Development of a breast statistical phantom prototype for mammographic image qualification

Abstract

Tissue simulators (or phantoms) are often used in radiology diagnostic services to evaluate images quality. This work deals with the development of a prototype of breast statistic phantom that allows to change the image density - by varying the phantom thickness - as well as to distribute randomly the structures. The purpose is obtaining an image so that tumor masses, microcalcifications and fibers can be detected by visual inspection similarly to a mammography exam. The main advantage of this prototype is that it can avoid (or, at least, minimize) the subjectivity usually associated to the use of phantoms based on fixed structure distribution. Tests have shown quite similarity between regions of interest extracted from this phantom images and extracted from actual mammograms. In addition, it is useful to improve quality assurance tests sensitivity, since the random variation of inserted structures minimizes the probability of systematic errors in the images evaluation. © 2009 Springer-Verlag.