Biomedical Engineering Reference
In-Depth Information
a level set framework. We give an overview of these two families of methods in
the next section.
2.2.6
Level Set Speed Functions with Regularizers
Suri
et al.
review in [9] recent works on the fusion of classical geometric and
geodesic deformable models speed terms with regularizers, i.e., regional statis-
tics information from the image. Regularization of the level set speed term is
desirable to add prior information on the object to segment and prevent seg-
mentation errors when using only gradient-based information in the definition
of the speed. Four main types of regularizers were identified by the authors of
the review:
1. Clustering-based regularizers
2. Bayesian-based regularizers
3. Shape-based regularizers
4. Coupling-surfaces regularizers
We give in the next section a brief overview of each method.
(1)
Clustering-based Regularizers
: Suri proposed in [28] the following en-
ergy functional for level set segmentation:
∂φ
∂
t
=
(
εκ
+
V
p
)
|∇
φ
|−
V
ext
∇
φ,
(2.19)
where
V
p
is a regional force term expressed as a combination of the inside and
outside regional area of the propagating curve. This term is proportional to a
region indicator taking value between 0 and 1, derived from a fuzzy membership
measure as described in [29].
(2)
Bayesian-based Regularizers
: Recent work from Baillard
et al.
[30] pro-
posed an approach similar to the previous one where the level set energy func-
tional expressed as:
∂φ
∂
t
=
g
(
|∇
I
|
)(
κ
+
V
0
)
|∇
φ
|
(2.20)
uses a modified propagation term
V
0
as a local force term. This term was de-
rived from the probability density functions inside and outside the structure to
