Texture Analysis for Nematode Genera Classification

Nematodes are the most diverse animals and the most numerous multi-cellular organisms on earth. They exist in almost every ecological niche and vary in size from microscopic round worms to large classes such as Cestoda, which can grow to over 30 meters. Not only are nematodes greatly abundant, but they have significant impacts on human life. Nearly two billion people worldwide are infected by a single nematode genus and billions of dollars of yearly agricultural damage can be attributed to just a handful of nematode species. Because these affects originate from a small subset of the phylum, classification of nematodes is necessary to assist disease prevention and eradication. Genetically based pesticides rely on accurate identification and monitoring of nematode populations. Classification of a single nematode by field experts requires several days, an automated system would improve the accuracy, throughput, and efficiency of this process. This paper focuses on classifying microscopic nematodes. This presents additional challenges in classification due to the size, lack of visual features, and great abundance of the subjects. Microscopic nematodes have simple, transparent bodies. A texture based analysis is conducted towards classification of a sample's genus due to their structure. Nematode images are attained using video capture and editing microscopy (VCE) techniques, resulting in a series of multifocal images per nematode sample. By applying texture based methods on multifocal image stacks of nematodes, an accurate image based classifier is constructed.