Rough Hybrid Scheme: An application of breast cancer imaging - Rough Fuzzy Image Analysis: Foundations and Methodologies - page 91

image1 HE FHH Fuzzyrule

image2 HE FHH Fuzzyrule

image3 HE FHH Fuzzyrule

TABLE 5.1 Contrast enhancement results

where τ r is the number of times a rule appears in all reduct and ρ r is the number of
reduct sets.
The quality of rules is related to the corresponding reduct(s) which are generating rules
that cover the largest parts of the universe U . Covering U with more general rules implies
smaller size of a rule set. Importance rule criteria introduced in (Aboul Ella and Dominik,
2006) were used to study the rules' importance.

5.4 Results and Discussion

In this section, the results of all processes using the proposed hybridization technique are
discussed. To evaluate the visual performance of the algorithm, a number of images contain-
ing masses from the Mammographic Image Analysis Society (MIAS) database were selected
(MIA, 2003).

Table 5.1 illustrates the contrast enhancement results: (a) images 1, 2, 3, show the original
images; (b) the HE images show the histogram equalization enhancement (HE) result; (c)
the FHH images show the Fuzzy Histogram Hyperbolization (FHH) enhancement result;
(d)the fuzzy rule images represent the fuzzy rule based result.
Table 5.2 depicts S-FCM and M-FCM visual clustering results with different initiation
parameters. The weight parameter of cost function is ranging from 0.001 to 0.0000001 and
the clusters' number ranging from three to eight. From the obtained results, we observe that
both algorithms get the same good results with higher number of clusters and small weight
of the cost function. The average of segmentation accuracy of the introduced algorithm is
about 3.9837% error, which means that it is robust enough.