The target procedure, the endoscopic endonasal approach, or EEA, is a minimally invasive technique that reaches tumours at the base of the skull through the nasal passages, using an endoscope inserted through the nostrils and often a two-surgeon, “four hands” technique mirroring the operating theatre. It is among the most technically demanding corridors in surgery because the working field is a two-dimensional endoscopic view through a narrow, labyrinthine space crowded with critical structures. The proximity of the internal carotid arteries, which run lateral to the sella turcica; the optic nerves and chiasm, which sit directly above the tumour; and the brain itself means that complications can be catastrophic. Training for EEA has historically required cadaveric dissection laboratories, which are expensive to maintain, subject to strict sterility and ethical constraints around body donation, and simply inaccessible to many surgeons worldwide. Surveys cited by the authors even suggest that many healthcare workers, while willing to donate organs, are reluctant to donate their bodies, further straining the supply of anatomical specimens.

The Spanish team, drawn from Puerta del Mar University Hospital and the University of Cádiz’s Institute of Research and Biomedical Innovation, began with routine clinical imaging rather than any special acquisition. DICOM data from magnetic resonance imaging, with and without contrast, and CT angiography were used, with the CT angiography acquired at a slice thickness of 0.5 millimetres, a tube voltage of 120 kVp and a tube current of 200 mA. The magnetic resonance protocol specified a post-contrast, volumetric 3D T1-weighted sequence, such as MPRAGE, with isotropic 1-millimetre voxels, no interslice gap, approximately 170 to 220 slices covering the cranium, a field of view of roughly 240 to 260 millimetres, and a 256 by 256 matrix. Computed tomography supplied the data for segmenting the skull, nasal structures and carotid arteries, while magnetic resonance was used to delineate the tumour and the optic apparatus, an intentional division of labour that exploits the complementary contrast properties of the two modalities.

Segmentation and reconstruction were carried out in the open-source platform 3D Slicer, version 4.10.2, following a ten-step workflow designed for reproducibility. Native DICOM data were imported with spatial metadata preserved, orientation was verified through volume rendering, and a preliminary mask of the cranial structures was generated using threshold-based extraction in the Segment Editor. Disconnected components and unwanted regions were removed with the Islands tool, the anatomical volume of interest was isolated with the Scissors cropping function, and a controlled smoothing operation was applied to suppress imaging noise while preserving genuine anatomical detail. The refined segmentation was validated across axial, coronal and sagittal planes in the 3D Viewer before being exported as a printable surface mesh in STL or OBJ format. A final mesh-repair step, performed in the open-source program Meshmixer, closed residual holes, corrected surface normals and guaranteed manifold geometry suitable for printing, while a posterior support structure was added to stabilise the skull during simulation.

Printing was performed at 1:1 scale on an Ultimaker S5 Pro Bundle using fused filament fabrication, with toolpaths generated in UltimakerCura. The choice of materials reflected the mechanical character of the tissues being imitated: the three cranial bones were printed in rigid polylactic acid, producing a stable structure that faithfully represents bone, while the internal carotid arteries, optic nerve and chiasm, tumour and nasal cavity were printed in flexible thermoplastic polyurethane, which better reproduces the compliance of soft tissue. For the skulls, a water-soluble polyvinyl alcohol support interface was used, yielding a superior surface finish and allowing supports to be removed easily without damaging delicate bony architecture. The parts were printed separately and then assembled anatomically: the tumour was fixed within the sella turcica, the two carotid arteries were threaded through their respective foramen lacerum to follow their natural course lateral to the tumour, and the optic chiasm and nerves, printed as a single integrated unit, were seated superior to the tumour at their entry into the optic canal. The entire bill of materials came to just 8.40 euros.

Validation followed a formal endoscopic rehearsal and a structured questionnaire. Twelve evaluators, three neurosurgery consultants, three neurosurgery residents, three ENT consultants and three ENT residents, performed the endoscopic endonasal approach on the models, introducing the endoscope through one nostril and a high-speed drill through the other, drilling the bony corridor to expose the tumour in the sella, exactly as in theatre. Each judge scored items covering anatomical realism, haptic and drilling sensation, and clinical usefulness on a 0-to-10 Likert scale. Statistical analysis in R applied the non-parametric Kruskal-Wallis test to compare consultants with residents and neurosurgeons with ENT specialists, together with Kendall’s W coefficient of concordance to measure agreement among raters and Spearman correlations to examine the influence of seniority, with significance set at a two-sided p value below 0.05.

The results were broadly encouraging. Scores were favourable across nearly all categories for both consultants and residents, and there were no significant differences between the neurosurgery and ENT departments in any variable assessed, suggesting the model holds value for both specialties that share the endonasal corridor. The only statistically significant differences between consultants and residents appeared in the items on anatomy and on the surgical view, where residents scored markedly higher, 9.00 versus 6.12 for anatomy and 8.50 versus 5.62 for the surgical visualisation. The authors attribute this pattern to experience: consultants already know the approach and therefore gain less from a training replica, whereas residents, early on their learning curve, found the model especially valuable for understanding anatomical landmarks and the operative corridor. Kendall’s W returned a borderline p value of 0.0505, which did not reject the null hypothesis of agreement among raters, and the consistently high mean scores reinforced the models’ practical utility. A strong positive correlation among usefulness, perceived clinical applicability and expected skill improvement was also reported.

The study was candid about its limitations. The original CT scan contained insufficient slices to fully print the ethmoid and sphenoid sinuses and the lamina papyracea, highlighting how dependent the technique is on imaging protocols designed for diagnosis rather than fabrication; the authors argue that acquisition standards must change if per-case printing is to become routine. They also observed that prolonged drilling melted the thermoplastic due to heat from the high-speed burr, although this problem was neatly solved by irrigating with saline, exactly as surgeons already do in the operating theatre. Negative Spearman correlations, though not significant, hinted that more experienced clinicians rate such simulators slightly lower, and the authors propose that future refinement with additional post-processing tools could improve the challenging sinonasal regions.

Even so, the cost and speed figures are difficult to ignore. Cadaver laboratories carry heavy financial burdens in specimen procurement, preservation and sterilisation, alongside ethical obligations; animal models, such as porcine donors described in other work, sidestep some ethical issues but not the expense or the risk of cross-species contamination; and improvised trainers built from capsicum peppers and tomatoes familiarise trainees with instruments but convey nothing of real nasal anatomy or tissue feel. Against these options, a patient-specific, multi-material model fabricated in two days for 8.40 euros, using hardware already present in many hospitals and engineering departments, positions 3D printing as a scalable substitute for body-donor training. Beyond rehearsal, the same models have documented value in preoperative planning and patient counselling, with prior studies showing that visualising a printed replica of one’s own tumour measurably improves patients’ understanding of their pathology, the planned approach and the potential complications.

The Cádiz team concludes that collaboration between engineering departments and skull base surgical units can raise the technical capability of practitioners and extend the complexity of procedures safely offered at their centre. If imaging protocols are standardised and printed materials further optimised, the authors argue, every neurosurgical and ENT department could one day rehearse each individual patient’s tumour before the first incision, converting the most hazardous moments of skull base surgery into procedures a surgeon has already, in effect, performed. For a field where the learning curve is measured in nervous structures rather than in minutes, that arithmetic may prove decisive for patient safety.