It contains the codes for the paper "Photo response non-uniformity correction for digital mammography systems", submitted to the CBEB 2024 conference.
Disclaimer: For education purposes only.
Denoising is a crucial tool for improving image quality and reducing radiation exposure in medical imaging. The effectiveness of denoising techniques relies on accurate noise modeling and understanding equipment characteristics. In previous studies, our research group developed a denoising pipeline that considers the Poisson-Gaussian noise model, including both signal and spatial-dependent quantum noise and electronic noise. However, this work did not address another noise source known as Photo Response Non-Uniformity (PRNU), or structural noise. In this work, we propose a correction for PRNU in Full-Field Digital Mammography (FFDM) images and analyze the impact of this correction on the denoising pipeline for mammography images. Our results indicate that correcting for structural noise enhances the effectiveness of our approach in removing noise from mammography images in terms of signal loss, especially in regions characterized by higher concentrations of high-gray-level pixels.
This work was supported by the São Paulo Research Foundation (FAPESP) grant #2021/12673-6 and by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - Brasil (CAPES) - Finance Code 001.