Performance of Color Cascading Framework on Different Color-Space for Malaria Identification

Rapid and accurate detection of malaria infection lead to better treatment for the patients. It can help in reducing the mortality due to the illness. We analyze the effectiveness of using several color-spaces, i.e., RGB, HSV, YC
_b
C
_r
, CIELAB, and grayscale in the Color Cascading Framework. Since each color space consists of several color channels, it is necessary to choose the right color channel to be used as the object of analysis so that the best system performance can be obtained. We propose measurements on each color channel by using a measurement of pixel intensity homogeneity using entropy method. From the experimental results, it was found that the color channel entropy value shows a direct relationship with the accuracy of the system. The highest color channel entropy value corresponded to the best accuracy value. The results of testing on each color-space revealed that the most suitable color channel for Color Cascading Framework was a Green color channel from RGB color-space, with performance accuracy, specificity, and sensitivity are 98.26%, 97.91%, and 98.61% respectively.