A Statistical Approach to Modelling Breast Tissue Appearance in Mammograms

In this paper, we present an approach to build models of breast tissue appearance in mammograms. Mammographic tissue is modelled based on a statistical analysis of local appearance. We investigate five strategies by using different types of local features, covering aspects of intensity, texture, and geometry. A visual dictionary is generated to summarise local tissue appearance with descriptive “words”. The global appearance of the breast is represented as an occurrence histogram over the dictionary. The resulting histogram models can be applied to breast density classification. The validity is qualitatively and quantitatively evaluated using the full MIAS database. The consensus of three experts according to the BIRADS criterion is used as the classification ground truth. We test the performance of each individual strategy and the combination of all strategies. The results indicate that our approach has potential for mammographic risk assessment.