Overlap Analysis of Segmentation Results

Using overlap analysis, both sensitivity and the average number of false-positives per image can be determined for each predicted breast group. The results from overlap analysis are shown in Table 11.18. From this table, it can be seen that the average number of false positives over all breast types has risen slightly with

Table 11.18: Sensitivity and average number of false positives per image after segmentation of 200 abnormal and 200 normal mammograms using the adaptive knowledge-based model

Type

Sensitivity

FP/i

1

0.79

207.26

2

0.80

162.68

3

0.96

161.86

4

0.80

136.45

Mean

0.84

167.01

the inclusion of the 200 normal mammograms compared with the results presented in Table 11.16. The aim of the false-positive reduction knowledge-based component described is to reduce the false-positive count, while maintaining sensitivity in the detection of lesions. The next section describes how this is achieved in this configuration of the adaptive knowledge-based model.

0 0

Post a comment