Improvement of Image Characteristics in Digital Breast Tomosynthesis by Incorporating a Compressed-sensing (CS) Deblurring Framework: Simulation Study

摘要

In this work, we considered a compressed-sensing (CS)-based framework with the total-variation regularization penalty for image deblurring of high accuracy in digital breast tomosynthesis (DBT). We implemented the proposed algorithm and performed a systematic simulation to demonstrate its viability for improving the image characteristics in DBT. In the simulation, blurred noisy projection images of a 3D breast phantom were generated by convolving their original (or exact) version by a 2D Gaussian blur kernel (σ = 2 in pixel unit, kernel size = 11 × 11), followed by adding Gaussian noise (m = 0, σ~2 = 0.05), and deblurred by using the proposed algorithm before performing DBT reconstruction.