Biomedical Engineering Reference
In-Depth Information
where
X
and
Y
denote two ra
nd
om v
ar
iables;
S
X
and
S
Y
denote the standard devia-
tion of
X
and
Y
, respectively;
X
and
Y
denote the means of
X
and
Y
, respectively; n
denotes the number of measurements of
X
and
Y
.
As a res
ult,
th
e co
mputed Pearson correlation of
FOC
and
FOM
,
repre-
sented by
r
(
FOM
)
,
amounts to
0.806; The
computed Pearson correlation of
FOC
and
F R AG
,
repr
esente
d by
r
FOC
(
F R AG
)
amounts to
0.82
4
; The
computed
Pearson correlation of
FOC
an
d
F
R
AG
,
r
epresented by
r
FOC
(
F R AG
)
amounts
FOC
to 0.707. It was expected that
r
(
F R AG
)
was of high correlation because both
of them depend on each other; if the segmented region has many incorrect labe-
ling, incorrect hand st
ructu
r
es are
expect
ed to
occur,
and vice versa. Similarly, the
high correlation in
r
FOC
(
F R AG
)
are expected since if edges
or regions were
not we
ll delineated, then it indicated that it has high likelihood to
suffer errors in
F R AG
,
and vice versa.
(
F R AG
)
and
r
FOC
FOC
4.4 Summary
In this chapter, the functionality of proposed MBOBHE and the modified aniso-
tropic diffusion have been justified. Both techniques create a suitable environment
for the subsequent segmentation modules by reducing the disturbance of uneven
illumination, sharpening the edge and smoothing the texture; at the same time,
MBOBHE is capable of enhancing the features of ossification sites to improve the
performance of computerized BAA. After that, an analytical comparison of AAM
segmentation framework with the proposed segmentation framework is performed.
The result showed that AAM requires a large number of user-specified parameters.
Furthermore, the nature of the parameters complicated the potential of being auto-
mated and hence reduces the automaticity. From the perspective of computerized
BAA, high automaticity implies that the proposed segmentation framework does
not require an expert to perform the segmentation after it has been finished mod-
eling. On the contrary, the AAM requires an expert to perform the task because
non-expert does not possess the knowledge of choosing the model, searching for
optimized location and determine the number of iterations. This in turn indicates
that the proposed segmentation framework can be operated by even non-expert,
and in fact, in this topic, the fully automated segmentation framework where the
users need not any extra knowledge to execute has been designed of which pos-
sesses perfect repeatability. In terms of accuracy, the segmentation improvement
of both the automated fuzzy quadruple division scheme and the quality assurance
process in the proposed segmentation framework has been justified. Then, the pro-
posed segmentation framework has been justified to possess high level of accuracy
and low level of errors despite using much less number of parameters than AAM.
The low standard deviation indicated high adaptability to radiographs contain-
ing different type of hand bones found in different age groups. The AAM, despite
having the potential to gain higher accuracy of segmentation, is perceived being
productively unrealistic in terms of production theory [
13
] because the added
Search WWH ::
Custom Search