Comparison of Automatic and Semiautomatic Approach for the Posterior Nipple Line Calculation

Early diagnosis of breast cancer is an important prerequisite for a better prognosis and improved life expectancy. An accurate diagnosis also depends on the quality of the mammographic image which depends on the correct positioning of the breast during the radiological examination. From this perspective, automatic tools supporting technologists and radiologists and guiding them in evaluating breast positioning from mammographies can be very useful especially if they can be based on quantitative parameters. However, there is no clear consensus concerning the evaluation of breast positioning from mammographies: several scores have been proposed and commercial software exists with their implementations. In general breast positioning scores take into account a small number of parameters on which a consensus has emerged. In this study, we compared an automatic and semi-automatic method for calculating the posterior nipple line (PNL), which represents an important quantitative parameter useful for guiding positioning. From the comparison of the automatic method with the semi-automatic method, an excellent agreement emerged in the case of the CC view (ICC = 0.92) and to be improved in the case of the MLO view (ICC = 0.73). This study therefore offers encouraging results for the use of the PNL as a quantitative parameter to guide correct breast positioning.