Plan of processing with filtering of noise related to contrast enhancement techniques applied to images of dense breasts

Abstract

Dense breast images represent a continuous challenge to the stages of imaging processing involved in CAD schemes ("Computer Aided Diagnosis") mainly due to their low contrast and its implications to the human and computational visualization. The noise present in radiological images, specifically mammograms, represents and impediment in the diagnostic process. The main purpose of this work is to use a selective non-linear model of contrast enhancement with a density continuous-automatic classification regarding digital mammographic images associated with a model of noise removal and the preservation of edges. Two different databases were used - composed by regions of interest (ROIs) obtained from dense breasts images and ROIs of phantoms. Initial tests performed have pointed out a remarkable improvement in the dense breast images contrast apart from a significant reduction in the number of false positives. © Springer-Verlag Berlin Heidelberg 2007.