Comparing augmented reality-assisted and freehand external ventricular drain placement: a multicenter randomized controlled crossover phantom study

BACKGROUND: External ventricular drain (EVD) placement is a common neurosurgical procedure with high rates of misplacement when performed using the freehand technique. With augmented reality (AR), the accuracy of EVD placement could be improved by providing a 3-D overlay, guiding optimal placement using a virtual trajectory superimposed over the patient. In this study, we aimed to assess the efficacy and usability of an AR application for assisting EVD placements which supported trajectory planning, point-based image-to-patient registration and 3-D stereoscopic projection.

METHOD: We conducted a randomized controlled crossover trial involving 15 neurosurgical residents and one neurosurgeon, who performed 236 EVD procedures (118 AR-assisted and 118 freehand) on biomimetic phantoms. EVD placement accuracy was evaluated using the Kakarla scale, distance-to-target, angular inaccuracy, and depth inaccuracy. The total procedural time was recorded. The user experience was evaluated using the NASA Task Load Index (NASA-TLX) and the Usefulness, Satisfaction, and Ease of Use (USE) questionnaire.

RESULTS: AR-assisted placement achieved significantly higher rates of optimal placement (Kakarla grade 1: 57.6% vs 37.3%; p < .001), lower rates of erroneous placement (Kakarla grade 3: 21.2% vs 40.7%; p < .001), a lower distance-to-target (median, 7.2 mm vs 11.4 mm; p < .001) and lower angular inaccuracy (median, 5.58° vs 7.60°; p < .001). Procedural time was longer for AR (median, 7 min 30 s vs 1 min 11 s; p < .001). Participants rated the AR system favorably on the USE for ease of learning (mean, 6.09/7 [SD, 0.94]) and satisfaction (mean, 6.45/7 [SD, 0.69]), while NASA-TLX scores indicated similar workloads between AR and freehand techniques.

CONCLUSIONS: AR improves the accuracy of EVD placement compared to the freehand technique, which is expected to improve the efficacy in clinical settings. It increases total procedural time but remains within clinically acceptable limits and provides favorable usability.