Automatic Fibrosis Quantification By Using a k-NN Classificator

This work presents an automatic algorithm to measure fibrosis in muscle sections of mdx mice, a mutant species used as a model of the Duchenne dystrophy. The algorithm described herein automatically segments three different tissues: Muscle cell tissue (MT), Pure collagen fiber deposit (CD) and cellular infiltrates surrounded by loose collagen deposit (CI), by using a statistical classifier based on the k-Nearest Neighbour (k-NN) decision rule in the RGB color space. The algorithm is trained by selecting a number of correctly classified pixels from each class. The k-NN rule classifies other pixels in the class that is most represented among the k nearest training samples in the RGB space, which is efficiently implemented with a fast k-distance transform algorithm. All extracted areas are quantified in absolute (micrometer squared) and relative (%) values. For validation of this method, the different tissues were manually segmented and their qualifications statistically compared with those obtained automatically. Statistical analysis showed interoperator variability in manual segmentation. Automatic qualifications of the same areas did not differ significantly from their mean manual evaluations. In conclusion, this method produce fast, reliable and reproducible results. Papers from the 23rd Annual International Conference of the IEEE Engineering in Medicine and Biology Society, October 25-28, 2001, held in Istanbul, Turkey. See also ADM001351 for entire conference on cd-rom. The original document contains color images.