2026 Proffered Presentations
S180: PRACTICAL APPLICATIONS OF AI-DRIVEN 3D DESIGN AND PRINTING IN SURGICAL NEUROANATOMY: A LOW-COST SOLUTION FOR DEVICE PROTOTYPING IN THE SKULL BASE LAB
Marco Obersnel, MD; Chiara Angelini, MD; Hao Tang, MD; Roberto Rodriguez Rubio, MD; UCSF
Introduction: Cadaveric anatomical laboratories often face challenges due to limited or damaged instruments, and acquiring new tools can be costly and time-consuming. Designing novel surgical approaches requires the use of instruments that may be unavailable or unsuitable for cadaveric use. Recently, 3D printing has emerged as a valuable tool, primarily for creating educational models and conducting surgical planning. Studying 3D-printed anatomy enables trainees to understand spatial relationships better and approach specimens with clearer expectations, reducing specimen use and rapidly offsetting the 3D printer cost. However, its practical use for prototyping devices in a surgical simulation laboratory remains unexplored. We therefore investigated available user-friendly options to develop a customized instrument using 3D printing and evaluated the broader potential of this technology for surgical training and instrument development.
Methods: An initial test dissection of a submandibular retropharyngeal approach to the anterior clivus shown the need for a tubular retractor not commercially available for the long, minimally invasive corridor studied.
Three 3D modeling approaches were used to design a tubular retractor of approximately 10 cm length, 21 mm diameter, and with an angled tip:
- SDF Modeler 0.5 (Signed Distance Fields): digital construction using Boolean operations (Figure1A);
- Zoo Design Engineering Copilot 1.0.21 (Zoo, formerly KittyCAD, Los Angeles, USA): AI-powered Text-to-CAD modeling from a descriptive prompt (Figure1B);
- Tripo AI 3.0 (VAST, Dongcheng, China): generation of a 3D model from a 2D reference photograph (Figure1C).
Meshes were imported into Blender 4.5 (Blender Foundation, Amsterdam, Netherlands) for quick dimensional and quality assessment (Figure2), then prepared for printing in Bambu Studio (Figure3) and printed on a Bambu Lab P1S 3D printer (Bambu Lab, Shenzhen, China) using PLA and 90A TPU filaments (Figure4).
Results: Prototypes were printed in 1.34–1.57 hours, with filament costs of 0.25–0.44 USD, enabling immediate evaluation and iterative refinement. After three design iterations, a fully functional retractor was obtained. PLA offered greater rigidity, whereas TPU provided increased flexibility, improving comfort and ease of use in cadaveric tissue.
The retractors were successfully used in thirty dissections without signs of damage or physical deformation.
Among the modeling tools, SDF Modeler is open-source and user-friendly for organic modeling, Zoo Design Engineering Copilot provides professional-grade CAD outputs, and Tripo AI generated useful starting meshes from photographs and led to the better result in our case. Finally, Blender provided a versatile rapid qualitative comparison of models.
Following this experience, additional customized instruments, such as spatulas for white matter dissection, were produced, balancing the delicacy of wooden spatulas with the precision of steel instruments and offering multiple sizes for different dissection phases.
Conclusions: Integrating 3D printing into cadaveric lab workflows enables rapid, low-cost production of customized surgical instruments, particularly valuable in limited-resource settings. Basic 3D modeling skills suffice for instrument design and refinement, while AI-assisted tools expand possibilities for rapid prototyping and innovation. TPU represents an affordable, practical filament for flexible models, further broadening 3D printing applications.
A potential limitation involves intellectual property concerns when using AI tools that may replicate existing designs, primarily for commercial purposes.
