Diffusion Tensor Imaging (DTI) is an MRI technique that measures water displacement to infer fiber architecture and integrity based on the assumption that water diffusion within intact fibers will be anisotropic (Bucci et al., 2013). This in vivo technique has implications in the clinical and research world for diagnosis and treatment of neurological disorders, however there is no gold standard for DTI that can be used to ensure reliability across scanners, parameters, and processing methods that simulates the brain’s complexity. In this work a DTI phantom containing complex geometries and anisotropic microstructure to mimic fiber bundles is presented. It is the end goal of this work to provide a commercially available DTI phantom to serve as a quality assurance tool in order to improve the application of diffusion imaging.

The DTI phantom (Fig. 1) consists of ten individual fiber bundles distributed in different orientations (linear, crossing, interweaving) to best test and confirm proper DTI processing. Each fiber bundle contains 4.4 million fibers within a flexible casing with an overall diameter of 6 mm. The phantom was imaged using a 2D EPI DTI sequence: 3T Siemens Skyra; 32 channel head coil; TE/TR = 118.0/20300 msec; Matrix = 160x160; FOV = 20.8 cm, 110 slices; slice thickness = 1.3 mm (no gap); grappa = 2; partial Fourier off; 20 diffusion directions; b = 1000 mm2/sec, and a 2D T2-weighted sequence: TE/TR = 180.0/15900 msec; Matrix = 320x320; 110 slices; slice thickness = 1.0 mm (no gap). Analysis: Data processing was performed using BrightMatter™ Plan (Synaptive Medical, 2015), a data processing tool that co-registers EPI DTI data to a T1-weighted, T2-weighted, or CT image and performs whole brain 3D tractography. In this work the DTI dataset was co-registered to the acquired T2-weighted images. FA values for two fiber bundles (Head – Foot [H-F] and Anterior – Posterior [A-P] directions) were measured along the length of the bundles in order to qualify consistency in diffusion. FA map slices were analyzed in ImageJ, an open source image processing and analysis tool, where data points, comprising an average FA of 27 voxels (i.e. 3x3x3 voxel cube), were taken along the length of the fiber bundle.An example slice of the T2-weighted dataset along with the co-registered DTI dataset are shown in Figure 2. As one can see from the figure, there is good agreement between fiber location and their generated tractography. FA values averaging 0.51 were within +/- 10% of the overall FA average within both fibers demonstrating consistent diffusion throughout the fiber bundles (Fig. 3). The sharp drop-off in FA value in the plot corresponds to each end of the fiber, indicating a low background FA value for the matrix. Complete phantom tractography was generated as shown in Fig. 4. Fiber tracts are present along all fiber bundles, which is especially visible for the interweaving fibers.This novel DTI phantom has demonstrated anisotropic diffusion using highly organized and flexible fiber bundles in complex orientations. This phantom can be used for potential validation of DTI processing, allowing for the expanded application of this useful in vivo imaging technique in the understanding of disease processes. With the rise in large, multi-site and longitudinal research studies, and the increased use of diffusion imaging in a clinical setting, the importance of standardization has never been more important.