Siemens Healthcare GmbH, 91301 Forchheim, Germany. Erlangen Centre for Astroparticle Physics, FAU Erlangen-Nuremberg, 91058 Erlangen, Germany.
Phys Med Biol. 2018 Jul 3;63(13):135018. doi: 10.1088/1361-6560/aacb07.
Talbot-Lau x-ray imaging (TLXI) is an innovative and promising imaging technique providing information about the x-ray attenuation, scattering, and refraction features of objects. However, the method is susceptible to vibrations and system component imprecisions, which are inevitable in clinical and industrial practice. Those influences provoke grating displacements and hence errors in the acquired raw data, which cause moiré artifacts in the reconstructed images. We developed an enhanced reconstruction algorithm capable of compensating these errors by adjusting the grating positions and thus suppressing the occurrence of moiré artifacts. The algorithm has been developed with regard to a future application in medical practice. The capability of the algorithm is demonstrated on a medical data set of a human hand (post-mortem) acquired under clinical conditions using a pre-clinical TXLI prototype. It is shown that the algorithm reliably suppresses moiré artifacts, preserves image contrast, does not blur anatomical structures or prevent quantitative imaging, and is executable on low-dose data sets. In addition, the algorithm runs autonomously without the need of interaction or rework of the final results. In conclusion, the proposed reconstruction algorithm facilitates the use of TLXI in clinical practice and allows the exploitation of the method's full diagnostic potential in future medical applications.
塔波特-劳 X 射线成像（TLXI）是一种创新且有前途的成像技术，可提供有关物体的 X 射线衰减、散射和折射特性的信息。然而，该方法容易受到振动和系统组件不精确的影响，这在临床和工业实践中是不可避免的。这些影响会引起光栅位移，从而导致采集到的原始数据出现误差，这会导致重建图像中出现莫尔伪影。我们开发了一种增强型重建算法，能够通过调整光栅位置来补偿这些误差，从而抑制莫尔伪影的发生。该算法是针对未来在医学实践中的应用而开发的。该算法的功能已在使用临床前 TLXI 原型在临床条件下获取的人体手部（死后）的医学数据集上进行了验证。结果表明，该算法能够可靠地抑制莫尔伪影，保留图像对比度，不会使解剖结构模糊或阻止定量成像，并且可以在低剂量数据集上执行。此外，该算法可以自主运行，无需对最终结果进行交互或返工。总之，所提出的重建算法促进了 TLXI 在临床实践中的应用，并允许在未来的医学应用中充分利用该方法的诊断潜力。
