Retrospective Comparative Analysis of Diffusion Tensor and Constrained Spherical Deconvolution Tractography Models in a Clinical Neurosurgical Navigation Platform

抄録

Tractography is a technique for visualizing the trajectories of white matter pathways. The most widely used model, the diffusion tensor imaging, assumes a single fiber orientation within each voxel. In contrast, the constrained spherical deconvolution model enables the estimation of multiple fiber orientations. Recently, constrained spherical deconvolution was added to a widely used neurosurgical navigation platform alongside the diffusion tensor imaging model, allowing both approaches to be applied within the same clinical environment. In this study, we compared the volumes of white matter tracts reconstructed using these 2 models. Magnetic resonance imaging data from 37 patients were used to reconstruct 5 white matter tracts-the corpus callosum, the bilateral inferior fronto-occipital fasciculi, and the bilateral corticospinal tracts. Volumetric and visual analyses were conducted to assess differences in tract reconstruction between constrained spherical deconvolution and diffusion tensor imaging. Across all bundles, constrained spherical deconvolution consistently yielded significantly greater tract volumes, with a mean percentage volume difference of 158.6% (p < 0.001), most prominently in the inferior fronto-occipital fasciculi. Visual assessments favored constrained spherical deconvolution-based reconstructions in approximately 80% of the cases, and volumetric and visual measures demonstrated strong concordance (Spearman's ρ = −0.766, p < 0.001). In contrast, differences between constrained spherical deconvolution and diffusion tensor imaging were less pronounced in the corticospinal tracts. Lowering the fractional anisotropy threshold in diffusion tensor imaging did not improve reconstruction quality and produced anatomically implausible streamlines. Tract type was the primary determinant of inter-method differences (p < 0.001). Integration of constrained spherical deconvolution into clinical navigation can enhance visualization of white matter organization and may improve preoperative planning and risk evaluation in neurosurgical practice.