Radiation dose reduction in computed radiography (CR) mammography through image processing

Abstract

X-ray mammography remains as the gold standard for breast cancer screening and diagnosis around the world. Even though recent developments in the field of breast imaging allowed the clinical use of systems based on digital radiography (DR) and digital breast tomosynthesis (DBT), the brazilian public health system still relies on the CR technology for breast imaging. Recent works have raised concerns about image quality and excessive radiation dose in breast images acquired by CR systems in brazilian health centers. In this scenario, the goal of this project is to propose a restoration algorithm capable of recovering the quality of breast images acquired by CR systems at low radiation doses to achieve equivalent or superior quality compared to images acquired using the current protocols. To reach this goal, we will propose a dedicated noise model to describe the degradation found in breast CR images. Furthermore, new variance stabilizing tools will be created to allow the use of state-of-the-art denoising techniques in CR breast images. An image bleding technique will be created to minimize signal smearing and blur, which naturally arises from the filtering process. The final validation of this work will be conducted on patient clinical data from the Clinics Hospital, at the School of Medicine of Ribeirão Preto, with the assistance of radiologists.