A Comparative Study of Unsharp Masking Filters for Enhancement of Digital Breast Tomosynthesis Images

Abstract

Microcalcification is the major focus in the early stages of breast cancer detection; thus, microcalcification detection is essential in early treatment and increases the survival rate. Since Digital Breast Tomosynthesis (DBT) images have been shown to improve the overlapping issue in mammograms, the use of this screening process is important to obtain a better perspective of microcalcifications. However, the DBT screening techniques produce blurry artifacts and noises leading this study to propose a stage for DBT image enhancement. Hence, this study proposes an enhancement method based on Non-Linear Unsharp Masking filters (NLUM). The NLUM needs a filter to complete the element of non-linear in the algorithm as Median Filter in conventional NLUM. Previously, the Hybrid Maximum Filter (H3F) and Hybrid Sigma Filter (H4F) have been proposed and demonstrated by other researchers to improve medical images, thus these filters can be adapted to the NLUM and replaced the conventional filter. Following that, the performance of the enhancement process will be assessed using Mean Square Error (MSE), Peak Signal to Noise Ratio (PSNR), and Structural Similarity Index Measure (SSIM). The results show that the H4F is the best filter to use in NLUM successfully enhances the DBT images when compared to Median Filter and H3F, with MSE, PSNR, and SSIM averages of 0.0198, 66.4000, and 0.9417, respectively.