Neher Peter F, Laun Frederik B, Stieltjes Bram, Maier-Hein Klaus H
Computational Disease Analysis Group, Div. Medical and Biological Informatics, German Cancer Research Center (DKFZ), Heidelberg, Germany.
Magn Reson Med. 2014 Nov;72(5):1460-70. doi: 10.1002/mrm.25045. Epub 2013 Dec 9.
Phantom-based validation of diffusion-weighted image processing techniques is an important key to innovation in the field and is widely used. Openly available and user friendly tools for the flexible generation of tailor-made datasets for the specific tasks at hand can greatly facilitate the work of researchers around the world.
We present an open-source framework, Fiberfox, that enables (1) the intuitive definition of arbitrary artificial white matter fiber tracts, (2) signal generation from those fibers by means of the most recent multi-compartment modeling techniques, and (3) simulation of the actual MR acquisition that allows for the introduction of realistic MRI-related effects into the final image.
We show that real acquisitions can be closely approximated by simulating the acquisition of the well-known FiberCup phantom. We further demonstrate the advantages of our framework by evaluating the effects of imaging artifacts and acquisition settings on the outcome of 12 tractography algorithms.
Our findings suggest that experiments on a realistic software phantom might change the conclusions drawn from earlier hardware phantom experiments. Fiberfox may find application in validating and further developing methods such as tractography, super-resolution, diffusion modeling or artifact correction.
基于体模的扩散加权图像处理技术验证是该领域创新的重要关键，且应用广泛。可公开获取且用户友好的工具，用于灵活生成针对手头特定任务的定制数据集，能够极大地便利世界各地研究人员的工作。
我们提出了一个开源框架Fiberfox，它能够实现：（1）直观定义任意人工白质纤维束；（2）通过最新的多室建模技术从这些纤维生成信号；（3）模拟实际的磁共振采集，从而能够将与磁共振成像相关的现实效应引入最终图像。
我们表明，通过模拟著名的FiberCup体模的采集，可以非常接近真实采集。我们还通过评估成像伪影和采集设置对12种纤维束成像算法结果的影响，进一步证明了我们框架的优势。
我们的研究结果表明，在逼真的软件体模上进行实验可能会改变从早期硬件体模实验得出的结论。Fiberfox可能会应用于验证和进一步开发诸如纤维束成像、超分辨率、扩散建模或伪影校正等方法。
