There may be value in combining multiple image features in an automated algorithm for mammographic density. Using a Bayesian classifier , the ranges in value of the fractal and regional skewness measures described earlier that corresponded to the classification of SCC by radiologist were identified. The data were then partitioned on the basis of these ranges to automatically assign a classification of mammo-graphic density. It was found that a two-dimensional, regional skewness-fractal classifier (S-FC) provided better correlation with subjective mammographic density assessment by SCC than was obtained with either the regional skewness (SC) or fractal classifier (FC) alone .
Was this article helpful?