Directional Total Variation Regularized Breast Tomosynthesis Image Reconstruction

Abstract

Digital Breast Tomosynthesis (DBT) is an effective breast imaging modality that provides 3D cross-sectional images of the breast. However, reconstructed images from the DBT have some distortions due to the limited scan angle and number of projections. In this study, a 3D Directional Total Variation (DTV) algorithm was developed by changing partial gradient terms. Four neighboring pixels along the y axis and 4 partial gradient terms were used in DTV. The effect of strengthening the partial gradient terms according to the DBT geometry were investigated. Contrast to noise ratio (CNR), image intensity profiles and visual assessment were used to assess the performance of the proposed method. It can be clearly seen that proposed DTV algorithm successful to suppress background noise according to visual assessment and image intensity profiles. Besides, it can be said that contrast resolution of the masses and micro calcifications can be improved according to CNR measurements by using DTV algorithm. The obtained results clearly show that the proposed DTV algorithm is effective in decreasing the artefacts in reconstructed DBT images.