Neurosurg Focus. 2026 Nov 1;61(5):E4. doi: 10.3171/2026.8.FOCUS26434.
ABSTRACT
OBJECTIVE: Virtual reality (VR)-based platforms for aneurysm clipping are increasingly used for preoperative visualization, but objective validation against the final clip construct and postoperative imaging remains limited. The aim of this study was to evaluate the implementation of a patient-specific VR preoperative clip planning environment by comparing blinded simulated clip selections with operative clip choice and postoperative day 1 CTA in patients undergoing microsurgical clipping of unruptured anterior circulation aneurysms.
METHODS: The authors conducted a single-center retrospective validation study that included 10 patients who underwent elective clipping for an unruptured anterior circulation aneurysm between 2021 and 2025. Three vascular neurosurgeons independently performed 1 anonymized VR planning session per case at least 6 months postoperatively and were blinded to clinical data, operative documentation, clip records, and postoperative imaging. Life-size Yasargil clip models mirrored the institutional tray. Raters selected primary clip type, primary blade length, single- versus multiple-clip strategy, and anticipated dog-ear remnant; planning time was recorded. The operative reference standard was derived from operative documentation and clip records. Aneurysm occlusion, remnant status, and clip pose validation were based on postoperative day 1 CTA. The primary endpoint was a 4-point VR-Clip Concordance Index (VCCI).
RESULTS: Ten patients (6 female, median age 63.5 years [IQR 60.0-69.0 years]) who underwent clipping for an unruptured anterior circulation aneurysm and 30 corresponding VR simulations were included in this analysis. The mean VCCI was 2.00 ± 1.11 on a 4-point scale, and 10 of 30 simulations (33.3%) achieved VCCI ≥ 3 using exact primary blade-length agreement. Of 30 simulations, agreement was 17 (56.7%) for primary clip type, 21 (70.0%) for single- versus multiple-clip strategy, 18 (60.0%) for remnant expectation, and 4 (13.3%) for exact primary blade length. Blade-length agreement increased to 5 of 30 simulations (16.7%) with a ±0.5-mm tolerance and to 10 of 30 simulations (33.3%) with an exploratory ±1.0-mm tolerance. The mean VR planning time was 2.23 ± 1.57 minutes. The median positional deviation between simulated and postoperative clip poses was 2.41 mm (IQR 1.67-3.70 mm); rotational deviation was larger and more variable (median 44.42°, IQR 31.13°-64.95°). On postoperative day 1 CTA, 9 of 10 patients had aneurysms were that were completely occluded and 1 had a small dog-ear remnant.
CONCLUSIONS: This patient-specific VR environment for clip planning was feasible and fast; however, it was only partially concordant with the operative clip construct. There was stronger agreement for broad clip strategy than exact blade length, recognition of nonstraight clip geometries, multiclip reconstruction, or remnant prediction, which were identified as priority engineering targets before prospective clinical implementation.
PMID:42967353 | DOI:10.3171/2026.8.FOCUS26434