Digital Signal Processing Reference
In-Depth Information
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Figure 11.1
Cluster assignment maps for the “neural gas” network of a dynamic perfusion MRI
study in a subject with a stroke in the right basal ganglia. Self-controlled
hierarchical neural network clustering of PTCs
x
(
τ
) was performed by the “neural
gas” network employing 16 CVs (i.e., a maximal number of 16 separate clusters at
the end of the hierarchical VQ procedure). For a better orientation, an anatomic
EPI scan of the analyzed slice is underlaid.
apixel
ν
characterized by the PTC
x
ν
to
a specific cluster
j
is based on a minimal distance criterion in the
n
-
dimensional time series feature space (i.e.,
ν
is assigned to cluster
j
), if
the distance
=(
x
ν
(
τ
))
,τ
∈{
1
,...,n
}
is minimal, where
w
j
denotes the CV belonging
to cluster
j
. Each CV represents the weighted mean value of all the
PTCs belonging to this cluster.
Self-controlled hierarchical neural network clustering of PTCs
x
(
τ
)
x
ν
−
w
j
Search WWH ::
Custom Search