Biomedical Engineering Reference
In-Depth Information
n
(
C
1
)
n
(
C
1
) +
n
(
C
2
)
log
2
n
(
C
1
)
n
(
C
1
) +
n
(
C
2
)
CER
(
n
(
C
1
)
,
n
(
C
2
)) =−
n
(
C
1
)
n
(
C
1
) +
n
(
C
2
)
n
(
C
1
)
n
(
C
1
) +
n
(
C
2
)
+
1
−
log
2
1
−
(3.58)
n
(
C
1
)
n
(
C
1
)+
n
(
C
2
)
onto the interval (0 1] to model the strenght of
the equality ratio of both clusters: If the
n
(
C
1
)
amounts to
n
(
C
2
)
, then a value of
unity is associated with it; on the contrary, as the
n
(
C
1
)
significantly deviates from
n
(
C
2
)
, then a value of approaching zero is associated with it, however, it will never
become zero as the minimum number of pixels reside in any cluster is one pixel.
This enables the logarithm function to be utilized in this modeling. The
CER
func-
tion versus ratio
n
(
C
1
)
n
(
C
1
)+
n
(
C
2
)
CER
map the ratio
is shown in Fig.
3.11
.
The
CER
function provided a numerical evaluation of the block content about
the ratio between the first cluster and the second cluster. If the ratio is relatively
high or relatively low, the
CER
results in low value, then the dividing process is
Input: cluster
members
Normalized Difference
Expectation (NDE)
Indicates Case 2,
further analysis is
required
Indicates Case 3,
further analysis is
required
Indicates
Case 1,
Close to each other
Bright
Quadruple
Division
: least likely
Homogeneity
Discrepancy Test(HDT) using
resultant region from ACR
using k=2 and k=3
Dark
Normalized Cluster Location
(NCL)
Far apart
Moderate
High
Low
Clustering Efficacy Ratio
(CER)
High
medium
Quadruple Division:
less likely
Quadruple
Division:
most likely
Quadruple
Division : less
likely
Quadruple
Division:
likely
Low
Output: new blocks
Fig. 3.12
The decision tree for quadruple division scheme
Search WWH ::
Custom Search