Comparison of probabilistic and deterministic fiber tracking of cranial nerves

• DOI: 10.3171/2016.8.jns16363
• Publication Date: 2016-12-16T15:54:04Z
• Authors: Amir Zolal; Stephan B. Sobottka; Dino Podlesek; Jennifer Linn; Bernhard Rieger; Tareq A. Juratli; Gabriele Schackert; Hagen H. Kitzler
• Keywords: Cranial nerves

OBJECTIVE The depiction of cranial nerves (CNs) using diffusion tensor imaging (DTI) is great interest in skull base tumor surgery and DTI used with deterministic tracking methods has been reported previously. However, there are still no good usable for the elimination noise from resulting depictions. authors have hypothesized that probabilistic could lead to more accurate results, because it efficiently extracts information underlying data. Moreover, adapted a previously described technique gradual threshold increases tracking. To evaluate utility this new approach, comparison provided work between increase method CNs. METHODS Both were depict CNs II, III, V, VII+VIII bundle. Depiction 240 was attempted each above 30 healthy subjects, which obtained 2 public databases: Kirby repository (KR) Human Connectome Project (HCP). Elimination erroneous fibers by gradually increasing respective thresholds (fractional anisotropy [FA] index connectivity [PICo]). results compared predefined ground truth images based on corresponding anatomical scans. Two label overlap measures (false-positive error Dice similarity coefficient) success both depicting CN. differences these parameters KR HCP (with higher angular resolution) databases evaluated. Additionally, visualization 10 5 clinical cases evaluated comparing depictions intraoperative findings. RESULTS Maximum coefficients significantly (p < 0.001; Wilcoxon signed-rank test). false-positive last also lower than 0.001). data yielded better terms coefficient 0.001, Mann-Whitney U-test) = 0.02). errors smaller 0.001) showed strong trend toward significance 0.06). In cases, visualized 7 accurately, 3 correct CONCLUSIONS High resolution scans preferable DTI-based nerves. Probabilistic PICo effective task FA might represent useful since eliminates without manual intervention.