Nonlinear image analysis for fuzzy classification of breast cancer

Nonlinear image processing proves to be a powerful tool for segmenting images preserving only the interesting regions; those are the cells' nucleus (CN) in this application.

Abstract

Nonlinear image processing proves to be a powerful tool for segmenting images preserving only the interesting regions; those are the cells' nucleus (CN) in this application. All CN are preserved and identified even if they are in a cluster, while the rest of the image is considered to be part of the background. The morphological segmentation makes it possible to calculate outstanding features that could not be obtained by simple observation. These are passed to a fuzzy classifier which decides the probability of the biopsy to belong to a high or low cancer level. Obtaining these measures by human observation is a hard, and nonprecise task. The extracted features will make it possible to reach a parametric classification that is more efficient than the subjective classification made by human observation.