Quantitative Validation of White Matter Fiber Tractography by use of an Anatomically Realistic Synthetic Diffusion Tensor Phantom

S. Delputte, E. Fieremans, Y. Dedeene, Y. D'Asseler, R. Achten, I. Lemahieu, R. Van de Walle Medisip, Department of Electronics and Information Systems, Ghent University, Ghent, Belgium, Department of Radiotherapy, Ghent University Hospital, Ghent, Belgium, Department of Neuroradiology, Ghent University Hospital, Ghent, Belgium Introduction Diffusion Tensor (DT) Imaging can disclose the 3D organization of fibrous tissue. Although Diffusion Tensor Tractography (DTT) is a very promising non-invasive method for reconstructing the white matter axonal pathways, its current clinical use is limited due to the lack of a golden standard to validate this technique. Therefore we propose a framework to construct a noise-free synthetic diffusion tensor dataset, originating from a known fiber distribution that resembles the true white matter anatomy as accurate as possible (including slender as well as thicker, crossing and “kissing” fasciculi with realistic curvature and torsion parameters). These ground-truth fibers are obtained by performing Density Regularized Fiber Tracking (DRFT, [1]) on in vivo data. The environmental architectural information proffered by the DRFT results, together with the pointwise measured FA (Fractional Anisotropy) and Mean-ADC (Apparent Diffusion Coefficient), were modeled into an anatomically realistic noise-free synthetic DT dataset, which eventually was used to optimize and quantitatively validate several existing tractography algorithms.