Skip to main content

Table 6 Processing costs for calculating feature sets as well as the performance of each feature set. The costs were averaged on ten random images selected from all classes of both 2D and 3D fluorescence microscope images. Costs shown for feature sets include preprocessing (and segmentation) costs.

From: Boosting accuracy of automated classification of fluorescence microscope images for location proteomics

Operation

Number of Features

CPU time (s)

Overall Accuracy (%)

2D preprocessing

N/A

0.6

N/A

3D preprocessing

N/A

21.1

N/A

3D segmentation

N/A

6.8

N/A

SLF8 (2D)

32

13.2

89.4

SLF15 (2D)

44

68.3

91.5

SLF13 (2D DNA)

31

10.8

90.7

SLF16 (2D DNA)

47

66.3

92.3

SLF14 (3D)

14

31.5

89.8

SLF10 (3D DNA)

9

32.0

95.8