📄 medicine

Validation of a Mixed Reality Neuronavigation System Using Standardized 3D-Printed Skull Phantoms

This study validates a mixed reality neuronavigation system using 3D-printed skull phantoms, demonstrating that it achieves clinically acceptable accuracy and low latency, making it a feasible, cost-effective alternative to conventional optical systems.

Original authors: Jia-yan Li, Xun-jie Ma, Zhi-Jian Weng, Zhi-Heng Jian, Ji-you Wu, Jun Wang, Zhong-jie Shi, Gang Chen

Published 2026-08-25

📖 3 min read☕ Coffee break read

CC BY 4.0

Original authors: Jia-yan Li, Xun-jie Ma, Zhi-Jian Weng, Zhi-Heng Jian, Ji-you Wu, Jun Wang, Zhong-jie Shi, Gang Chen

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

In the operating room, a neurosurgeon's most critical task is often to find a tiny, hidden target deep inside the brain and remove it without disturbing the delicate, healthy tissue surrounding it. To do this safely, surgeons rely on navigation systems that act like a high-tech GPS, showing them exactly where their instruments are in relation to the patient's anatomy. For decades, these systems have worked by projecting a two-dimensional image onto a separate monitor. This forces the surgeon to constantly look away from the patient's head to check the screen, then look back to operate, breaking their focus and adding mental strain. A newer approach, known as mixed reality, attempts to solve this by projecting a three-dimensional hologram of the brain directly into the surgeon's view, allowing them to see the internal structures as if looking through the patient's skull. However, for this technology to be trusted in life-or-death situations, it must be proven to be just as accurate and responsive as the established systems currently in use.

Researchers at Zhuhai's People's Hospital and their collaborators set out to test whether a custom-built mixed reality navigation system could meet these strict standards. Instead of testing on human patients, which would be ethically complex and risky for a new device, they created a controlled testing ground using five highly precise 3D-printed models of human skulls. These models were fabricated from real medical scan data and included small steel nails placed at specific, known locations to serve as reference points. The team used these models to compare their new mixed reality system against a standard, commercial optical navigation system that is already widely used in hospitals. They measured how closely the digital map matched the physical model and how quickly the system responded when the surgeon moved their tools.

The results showed that the mixed reality system performed remarkably well. When the researchers measured the distance between the digital target and the actual steel nail, the new system was off by an average of 1.37 millimeters. The established commercial system was slightly more precise, with an average error of 1.1 millimeters. While the new system was not quite as perfect as the traditional one, both fell well within the safety margin that doctors consider acceptable for brain surgery, which is generally under 2 millimeters. This means the holographic guidance was accurate enough to be used safely in a real operating room.

Speed was another critical factor, as even a tiny delay between moving a tool and seeing it move on the screen could be dangerous. The researchers measured this delay by recording the system's response with high-speed video. They found that the mixed reality system reacted in about 108 milliseconds. This is fast enough to feel instantaneous to a human surgeon, ensuring that the holographic image moves in perfect sync with the physical instruments. The study concluded that this 3D-printed testing method provided a reliable way to validate the technology, proving that the mixed reality system offers a viable, cost-effective, and user-friendly alternative to the expensive and complex traditional navigation tools. By allowing surgeons to see the brain in three dimensions without looking away, this technology could reduce fatigue and improve the precision of future neurosurgical procedures.

Technical Summary: Validation of a Mixed Reality Neuronavigation System Using Standardized 3D-Printed Skull Phantoms

Problem Statement
Conventional optical neuronavigation systems, while indispensable for precise neurosurgery, present significant limitations: they require surgeons to frequently shift their gaze between the surgical field and a separate 2D monitor, increasing cognitive load and procedural complexity. Furthermore, these systems are characterized by high costs, complex hardware setups, and steep learning curves. While Mixed Reality (MR) navigation offers a potential solution by overlaying 3D holographic anatomical reconstructions directly onto the surgical field, rigorous quantitative validation of its accuracy and latency in neurosurgical settings remains limited. Existing studies often lack standardized testing environments or rely on early-generation hardware with reported errors ranging from 4.0 mm to 4.1 mm.

