Evanivaldo Castro Silva Júnior,  Homero Schiabel. 

One of the most important stages in the processes ofinterpreting, classifying and diagnosing analyses forComputer Aided-Diagnosis schemes is segmentingstructures of interest. In this work is develop a model forclassification of regions of interest (ROI) possiblecontaining microcalcifications clusters for the processingof images in full mammograms. The model was based onfractal analysis, specifically fractal dimension, and isapplied on a previous model proposed by Silva Jr. et. al.as complementary stage of classifying structures likeclusters of microcalcification for the final determinationof the true-positive finds (TP), being looked for, like this,the decrease of the rate of false-positive (FP) ones.Computational tests was accomplished in order toanalyze the preliminary results. Finally, severalcomputational tests was accomplished in three groups ofimages with different compositions being the first formedby ROI's of phantoms, the second by ROI's ofmammograms and the third for full mammograms.

http://www.lbd.dcc.ufmg.br/colecoes/wvc/2009/0020.pdf

Processando consulta.