Leukocyte Detection Through an Evolutionary Method - Complex System Modelling and Control Through Intelligent Soft Computations - page 155

Table 2 Comparative leukocyte detection performance of the BSV approach, the IO method, the
Wang algorithm, the BGA detector and the proposed DE method over the data set which contains
30 images and 426 leukocytes
Leukocyte type Method Leukocytes

detected

Missing False

alarms

DR
(%)

FAR
(%)

Bright leukocytes
(287)

BSV [3] 130 157 84 45.30 29.27

IO [4] 227 60 73 79.09 25.43

Wang [5] 231 56 60 80.49 20.90

GAB [16] 220 67 22 76.65 7.66

DE-based 281 6 11 97.91 3.83

Dark leukocytes
(230)

BSV [3] 105 125 59 46.65 25.65

IO [4] 183 47 61 79.56 26.52

Wang [5] 196 34 47 85.22 20.43

GAB [16] 179 51 23 77.83 10.00

DE-based 227 3 3 98.70 1.30

Overall (517) BSV [3] 235 282 143 45.45 27.66

IO [4] 410 107 134 79.30 25.92

Wang [5] 427 90 107 82.59 20.70

GAB [16] 399 118 45 77.18 8.70

DE-based 508 9 14 98.26 2.71

and data acquisition systems. Therefore, the detection results depend on the algo-
rithm s ability to cope with different kinds of noises. In order to demonstrate the

robustness in the WBC detection, the proposed DE approach is compared to the
BSV approach, the IO method, the Wang algorithm and the BGA detector under
noisy environments. In the test, two different experiments have been studied.
The

'

first inquest explores the performance of each algorithm when the detection
task is accomplished over images corrupted by Salt and Pepper noise. The second
experiment considers images polluted by Gaussian noise. Salt and Pepper and
Gaussian noise are selected for the robustness analysis because they represent the
most compatible noise types commonly found in images of blood smear (Landi and
Piccolomini 2012 ). The comparison considers the complete set of 50 images pre-
sented in Sect. 6.1 containing 517 leukocytes which have been detected and
counted by a human expert.
The added noise is produced by MatLab © , considering two noise levels of 5 and

10 % for Salt and Pepper noise whereas 10 are used for the case of

Gaussian noise. Such noise levels, according to (Tapiovaara and Wagner 1993 ),
correspond to the best trade of between detection dif

r ¼ 5and r ¼

culty and the real existence in
medical imaging. If higher noise levels are used then the detection process would be
unnecessarily complicated without representing a feasible image condition.