Kim Hyeongseok, Lee Taewon, Hong Joonpyo, Sabir Sohail, Lee Jung-Ryun, Choi Young Wook, Kim Hak Hee, Chae Eun Young, Cho Seungryong
Department of Nuclear and Quantum Engineering, Korea Advanced Institute of Science and Technology (KAIST), Daejeon, 34141, Korea.
School of Electrical and Electronics Engineering, Chung-Ang University, Seoul, 06974, South Korea.
Med Phys. 2017 Feb;44(2):417-425. doi: 10.1002/mp.12078. Epub 2017 Feb 2.
Nonlinear pre-reconstruction processing of the projection data in computed tomography (CT) where accurate recovery of the CT numbers is important for diagnosis is usually discouraged, for such a processing would violate the physics of image formation in CT. However, one can devise a pre-processing step to enhance detectability of lesions in digital breast tomosynthesis (DBT) where accurate recovery of the CT numbers is fundamentally impossible due to the incompleteness of the scanned data. Since the detection of lesions such as micro-calcifications and mass in breasts is the purpose of using DBT, it is justified that a technique producing higher detectability of lesions is a virtue.
A histogram modification technique was developed in the projection data domain. Histogram of raw projection data was first divided into two parts: One for the breast projection data and the other for background. Background pixel values were set to a single value that represents the boundary between breast and background. After that, both histogram parts were shifted by an appropriate amount of offset and the histogram-modified projection data were log-transformed. Filtered-backprojection (FBP) algorithm was used for image reconstruction of DBT. To evaluate performance of the proposed method, we computed the detectability index for the reconstructed images from clinically acquired data.
Typical breast border enhancement artifacts were greatly suppressed and the detectability of calcifications and masses was increased by use of the proposed method. Compared to a global threshold-based post-reconstruction processing technique, the proposed method produced images of higher contrast without invoking additional image artifacts.
In this work, we report a novel pre-processing technique that improves detectability of lesions in DBT and has potential advantages over the global threshold-based post-reconstruction processing technique. The proposed method not only increased the lesion detectability but also reduced typical image artifacts pronounced in conventional FBP-based DBT.
在计算机断层扫描（CT）中，投影数据的非线性预重建处理通常不被提倡，因为准确恢复CT值对诊断很重要，而这种处理会违反CT成像的物理原理。然而，在数字乳腺断层合成（DBT）中，可以设计一个预处理步骤来提高病变的可检测性，由于扫描数据的不完整性，在DBT中根本不可能准确恢复CT值。因为检测乳房中的微钙化和肿块等病变是使用DBT的目的，所以一种能提高病变可检测性的技术是有价值的。
在投影数据域中开发了一种直方图修改技术。原始投影数据的直方图首先被分为两部分：一部分用于乳房投影数据，另一部分用于背景。背景像素值被设置为一个代表乳房和背景之间边界的单一值。之后，两个直方图部分都偏移适当的量，并且对经过直方图修改的投影数据进行对数变换。使用滤波反投影（FBP）算法进行DBT的图像重建。为了评估所提出方法的性能，我们从临床获取的数据中计算了重建图像的可检测性指数。
使用所提出的方法，典型的乳房边界增强伪影得到了极大抑制，钙化和肿块的可检测性增加。与基于全局阈值的重建后处理技术相比，所提出的方法在不产生额外图像伪影的情况下产生了对比度更高的图像。
在这项工作中，我们报告了一种新颖的预处理技术，该技术提高了DBT中病变的可检测性，并且相对于基于全局阈值的重建后处理技术具有潜在优势。所提出的方法不仅提高了病变的可检测性，还减少了传统基于FBP的DBT中明显的典型图像伪影。
