gms | German Medical Science

Objective: Augmented reality (AR) headsets allow to superimpose 3D holograms derived from imaging onto body parts, e.g. the cranial anatomy onto a head. External ventricular drainages (EVDs) are usually placed under guidance of anatomical landmarks, resulting in unfavourable positions in 30-40% of cases. The presented project provides an approach on improving this rate through holographic visualisation of the cranial anatomy, entry point and target area for the EVD placement. Commercially available AR glasses were used as the sole active component in this setting. The goal was minimal disruption to the workflow of this standard procedure.

Methods: A modular head model and segmentation algorithm were developed. Precision of the setup was confirmed. Routine computed tomography (CT) scans of the models were used. Entry point and target for EVD placement are computed automatically. A passive marker system was attached to the head model to be tracked by the headset.

Eleven surgeons conducted two ventriculostomies per head to place a total of 110 EVDs. Four of the participants did two additional series to derive a possible learning curve. Abnormalities such as midline shifts or deformities were included in the individually shaped ventricles. EVD catheters were left in place for CT-scans to check for placement accuracy.

Results: The setup worked reliably and showed stable performance. The overall rate of successful ventriculostomies was 68.8%. The mean offset of the EVDs from the displayed reference trajectory was 5.2±2.6mm (mean ± standard deviation). Overall feedback was positive. For the majority it was the first encounter with AR headset technology. General concerns were raised about reliability of the AR hardware in regard to surgical procedures.

The subgroup with repeated puncture series showed a significant improvement in precision and successrate.

Conclusion: AR guided EVD placement in head models with varying ventricular anatomy was shown to be feasible. Precision of AR guided EVDs was satisfactory.

To our knowledge, this is the first report on AR guided ventriculostomy based on a marker system. The results achieved in this project are promising. Good acceptance of this compact navigation tool could be expected in a presumed clinical application. The learning curve with the proposed setup seems to be steep.

Before a translation into clinical application and use in humans is safe, further development on the marker system and visuo spatial uncertainties have to be addressed.