Edge detection of intestinal parasites in stool microscopic images using multi-scale wavelet transform

Biomedical image processing is experiencing a significant progress with many applications. However, automatic recognition of microscopic pathogens from their images remains a challenge that will allow clinical laboratories to increase both the speed of tests and the reliability of diagnoses. We present an algorithm for edge detection of parasites in microscopic images of stools, using the multi-scale wavelet transform. This method is an evolution of the Canny–Mallat detector which gives the possibility to vary the frequency of the analysis in order to find the outlines of the most significant edges. The various contours obtained are chained across the scales from the coarsest to the finest. Using this algorithm, we were able to correctly represent the contours of the features of parasites found in microscopic images. The results obtained were compared with those produced by classical edge detectors on the same images. It comes out from both subjective and objective quantitative performances evaluation that our detector, better than all others, can clearly mark the outlines of the structures of the pathogen on an image of stools.