Validation of Presurgical Simulation of White Matter Damage Using Diffusion Tensor Imaging

Abstract

The understanding of white matter (WM) was revolutionized by the emergence of tractography based on diffusion tensor imaging (DTI). Currently, DTI simulations are implemented in preoperative planning to optimize surgical approaches. The reliability of these simulations has been questioned and investigated seeking for correlation between neurological performance and anomalies in DTI parameters. However, the ability of preoperative WM simulations to predict a surgical injury has not been thoroughly evaluated. Our objective was to assess the reliability of preoperatively simulated WM injuries for conventional neurosurgical procedures. WM surgical damage was preoperatively simulated by creating a 3-dimensional volume representing the endoscope or the surgical trajectory. This volume was used as an additional region of interest in the fascicle reconstruction to be subtracted from the original fascicle. Simulated, injured fascicles were compared in terms of the number of fibers and volume to those created from postoperative DTI studies. Reliability was assimilated into the correlation between the simulation and the postoperative reconstruction; evaluated using the intraclass correlation coefficient or Lin's Concordance correlation coefficient (CCC), and represented on Bland-Altman plots. The preoperative and postoperative DTI studies of 30 patients undergoing various neurosurgical approaches were processed. The correlation between simulated injuries and postoperative studies was high in terms of fibers (Concordance correlation coefficient= Rho.C=0.989 [95% confidence interval= 0.979-0.995]) and volume (intraclass correlation coefficient=0.95 [95% CI= 0.89-0.97]). Bland-Altman plots demonstrated that the great majority of cases fell within the mean±2 Standard deviations. Presurgical simulation of WM fascicles based on DTI is consistent with postoperative DTI studies. These findings require further validation by neurophysiological and clinical correlation.