Biomedical Engineering Reference
In-Depth Information
delineating the anatomical structure; this was logical because if the edges of the
hand bone could have been completely detected, then it would not have needed
any segmentation techniques to partition the hand bone. Therefore, the Fig. 4.10 c
provided the main regions and the detected edges played the role as supplemen-
tary material that complete the hand bone outline as shown in Fig. 4.10 d. The
final segmented hand bone of Fig. 4.10 e was obtained after applying BRANEA in
Fig. 4.10 d.
4.3.2.3 Accuracy Evaluation of the Proposed Segmentation Framework
After inspecting the segmentation effect of BRANEA, quantitative evaluation
is conduct ed to f u rther justify its con sistency over all age groups of radiographs
by using FOM , FOC and F R AG of metrics described in Eqs. ( 4.11 - 4.13 )
This evaluation result amounted to evaluating the overall segmentation accuracy
of the entire proposed segmentation framework. The results were illustrated in
Figs. 4.11 , 4.12 , 4.13 , respectively, followed by res ult inte rpretations.
As shown in Fig. 4.11 , the expected value of FOM is around 0.9 (0.8958).
This suggested strongly that the segmented image contained most of the expected
hand skeletal anatomical borders of bones which is critical in computer-aided skel-
etal age scoring system in ossification localizatio n and t he bone age analysis as
well. Besides, it is noticeable that the dispersion of FOM is within a narrow range
between 0.8130 and 1.000. In terms of standard deviation, the data dispersion is
only 0.0502 which suggested that the precisions or consistency of detected bone
borders is very high. In other words, the segmentation framework could produce
1
0.8
0.6
0.4
0.2
Mean of FOM
Expected value of FOM over all Age group
0
0
2
4
6
8
10
12
14
16
18
Age Group
Fig. 4.11 The mean of FOM and the expected value
 
Search WWH ::




Custom Search