Quantitative evaluation of a mammographic software phantom generator

This paper presents an evaluation of an algorithm for generation of 3D breast software phantoms for mammography. For this purpose, thirty breast phantoms of different shape, size and composition and with no pathology included were generated. Mammographic images of these breast models were produced by simulating mammographic image acquisition. Quantitative texture analysis included calculation of fractal dimension, skewness, kurtosis, and power spectral analysis of simulated and real mammograms as well as comparison with similar calculations published in the literature. Calculated parameters from simulated images matched closely those calculated from real mammograms and published data. As such the mammographic software phantom generator can be used to create breast models suitable for carrying out feasibility studies in mammography.