Computerized simulation of breast ultrasound images

The present article proposes a methodology for simulating breast ultrasound (BUS) images for assessing segmentation algorithms. The main characteristics of the simulated BUS images are: (i) building an echogenicity map containing common tissues found in real breasts as well as their geometries and (ii) simulating the physical process of image acquisition. The validation procedure is based on measuring the correlation between the theoretical Rayleigh probability density function (PDF) with the one obtained from simulated speckle textures. The results pointed out that the PDF derived from simulated images are strongly correlated with the theoretical PDF with a coefficient of correlation of 0.999.