2025-07-02 –, Theil C1-4 (60p)
Accurate three-dimensional (3-D) appreciation of cranial anatomy is essential yet difficult to achieve with conventional two-dimensional (2-D) imaging. Lumi, an AI-driven, cloud-based mixed-reality (MxR) platform that originated from and is studied in close cooperation with the UMC Utrecht, automatically converts DICOM data into patient-specific holograms that can be explored on a head-mounted display or 2-D screen. In this presentation we summarise the current bench-to-bedside evidence base to quantify Lumi’s effect on spatial insight, surgical planning and task performance.
Methods
Evidence was drawn from an ex-vivo phantom study (n = 36), five prospective clinical series (n = 202) and a randomised crossover phantom trial of external-ventricular-drain (EVD) placement (n = 236). The predefined clinical end-point was the proportion of cases in which the surgical plan (derived from 2-D imaging) was meaningfully modified after review of the Lumi hologram in Mixed Reality (change in positioning, incision, craniotomy, or trajectory). Technical accuracy, workflow metrics and user-reported workload were also recorded.
Results
Ex-vivo study: Mean tumour-overlap improved by 50–72 % versus MRI alone and by 26–52 % versus a flat-screen 3-D viewer, irrespective of participant expertise.
Mixed-indication cohort (n = 107): A consecutive study of a mixed case series in neurosurgery (oncology, vascular, functional, skull base). Mixed Reality preparation prompted plan modification in 2.8 % without prolonging theatre preparation.
Posterior-fossa tumours (n = 21): Head position or craniotomy was altered in 3 patients (14 %), contributing to > 90 % resection in 81 % of cases (71 % mRS 0–1).
Extracranial–intracranial bypass (n = 10): Skin-incision adaptation occurred in 70 %, shortening arterial mapping and reducing donor-vessel injury to 0 % versus 30 % in historical controls.
Carotid endarterectomy (n = 39): Incision tailoring was noted in 13 patients (33 %), caused by depiction of bifurcation height and plaque length.
MCA aneurysm clipping (n = 25): Lumi altered approach or head orientation in 8 % . The mean skin-to-skin time was less that a matched historical cohort by 26 min (203 ± 64 vs 229 ± 78 min).
EVD placement phantom trial (n=236): Optimal catheter positioning (Kakarla-1) rose from 37.3 % to 57.6 % and gross misplacement fell from 40.7 % to 21.2 %; distance-to-target and angular error were significantly lower, at the cost of a 6-min median time penalty. NASA-TLX workload and SUS usability scores remained favourable throughout studies.
Conclusions
Across diverse cranial indications, Lumi mixed-reality provides three consistent advantages: (i) objectively superior 3-D anatomical insight, (ii) clinically meaningful modifications of operative strategy in 3–39 % of cases, and (iii) quantifiable gains in technical accuracy without workflow penalty. The platform thereby showed a low risk profile. Ongoing multicentre trials and structured post-market surveillance will further delineate long-term outcome effects, but current data already position Lumi with its combination of AI and MxR as a valuable adjunct in modern neurosurgery.
