An Improvement for Human Intestinal Parasites Detection Methodology using $k$-Means and Fast $k$-Means Clustering

Helminthiases disease is one of the diseases that leads to a significant impairment in mental and physical development in children and adolescents. It entail risks if not treated properly. Hence, it is very important to have a system that can diagnose the helminthiases disease efficiently. Hence, the goal of this research is to analyse the segmentation performance for the unsupervised colour image segmentation of human intestinal parasites based on helminth between standard
$k$
-means (KM) clustering algorithm and Fast
$k$
-means (FKM) clustering algorithm through modified global contrast stretching (MGCS) enhancement technique and colour conversion based on saturation (S) component. A total of 200 helminth parasite images have been analysed with the technique proposed, which consists of 50 images for each species of Ascaris Lumbricoides Ova (ALO), Entrobious Vermicularis Ova (EVO), Hookworm Ova (HWO), and Trichuris Trichiura Ova (TTO). Both KM and FKM have succeeded in segmenting the helminth parasites with almost similar results. However, FKM has surpassed KM in terms of processing time. FKM takes 694 seconds to finish the segmentation process, as compared to KM which takes 1112 seconds. Overall, FKM clustering has successfully segment the helminth parasites with less processing time and good segmentation performance of 99.33% for accuracy, 93.36% for sensitivity of and 99.43% for specificity. The results obtained shows that the approach technique is fast and suitable to segment the helminth parasites.