COMPUTING THE SCATTER COMPONENT OF MAMMOGRAPHIC IMAGES

This paper builds upon a technical report in which we proposed a model of the mammographic imaging process for which scattered radiation is a kev degrading factor. In this paper, we propose a way of estimating the scatter component of the signal at any pixel within a mammographic image, and we use this estimate for model-based image enhancement. Th e first step is to extend our previous model to divide breast tissue i nto ''interesting'' (fibrous/glandular/cancerous) tissue and fat. The scatter model is then based on the idea that the amount of scattered r adiation reaching a point is related to the energy imparted to the sur rounding neighbourhood. This complex relationship is approximated usin g published empirical data, and it varies with the size of the breast being imaged. The approximation is further complicated by needing to t ake account of extra-focal radiation and breast edge effects. The appr oximation takes the form of a weighting mask which is convolved with t he total signal (primary and scatter) to give a value which is input t o a ''scatter function'', approximated using three reference cases, an d which returns a scatter estimate. Given a scatter estimate, the more important primary component can be calculated and used to create an i mage recognizable by a radiologist. The images resulting from this pro cess are clearly enhanced, and model verification tests based on an es timate of the thickness of interesting tissue present proved to be ver y successful. A good scatter model opens the way for further processin g to remove the effects of other degrading factors, such as beam harde ning.