Result and Discussion - Segmentation of Hand Bone for Bone Age Assessment - page 121

Me an of quadruple scheme
Ex pe cted value of quadruple scheme
Me an of rigid scheme wi th row=4,column=4
Ex pe cted value of rigid scheme wi th row=4,column=4
Me an of rigid scheme wi th row=2,column=2
Ex pe cted value of rigid scheme wi th row=2,column=2
Me an of rigid scheme wi th row=3,column=3
Ex pe cted value of rigid scheme wi th row=3,column=3
Me an of rigid scheme wi th row=1,column=1
Ex pe cted value of rigid scheme wi th row=1,column=1

1.2

1

0.8

0.6

0.4

0.2

0

0 2 4 6 8 10 12 14 16 18

Ag e Group

Fig. 4.8 Comparisons of segmentation accuracy between manual rigid schemes with various
divisions and proposed automated fuzzy quadruple division scheme using the mean of FOC and
expected value over all age group

segmentation was not trivial but noticeable enough to justify the claimed utility of
the proposed scheme in retaining important regions of hand bones.
Figure 4.9 illustrated the comparison in terms of number of rela ted bones
that have been segmented. The improvement shown in result of FOC in above
experiments compar ed to othe r rigid schemes was even more tremendous than
the previous FOM and FOC . This result revealed that even without the qual-
ity assurance process, the proposed division scheme is capable of retaining most
of the critical type of bones that wou ld affec t the eventual bone age assessment
result. Quantitatively, the average of F R AG over all age groups is 0.6997, fol-
lowed by 0.1786 from rigid scheme of row = 4, column = 4, 0.1471 from
rigid scheme of row = 2, column = 2, 0.1198 from rigid scheme of row = 3,
column = 3, and 0.0844 from rigid schem e of row = 1, column = 1. The overall
average of these rigid schemes in terms of F R AG was 0.1324. Note that the over-
all ave rage of rigid scheme in this metrics are the lowest compared to the previous
FOM and FOC . This indicated that the rigid division scheme performed infe-
riorly especial ly in retaining the critical number of oss iicatio n sit es. In terms of
BAA, F R AG is considered to be more important than FOM and FOC because
a single missing of ossification sites would totally distort the final result of BAA.
As a summary to this sub-section, experimental result on proposed division
scheme with conventional rigid scheme, in a narrow sense, revealed that the pro-
posed scheme did play a pivotal role in the segmentation framework and thus this
justified the initial motivation of this scheme. In a broader sense, the result sug-
gested the general insight that merely an explicit alternation of the properties for