Bruun, Johan Musaeus3; Kapel, Christian M. O.3; Carstensen, Jens Michael4
1 Department of Informatics and Mathematical Modeling, Technical University of Denmark2 Image Analysis and Computer Graphics, Department of Informatics and Mathematical Modeling, Technical University of Denmark3 University of Copenhagen4 Department of Applied Mathematics and Computer Science, Technical University of Denmark
Eggs from the small, intestinal pig whipworm Trichuris suis constitute the active pharmaceutical ingredient in a novel type of medicine for chronic autoimmune diseases like Crohn's disease. The pharmaceutical potency of such an egg suspension can be assessed by microscopic inspection, as only eggs containing a viable, infective larva provoke the wanted immune reaction. Thus, a precise and objective estimate of the concentration of infective eggs is crucial for dosing the new medicine. In this paper, a vision-based method for detecting and classifying T. suis parasite eggs is described. The detection is based on matched filters and the classification is done using linear and quadratic discriminant analysis on a set of biologically inspired features, including the autocorrelation-based longitudinal anisotropy and the mean scattering intensity under dark field illumination. Despite the presence of impurities and overlapping eggs, the proposed method achieves cross-validated classification rates around 93%.
2012 9th Ieee International Symposium on Biomedical Imaging (isbi), 2012, p. 1627-1630
Main Research Area:
IEEE 9th International Symposium on Biomedical Imaging (ISBI 2012)