Patient-Specific 3D Liver Reconstruction for Surgical Planning: From Segmentation to Surgical Decision

Patient-specific three-dimensional (3D) reconstruction converts cross-sectional imaging into individualised anatomical models of the liver, its vasculature and tumours and is increasingly used to plan hepatectomy. Its appeal is intuitive: it externalises the spatial reasoning that surgeons must otherwise perform mentally from two-dimensional images. Whether that appeal is matched by evidence varies greatly across the pathway from image to outcome. This critical review appraises the evidence chain—acquisition, segmentation, anatomical and quantitative representation, surgical planning, decision correctness, intraoperative translation and clinical benefit—against the reference standards appropriate to each step. The evidence is strongest where a physical referent exists: whole-liver and parenchymal segmentation are reproducible, and whole-liver and future-liver-remnant volumetry are among the best-validated outputs, albeit with characterised systematic and patient-level errors. It weakens progressively for tumour, small-vessel, arterial, biliary and territorial reconstruction and for computed vascular territories, which are inferred rather than directly observed. Reconstruction reliably changes operative plans, but whether those changes are correct is poorly established, and the operation performed is not an independent reference standard. Intraoperative registration remains approximately centimetric in several clinical series, and perioperative outcome and cost-effectiveness evidence is largely observational. The central message is that the maturity of 3D reconstruction is unevenly distributed: technical feasibility and volumetry are comparatively mature, whereas decision correctness, intraoperative validity and demonstrated patient benefit are not. Progress now depends less on more elaborate models than on reference standards, surgically meaningful validation metrics and prospective outcome evidence.