Stochastic model and probabilistic decision-based classifier formass detection in digital mammography

摘要

We have developed a combined method utilizing morphologicaloperations, a finite generalized Gaussian mixture (FGGM) modeling, and acontextual Bayesian relaxation labeling technique (CBRL) to enhance andextract suspicious masses. A feature space is constructed based onmultiple feature extraction from the regions of interest (ROIs).Finally, a multi-modular probabilistic decision-based classifier isemployed to distinguish true masses from non-masses