Probabilistyczna rekonstrukcja drogi wzrokowej w celu lepszego przewidywania zaburzeń widzenia po operacji epilepsji skroniowej: międzynarodowe wieloośrodkowe badanie

OBJECTIVE: Postoperative visual field deficits (VFDs) following anterior temporal epilepsy surgery often result from Meyer's loop (ML) injury. Optic radiation (OR) tractography aids surgical planning, but accuracy varies by method. Deterministic diffusion tensor imaging (Det-DTI) tractography underrepresents fiber complexity, whereas constrained spherical deconvolution with probabilistic tracking (Prob-CSD) resolves crossing fibers and improves anatomical fidelity. The aim of this study was to compare Det-DTI and Prob-CSD for OR/ML delineation and to evaluate accuracy by correlating tract injury with postoperative VFD severity. METHODS: Adults who underwent anterior temporal epilepsy surgery at two centers (Alfred Hospital in Melbourne, Australia; Vanderbilt University Medical Center in Nashville, Tennessee, US) were included. All patients underwent pre- and postoperative T1-weighted and diffusion-weighted MRI and automated postoperative perimetry. Det-DTI and Prob-CSD tractography pipelines, using the same expert-guided regions of interest, were compared. Tract injury was quantified as volumetric overlap between reconstructed ORs and resection cavities. VFD severity was measured as quadrant-specific visual field (QsVF) scores from perimetry. Reconstructed tract and estimated injury volumes were compared using the Wilcoxon signed-rank test, and their correlations with QsVF were assessed using Pearson correlation. Linear regression was used to assess whether Prob-CSD-estimated tract injury independently predicted postoperative QsVF, with resection volume and surgical technique as covariates. RESULTS: Overall, 33 patients (19 female, median age 39.1 years; 30 right handed and 11 left sided) were included in the analysis. Surgery comprised 9 anterior temporal lobectomies, 21 selective amygdalohippocampectomies, and 3 lesionectomies. Compared with Det-DTI, Prob-CSD tractography delineated significantly larger OR volumes (mean 20.18 ± 4.76 cm3 vs 1.48 ± 1.05 cm3; p = 8.34 ×10-7) and identified greater tract injury volumes (mean 0.60 ± 0.81 cm3 vs 0.04 ± 0.16 cm3; p = 3.7 ×10-6). Det-DTI failed to reconstruct ML in all but one case. Prob-CSD tract injury volumes correlated strongly with postoperative QsVF scores (r = -0.63, p = 7.35 ×10-5), whereas Det-DTI showed a weaker nonsignificant association (r = -0.31, p = 0.08). Linear regression confirmed independent association between Prob-CSD tract injury and postoperative VFD severity (p = 0.002, SE = 7.23 ×10-5, adjusted R2 = 0.34). CONCLUSIONS: Prob-CSD tractography outperformed Det-DTI in reconstructing ML and reliably predicting postoperative VFDs, regardless of surgical technique and resection extent. Det-DTI should be avoided for ML tractography. This study strongly supports incorporating expert-guided Prob-CSD into preoperative planning to enhance patient counseling and reduce risk of visual morbidity following temporal resection.