Análise da qualidade das imagens mamográficas a partir de transformações de equalização de histograma e de aplicação de métricas de qualidade

Abstract

Breast cancer is characterized by the disordered growth of cells with abnormal characteristics, multiplying and causing the formation of nodules. Currently, breast cancer accounts for 29.7% of all cancers affecting Brazilian women the most. Therefore, mammography is of utmost importance as the primary method for disease screening. Through mammography, early diagnosis becomes possible, increasing the likelihood of less aggressive treatment and higher rates of satisfactory outcomes. To use mammography as a screening tool, it is essential for the images to have good quality for the differentiation of breast structures. Image processing tools based on histogram transformations can be used to enhance breast structures. One way to assess image quality is by applying image quality metrics. In this study, histogram equalization techniques were applied to phantom and patient breast images to evaluate the best processing parameters. Image quality was examined using metrics such as contrast, signal-to-noise ratio (SNR), and contrast-to-noise ratio (CNR). Ultimately, the obtained data were compiled into tables and compared. The results showed that the image with higher CNR corresponds to better visualization.