Fabry disease automatic recognition through image processing

Image processing algorithms are widely used nowadays, finding employment in the photographic field, in facial recognition, up to clinical applications. Diagnostic processes are based on digital images obtained using radiological tools capable of representing regions of the human body on 2D or 3D matrices. Image processing techniques are fundamental in order to obtain good results in terms of resolution, contrast and for a correct reconstruction of the image. Histological examinations are important for diagnosing pathologies too, analysing microscopic images of biological tissue. This is the case with Fabry disease, whose symptoms and complications can be severe (as kidney damage, heart attacks and strokes) with important effects on the quality of life of those affected. This work shows image processing algorithms as useful diagnostic support tools, capable to segment cellular tissue and to identify glycolipid bodies, typical of Fabry disease, in cellular tissue digital optical microscopic images. The obtained results are the number of intracellular accumulation structures with an estimated error lower than 10%, and the ratio of the areas of cellular tissue and glycosphingolipid accumulations, which constitute quantitative parameters for the diagnosis. Tools execution times are very low, from few seconds up to 10 minutes, with a substantial time saving compared to manual analysis. Also a mobile version of the algorithms is developed, as important application in the field of telepathology.