A robust segmentation for malaria parasite detection of thick blood smear microscopic images

Parasite Detection on thick blood smears is a critical step in Malaria diagnosis. Most of the thick blood smear microscopic images have the following characteristics: high noise, a similar intensity between background and foreground, and the presence of artifacts. This situation makes the detection process becomes complicated. In this paper, we proposed a robust segmentation technique for malaria parasite detection of microscopic images obtained from various endemic places in Indonesia. The proposed method includes pre-processing, blood component segmentation using intensity slicing and morphological operation, blood component classification utilising rule based on properties of parasite candidates, and parasite candidate formation. The performance was evaluated on 30 thick blood smear microscopic images. The experimental results showed that the proposed segmentation method was robust to the different condition of image and histogram. It reduced the misclassification error and relative foreground error by 2.6% and 45.5%, respectively. Properties addition to blood component classification increased the system precision. Average of precision, recall, and F-measure of the proposed method were all 86%. It is proven that the proposed method is appropriate to be used for malaria parasites detection.