We have studied an animal model of acute local inflammation in muscle induced by Aspergillus fumigatus by using magnetic resonance imaging (MRI) and magnetic resonance spectroscopy (MRS). We have compared our data to those found using histopathology and segmentation maps obtained by the mathematical processing of three-dimensional T2-weighted MRI data via a neural network. The MRI patterns agreed satisfactorily with the clinical and biological evidence of the phases of acute local infection and its evolution towards chronicity. The MRS results show a statistically significant increase in inorganic phosphate and a significant decrease in phosphocreatine levels in the inflamed region. Image segmentation made with a self-organizing, neural-network map yielded a set of ordered representatives that remained constant for all animals during the inflammatory process, allowing a non-invasive, three-dimensional identification and quantification of the inflamed infected regions by MRI.
Copyright 2002 John Wiley & Sons, Ltd.