Accuracy of corticospinal tract reconstruction: DTI-based deterministic versus probabilistic tractography with intraoperative validation

Introduction The precise preoperative mapping of the corticospinal tract (CST) is crucial for preserving motor integrity during brain tumor resection. While Diffusion Tensor Imaging (DTI) is widely utilized, the comparative accuracy of DTI-based deterministic versus probabilistic tractography algorithms, validated against intraoperative direct electrical stimulation (DES), remains a critical area of investigation. Methods We retrospectively analyzed a single-center experience about patients operated on for brain tumors involving the motor areas between July 2024 and December 2025. CST reconstructions were generated using Medtronic StealthStation S8 software. Statistical analysis was applied to determine the most reliable reconstruction method, and the distance measured by each DTI technique (mm) was compared to the stimulation threshold (mA), assuming a 1:1 ratio (1mA = 1 mm). Results 40 patients underwent surgical resection and 8 patients, (mean age 58.33) meet the inclusion criteria. 30 stimulation sites were analyzed. The mean stimulation threshold was 9.83 ± 5.49 mA (range: 5–20 mA). The probabilistic tractography reached a mean distance of 12.01 ± 8.92 mm (range: 1–30.1 mm), while the DTI-based deterministic tractography showed a mean distance of 27.71 ± 17.14 mm (range: 8–71.4 mm). Paired mean difference between DES threshold and probabilistic tractography was not significant (mean difference: -2.2, 95% CI -5.9; 1.6, p=0.24), while it was significant between DES threshold and DTI-based deterministic reconstruction (mean difference: -17.9, 95% CI -24.9; -10.9, p0.001) revealing that this method significantly overestimated the distances. Probabilistic tractography also held lower RMSE against the gold standard compared to deterministic (respectively 10.1 vs 25.7). Conclusions Integrating probabilistic tractography with DES enhances the surgical safety profile for tumors involving the motor pathway. The quantitative reliability of this combined approach allows for optimized maximal-safe resection while minimizing postoperative motor deficits.