Detection of the smallest microcalcifications for early diagnostic of breast cancer

Abstract

Even though breast cancer is a cancer with relatively easy early diagnostic and has an appropriate treatment, it has high mortality rates in Brazil. This is in part because the disease is diagnosed only in advanced stages, but also because the whole information contained in the mammograms is not used by physicians and radiologists. There are many parameters to be considered in assessing the quality of a mammogram image. Among these parameters are contrast, spatial resolution, the signal-to-noise ratio, and the efficiency of the applied dose. Even with the improvement of the quality of radiographs, many structures, such as small microcalcifications, are not always identified by radiologists in the images. To determine the lowest detectable structures in digital mammograms, we made a numerical analysis of a few digital mammography using simulators, determining the spatial and intensity resolutions, and studying the noise and its distribution. With this, we could determine the detection levels, quantifying the probability that any point is due to statistical noise or a real change in breast density. This is the first step towards early detection of microcalcifications. In our work, it was possible to detect even the smallest microcalcifications of the simulator, 0.18 mm in diameter, with false alarm probability smaller than 1/1000.