Abstract Pre-operative planning for complex liver surgery is constrained by 2D CT/MRI. Digital 3D liver models (3DL-RL) may improve decision-making, but multicenter evidence on decision-level effects and inter-surgeon variability is limited. A multicenter, case-based survey was conducted among 10 HPB surgeons from all five Norwegian centers. Nine anonymized cases were reviewed in two phases: CT/MRI only, then CT/MRI + 3DL-RL. Fifty-five responses were collected to assess plans, usefulness, and adoption. Agreement and paired comparisons used Krippendorff’s α , Fleiss’ κ , Shapiro–Wilk, Levene’s, paired t , and Wilcoxon tests. 3DL-RL was associated with a significant planning shift (Wilcoxon p = 0.006, r = 0.372). Tumor counts changed in 41/55 (75%), always upward; plans changed in 64% of those versus 25% when counts were unchanged. Frequent modifications included parenchyma-sparing strategies 11/55 (20.0%), switch between surgery and another modality 6/55 (10.9%), extended resection 4/55 (7.3%), and refinements 8/55 (14.5%). Overall inter-surgeon agreement on binary planning remained low (Krippendorff’s α 0.207 CT/MRI vs 0.177 3DL-RL). Per-case anatomical agreement showed that α remains higher in the 3DL-RL group (0.664), with Cohen’s d = 0.571. Across Norwegian HPB units, 3DL-RL meaningfully influences pre-operative planning, often enabling parenchyma-sparing or other targeted changes, while serving as a complementary MDT aid. Variability persists across surgeons, consistent with individualized, precision-oriented decision-making. Larger, prospective, outcome-linked studies and ML-assisted workflows are needed to confirm clinical and economic impact.
Radiya et al. (2026) studied this question.