Mammographic Lesions Discrimination Based on Fractal Dimension as an Indicator

In this paper it is investigated how fractal properties can be used to characterize a mammographic lesion. The idea is suggested by the similarity between the breast tissue and a synthetically generated fractal image. Fractals are pertinent tools to describe the complexity of a shape; meanwhile, radiologists use the complexity of the lesion's contour to classify the abnormality. Tests on 30 cases mammographic lesions shows that fractal dimension of the lesion's contour is higher in cancer cases and lower in benign cases. This could be an important observation in order to classify BI-RADS 4 lesions, with no need of further examination (biopsy).