Usability Study of Augmented Reality Visualization Modalities on Localization Accuracy in the Head and Neck: Randomized Crossover Trial

Yao Li ^1,2, Gijs Luijten ^3,4,5, Christina Gsaxner ^3,5

⁰Department of Oral and Maxillofacial Surgery, University Hospital RWTH Aachen, Pauwelsstrasse 30, Aachen, 52074, Germany, 49 2418088231.

Jmir Serious Games +

|

January 13, 2026

View abstract on PubMed

Summary

Superimposition (SI) augmented reality was more accurate and faster for point localization than virtual twin (VT) holography in head and neck phantom studies. However, VT may be suitable for scenarios not requiring high accuracy.

Area Of Science

Medical Imaging
Augmented Reality
Surgical Navigation

Background

Augmented reality head-mounted displays (AR-HMDs) aim to bridge the spatial gap between medical imaging and the surgical field, especially in complex head and neck anatomy.
While markerless AR-HMDs are cost-effective, their overlay accuracy (superimposition, SI) is limited.
Virtual twin (VT) holography presents a potential alternative, but its performance requires further investigation.

Purpose Of The Study

To compare the accuracy and efficiency of SI and VT augmented reality visualization modalities for anatomical localization in the head and neck.
To evaluate performance based on accuracy, workload, time, and user experience.

Main Methods

A randomized crossover trial involving 38 participants using HoloLens 2.
Localization of point, line-based, and volume-based anatomical structures on head phantoms.
Evaluation of accuracy (absolute, relative, surface distance, Hausdorff distance), workload (NASA-TLX), and time.

Main Results

SI demonstrated superior point localization accuracy compared to VT (P<.001).
Accuracy for line-based structures was comparable between SI and VT (P>.51).
VT showed significantly less deviation for volume-based structures than SI (P<.01).
Participants were faster using SI (P=.02), with no significant difference in workload (P=.79).

Conclusions

SI offers better point localization accuracy and speed but struggles with overlaid soft tissue and viewing challenges.
VT remains a viable option for surgical scenarios where high precision is not paramount.
Future research should enhance viewing angle guidance and anatomical target-to-skin surface linkage for AR systems.

Keywords:

computer-assisted surgery human-machine interface mixed reality preoperative planning visualization techniques