Jaouen, Vincent
González, Paulo
Stute, Simon
Guilloteau, Denis
Chalon, Sylvie
Buvat, Irène
Tauber, Clovis
2014-09-04
<p>In this paper, we extend the gradient vector flow</p>
<p>field for robust variational segmentation of vector-valued images.</p>
<p>Rather than using scalar edge information, we define a vectorial</p>
<p>edge map derived from a weighted local structure tensor of</p>
<p>the image that enables the diffusion of the gradient vectors in</p>
<p>accurate directions through the 4D gradient vector flow equation.</p>
<p>To reduce the contribution of noise in the structure tensor,</p>
<p>image channels are weighted according to a blind estimator of</p>
<p>contrast. The method is applied to biological volume delineation</p>
<p>in dynamic PET imaging, and validated on realistic Monte Carlo</p>
<p>simulations of numerical phantoms as well as on real images.</p>
https://doi.org/10.1109/TIP.2014.2353854
oai:zenodo.org:16053
Zenodo
https://zenodo.org/communities/inmind
https://zenodo.org/communities/eu
info:eu-repo/semantics/restrictedAccess
IEEE Trans Image Process, 23(11), 4773-85, (2014-09-04)
deformable models
dynamic PET
gradient vector flow
structure tensor
Variational segmentation of vector-valued images with gradient vector flow.
info:eu-repo/semantics/article