Article Preview
Top1. Introduction
The human intestinal worm is one of the most common infectious diseases in human, which is predominant in developing countries possing economic threat to public health (Haque, 2007; Harhay, Horton, & Olliaro, 2010). The early detection of these worms among children is vital for the diagnostic purpose; nonetheless, most health facilities are less equipped with a sophisticated tool and expect skills to manage their outbreak (Garcia, et al., 2017). Conventionally, medical excepts in this field use visual analysis based on experience to carry out their diagnostic task (Crowley, Naus, Stewart, & Friedman, 2003). Though this approach help with the diagnostic process, it is time-consuming and extremely tedious to the extent of possing other health-related hazards (Momcilovic, Cantacessi, Arsic-Arsenijevic, Otranto, & Tasic-Otasevic, 2019; Moody & Chiodini, 2001; Tavares, et al., 2011). With the advent of computational tools, several studies are being carried to search for a sustainable and cost-effective solution to the diagnostic problem (Weatherall, Greenwood, Chee, & Wasi, 2006). In the field of parasitology, concepts of pattern recognitions are employed to diagnose medically relevant parasites (Daugschies, Imarom, & Bollwahn, 1999; Sommer, 1996). The concept takes three forms, thus, image pre-processing, feature extractions (Gupta & Shanker, 2021; Aggarwal, Mittal, & Bali, 2021), and classification (Lim, 1990; Jahne, 2005). Techniques adapting these fundamentals concepts in the diagnosis of intestinal worm depending on the characteristic of the dataset are artificial neural networks (Yang, Park, Kim, Choi, & Chai, 2001; Goundar, Prakash, Sadal, & Bhardwaj, 2020), adaptive network-based fuzzy inference system (Dogantekin, Yilmaza, Dogantekin, Avcic, & Sengurc, 2008), MultiClass Support Vector Machine classifier (Avci & Varol, 2009; Panda, 2019; Goundar, Sam; Bhardwaj, Akashdeep, 2021), active contours (Gupta, Bharadwaj, & Rastogi, 2021) and Bayesian classification system (Castañon, Fraga, Fernandez, Gruber, & Costa, 2007). Though computational tools are proven to be usual, the story is slightly different with the African context given the type of dataset captured from the community. It is well observed that, these types of parasites are common in neglected communities where lifestyle and access to health care is poor. Despite these conditions, quite a good number of natives have access to smartphones but with poor resolution which can be leveraged on to enhance health care. The problem then arises with how to use this low resolution images and still provide to some level of precision good health care. This study therefore seeks to propose a framework for the automatic diagnosis of human intestinal worms captured with a smartphone which poses unique challenges.