Objective: To investigate the effect of detector performance during digital breast tomography (DBT) projection data acquisition on reconstructed image quality.
Methods: With reference to the traditional detector data correction method and the specific data acquisition pattern in DBT imaging, we utilized dark field correction, light field and its gain correction for processing the projection data collected by the detector. The reconstructed images were evaluated using iterative reconstruction method based on total generalized variation (TGV).
Results: In physical breast phantom experiment, the proposed method resulted in a reduced Heel effect caused by nonuniform photon number. The reconstructed DBT images after correction showed obviously improved image quality especially in the details with a low contrast.
Conclusion: The dark field correction, light field and its gain correction process for DBT image reconstruction can improve the image quality.
目的: 研究数字乳腺层析成像（DBT）投影数据采集过程中的探测器性能因素对重建图像质量的影响。
方法: 参照传统探测器的投影数据校正方式，结合DBT成像过程特有的数据采集模式，对探测器采集的投影数据进行暗场校正、光场及其增益校正等。而后，利用基于全广义变分正则化项的PWLS迭代重建算法进行DBT图像重建。
结果: 乳腺仿真物理体模数据的实验结果表明，相对于未校正的投影数据，本文校正方法对由光子数分布不均造成的投影数据足跟效应进行了有效抑制。重建图像结果显示，校正后投影数据的重建图像中低对比度结构的显像得到提升。
结论: 投影数据暗场校正、光场及其增益校正过程用于DBT图像重建可提高图像质量。