Abstract
In this paper we introduce a novel hybrid graph-based approach for Guide-wire tracking. The image support is captured by steerable filters and improved through tensor voting. Then, a graphical model is considered that represents guide-wire extraction/tracking through a B-spline control-point model. Points with strong geometric interest (landmarks) are automatically determined and anchored to such a representation. Tracking is then performed through discrete MRFs that optimize the spatio-temporal positions of the control points while establishing landmark temporal correspondences. Promising results demonstrate the potentials of our method.
Original language | English (US) |
---|---|
Pages (from-to) | 9-16 |
Number of pages | 8 |
Journal | Medical image computing and computer-assisted intervention : MICCAI ... International Conference on Medical Image Computing and Computer-Assisted Intervention |
Volume | 14 |
Issue number | Pt 1 |
State | Published - 2011 |
Externally published | Yes |
ASJC Scopus subject areas
- General Medicine