Design of a Virtual Reality Module for Learning Parotid Gland and Facial Nerve Anatomy

Surgeons who operate on the head and neck face a unique challenge: the human body is a three-dimensional puzzle, but most textbooks and medical images are flat. To learn the intricate map of muscles, nerves, and blood vessels, trainees have traditionally relied on dissecting human cadavers. While this remains the gold standard for understanding anatomy, it is expensive, difficult to access, and the tissue cannot be reused. In recent years, three-dimensional printing has offered a middle ground, creating physical models of specific body parts. However, these printed objects are static, costly to produce, and cannot easily incorporate other vital learning tools like surgical videos or dynamic imaging. Virtual reality, where a learner wears a headset to step inside a digital world, has emerged as a powerful alternative. It allows for immersive exploration of anatomy without the physical limitations of paper or plastic, yet until now, no such tool existed to teach the specific, complex region where the parotid gland and the facial nerve reside.

A team of researchers at the Mayo Clinic set out to build this missing piece of the educational puzzle. They focused on the parotid gland, the largest of the salivary glands located in front of the ear, and the facial nerve, which weaves through it to control the muscles of the face. This area is critical for surgeons performing parotidectomies, a procedure to remove tumors from the gland. If a surgeon accidentally damages the facial nerve during this operation, it can lead to permanent facial paralysis. The researchers wanted to create a virtual reality module that would allow students and residents to practice navigating this dangerous terrain safely, repeatedly, and without risk to a patient.

To build this digital environment, the team started with real medical data. They took de-identified CT scan images from a 29-year-old patient and used specialized software to trace and separate the different tissues, creating a high-fidelity digital model of the head and neck. This initial model was incredibly detailed, containing over 22 million tiny geometric shapes called polygons, which are the building blocks of 3D computer graphics. While this level of detail was perfect for a static image, it was far too heavy for a standalone virtual reality headset to process in real time. If the computer tried to render all those shapes at once, the image would lag or freeze, breaking the immersion.

The team then faced a significant technical hurdle: how to make the model light enough to run smoothly while keeping it accurate enough to be useful. They worked with a medical illustrator to sculpt and refine the digital anatomy, ensuring the shapes matched known human anatomy perfectly. Then, they systematically reduced the number of polygons, stripping away unnecessary detail until the entire model was down to about 806,000 polygons. This was a reduction of more than 96 percent. The result was a model that was light enough to run on a portable headset but still retained the clinical accuracy needed for surgical training. They created 65 distinct anatomical structures in total, though they ultimately used 55 of them in the final scene to ensure the software ran without errors.

The researchers did not just build a 3D model; they built a guided learning experience. They organized the module into a series of sequential steps, much like turning the pages of a book, but in a three-dimensional space. As the user moves through each step, a narrator explains the anatomy, and specific structures light up to draw the eye. For example, when the lesson focuses on the facial nerve, only that nerve and its immediate surroundings are highlighted, while the rest of the model remains dim. This technique, known as progressive disclosure, prevents the learner from being overwhelmed by too much information at once. The virtual room also includes other learning tools on the walls, such as a video of a real parotidectomy surgery and interactive viewers that let the user scroll through the original CT and MRI scans slice by slice. This allows the learner to connect the 3D virtual model with the flat images they would see in a real hospital.

To test their creation, the team invited 36 surgeons, residents, and medical students to try the module at various educational events. Most participants were able to navigate the virtual space on their own, moving through the steps and interacting with the controls without needing help. A few users, particularly those new to virtual reality, needed a moment to reorient themselves, suggesting that future versions might benefit from a slightly longer introduction to the controls. The researchers also encountered technical challenges during development, such as figuring out how to highlight specific parts of a bone or nerve that were part of a single, large digital object. They solved this by creating special digital textures that could change color instantly, effectively "painting" the highlighted areas without needing to rebuild the entire model.

The final product is a working proof-of-concept that demonstrates high-fidelity 3D models derived from real patient scans can be successfully adapted for immersive, standalone virtual reality. The module successfully integrated 65 anatomical structures, including the complex branching of the facial nerve, into a single, interactive environment. While the study did not measure whether students learned better with this tool compared to traditional methods, it established a clear, reproducible method for turning complex medical imaging into an accessible educational tool. The team showed that by combining expert narration, surgical video, and dynamic 3D models, they could create a learning environment that is both portable and rich in detail. This work fills a gap in medical education, offering a new way to teach the delicate anatomy of the face and neck to the next generation of surgeons.