Microcalcifications detection ushig wavelets and self-organized methods by nowcontextal pixels classification
José Miguel Barrón-Adame, Antonio Vega-Corona · World Automation Congress · 2004
We present an image segmentation based in pattern recognition for microcalifications (μCs)dectections. A feature Vector Set (FVS) that represents the microcalifications (μCs) is selected in order to train a classifier. Wavelet (WT) and Self Organized Map (SOM) have been combined in segementation process. Regions of Interest (ROIs) have been previously diagnosed and analyzed in order to extract a multidimensional FVS. Each pixel is represented by a mulitdimensional vector. A SOM method to chaster and lable the FVS in order to identify (μCs) pixwles have been applied. We give appropriate results segmenting the (μCs) from our images database.