide, Miriam Natanzon, Lior Weizman, Einav Nammer, Ofek Shilon, Judith Frank, Leo Joskowicz, and Jacob Sosna
contact: freiman@cs.huji.ac.il

Abstract:

We present a nearly automatic graph-based segmentation method for patient specific modeling of the aortic arch and carotid
arteries from CTA scans for interventional radiology simulation. The method starts with the watershed-based segmentation of the
aorta and the construction of a prior intensity probability distribution function for arteries. The carotid arteries are then
segmented with a graph min-cut method based on a new edge weighting function that adaptively couples voxel intensity, intensity prior,
and geometric vesselness shape prior. Finally, the same graph-cut optimization framework is used to interactively remove a few
unwanted vessel segments and to fill in minor vessel discontinuities caused by intensity variations.
We conducted three experimental studies on 71 multicenter clinical CTA datasets for whichmultiobserver ground truth segmentations
were manually generated. The first study is the automatic and nearly-automatic segmentation of the carotid bifurcation lumen
on 56 CTAs from the MICCAI 2009 3D Segmentation Challenge. The second study is the automatic and interactive refinement
segmentation of the entire carotid vasculature on 15 CTAs. Our method successfully segmented in all cases the carotid bifurcation,
the aortic arch, the left/right subclavian arteries, and the common, internal, and external carotids and their secondary vessels. The
segmentations accuracy is comparable or better than those of other methods, without any user initialization or parameters adjustments.
The third study simulates on the ANGIO MentorTM common interventional radiology procedures on four patient-specific
models generated by our segmentation method. The simulations ran flawlessly for over an hour, with users reporting great realism
and an improved experience. The entire segmentation and simulation model generation takes less than 10mins of computation time
on a standard PC and only 1-2mins of end-user interaction. This constitutes a proof-of-concept of practical patient-specific carotid
interventional radiology simulations from CTA scans in a clinical environment. 

Keywords: Carotid arteries, CTA, segmentation, model-based graph-cut

• Freiman. M.  Broide. N. Natanzon. M. Weizman. L. Nammer. E. Shilon. O. Frank. J. Joskowicz. L. Sosna. J.
  Vessels-Cut: a graph based approach to carotid arteries patient-specific modeling.  
  Proc. 2nd workshop on: 3D Physiological Human, 2009 (pdf,bib).
  
  
• "Nearly automatic vessels segmentation using graph-based energy minimization".
  M. Freiman, J. Frank, L. Weizman, E. Nammer, O. Shilon, L. Joskowicz, J. Sosna
  Proc. 3rd 3D Segmentation in the Clinic: Carotid Lumen segmentation challenge workshop, 
  12th Int. Conf. on Medical Image Computing and Computer-Assisted Intervention, MICCAI'09, London, UK, 2009.  (pdf).  
  
  
• "Patient-specific modeling of the carotid arteries for surgical simulation".  
  M. freiman, M. Natanzon, N. Broide, L. Joskowicz,L. Weizman. E. Nammer, O Shilon. 
  Proc. 23rd Int. Conf. Computer-Assisted Radiology and Surgery, CARS'2009, Berlin, Germany, June 2009.  (abstract)

Related materials:

• View movies:
  • Patient specific simualtion with our model (mpeg, 100mb).
  • Entire 3D model - case1: mpeg.
  • Entire 3D model - case2: mpeg.
  • Entire 3D model - case5: mpeg.
  • Entire 3D model - case7: mpeg.
  • 2D axial view, results (red) compared to ground truth (green) - case 1: mpeg.
  • 2D axial view, results (red) compared to ground truth (green) - case 3: mpeg.
  • Carotid bifurcation - case 002: mpeg.
  • Carotid bifurcation - case 102: mpeg.
  • Carotid bifurcation - case 107: mpeg.

Links: