Improved Accuracy and Lowered Learning Curve of Ventricular Targeting Using Augmented Reality—Phantom and Cadaveric Model Testing

Abstract

BACKGROUND: 

Augmented reality (AR) has demonstrated significant potential in neurosurgical cranial, spine, and teaching applications. External ventricular drain (EVD) placement remains a common procedure, but with error rates in targeting between 10% and 40%.

OBJECTIVE: 

To evaluate Novarad VisAR guidance system for the placement of EVDs in phantom and cadaveric models.

METHODS: 

Two synthetic ventricular phantom models and a third cadaver model underwent computerized tomography imaging and registration with the VisAR system (Novarad). Root mean square (RMS), angular error (γ), and Euclidian distance were measured by multiple methods for various standard EVD placements.

RESULTS: 

Computerized tomography measurements on a phantom model (0.5-mm targets showed a mean Euclidean distance error of 1.20 ± 0.98 mm and γ of 1.25° ± 1.02°. Eight participants placed EVDs in lateral and occipital burr holes using VisAR in a second phantom anatomic ventricular model (mean RMS: 3.9 ± 1.8 mm, γ: 3.95° ± 1.78°). There were no statistically significant differences in accuracy for postgraduate year level, prior AR experience, prior EVD experience, or experience with video games (P > .05). In comparing EVDs placed with anatomic landmarks vs VisAR navigation in a cadaver, VisAR demonstrated significantly better RMS and γ, 7.47 ± 0.94 mm and 7.12° ± 0.97°, respectively (P ≤ .05).

CONCLUSION: 

The novel VisAR AR system resulted in accurate placement of EVDs with a rapid learning curve, which may improve clinical treatment and patient safety. Future applications of VisAR can be expanded to other cranial procedures.