Methodology
This study employed a controlled experimental design to validate a custom-built MR neuronavigation system against a commercial optical navigation system (Anke ASA-610V).

Phantom Fabrication: Five standardized skull models were fabricated using high-precision 3D printing technology. These models were derived from thin-slice CT DICOM data of healthy adult subjects. The printing process utilized photosensitive resin and included the insertion of 1.1 mm steel nails to serve as fiducial markers and target points. The models were designed with specific anatomical landmarks (e.g., mastoid processes, supraorbital ridges) to ensure reproducibility.
System Architecture:
- Hardware: The MR system utilized Microsoft HoloLens 2 glasses, a notebook computer (Legion Y700P), a wireless router, a navigation probe, and high-reflective marker balls.
- Software: Preoperative models were processed using 3D Slicer and GeomagicWrap. The navigation software was a custom Windows application developed for HoloLens 2, supporting voice recognition, model transparency adjustment, and target point setting.
- Registration: The system employed a point-based rigid body registration method. Five non-coplanar virtual marker points on the 3D model were aligned with corresponding physical points (steel nails) on the phantom using a guide probe. A least-squares algorithm computed the transformation matrix to align the virtual and real spaces.
Calibration: The MR system was calibrated according to ASTM F2554-18 standards using a custom FR-554 multi-point calibration jig. This ensured tracking accuracy and corrected for optical distortion and tip offset.
Evaluation Metrics:
- Fiducial Planar Error (FPE): Measured the 2D planar deviation of registration points.
- Target Planar Error (TPE): Measured the 2D Euclidean distance between the virtual target point and the actual physical position of steel nails at specific anatomical landmarks.
- Latency: Quantified via frame-by-frame video analysis using two smartphones to measure the time delay ( $\Delta t = T_2 - T_1$ ) between physical probe movement and virtual response.

Key Results

Accuracy Comparison:
- The commercial optical navigation system achieved a mean FPE of 0.88 ± 0.04 mm and a mean TPE of 1.1 ± 0.44 mm.
- The custom MR system achieved a mean FPE of 1.29 ± 0.34 mm and a mean TPE of 1.37 ± 0.32 mm.
- Statistical analysis (paired t-test) revealed a significant difference in FPE between the two systems ( $p = 0.007$ ). However, both systems maintained mean errors well within the clinically accepted threshold of < 2 mm.
Latency Performance:
- The MR system demonstrated a mean latency of 108.38 ± 8.7 ms, corresponding to an update rate of approximately 9.2 Hz.
- This latency is below the 200 ms threshold considered clinically acceptable for real-time neurosurgical navigation.

Significance and Claims
The authors claim that this study provides evidence-based support for the clinical translation of MR navigation in neurosurgery. By utilizing standardized 3D-printed skull phantoms, the research successfully benchmarked a custom MR system against a conventional optical platform under controlled, reproducible conditions.

The paper concludes that the holographic MR navigation system offers acceptable accuracy and low latency, making it a feasible, cost-effective, and user-friendly alternative to traditional optical systems. The system addresses key limitations of conventional navigation by providing intuitive 3D visualization, reducing the need for gaze shifting, and utilizing a more portable hardware setup. The authors emphasize that while the MR system's accuracy is slightly lower than the optical standard, it remains within safe clinical limits, supporting its potential for broader adoption in routine neurosurgical practice.

Limitations Acknowledged
The study notes several limitations:

Measurement Surrogates: The use of vernier calipers to measure planar errors serves as a surrogate for true 3D registration error (FRE/TRE), potentially introducing manual bias.
Latency Measurement: Smartphone-based video recording is limited by frame rate resolution compared to high-speed imaging systems.
Scope of Metrics: Spatial pose accuracy (orientational errors) was not assessed, and the study did not evaluate dynamic latency drift over prolonged operation.

Future work is suggested to include standardized instruments for direct FRE/TRE quantification, high-speed imaging for latency profiling, and comprehensive spatial pose measurements.

Technical Summary: Validation of a Mixed Reality Neuronavigation System Using Standardized 3D-Printed Skull Phantoms

More like this