Neural Network Based Classification of Cell Images via Estimation of Fractal Dimensions
This paper1 presents a system for classifying cells in human resistance arteries via estimating and analysing fractal dimensions of normal and abnormal cell images. The use of fractal features helps characterise and differentiate between categories of cell images. The classification task is implemented using a multi-layer feedforward neural network, which maps estimated fractal feature patterns onto their underlying cell index. This system has been applied to a large database of images (which were taken from proximal and distal areas of subcutaneous resistance arteries, of patients suffering from critical limb ischaemia, by means of laser scanning confocal microscopy), with the overall classification rate reaching 90.5%.
Unable to display preview. Download preview PDF.
- Bishop CM. Neural networks for pattern recognition. Oxford University Press, Oxford, 1995Google Scholar
- Rumelhant DE, Hinton GE, Williams RJ. Learning internal representations by error propagating. In: Parallel Distributed Processing. Rumelhant E, McClelland JL. (Eds.), MIT Press, 1986Google Scholar
- Daly CJ, Gordon JF, McGrath JC. The use of fluorescent nuclear dyes for the study of blood vessel structure and function: novel applications of existing techniques. J. Cardiovascular Research 1992; 29:41–48Google Scholar
- Duda RO, Hart PE. Pattern classification and scene analysis. Jone Wiley and Sons, New York, 1993Google Scholar