Abstracts DGNC

Background:Augmented reality (AR) headsets are used in medicine for experimental and educational applications.They can superimpose 3D holograms derived from imaging over body parts, e.g. the anatomy of the skull over a head.Ventriculostomies (VST) with external ventricular drains (EVDs) are usually performed using anatomical landmarks, which leads to unfavorable positions in 30-40% of cases.The presented project aims to improve this rate through holographic visualization of the cranial anatomy, entry point and target area for EVD placement.Commercially available AR glasses are used as the only active component in order to minimize disruption to the workflow of this standard procedure. Materials and methods:A modular head model and automated segmentation algorithm were developed.The precision of the setup was confirmed before a series of VSTs were performed.Routine computed tomography (CT) scans of the head models were used.Entry point and target of EVD placement are automatically computed and could then be verified by the surgeon using AR vision.A passive marker system is attached to the head model.An initial manual registration is required before the AR goggles can track the marker (this was bypassed using pre-registered data in order to streamline the test procedure).A Microsoft HoloLens Gen. 1 served as the hardware platform.Eleven surgeons performed two VST per head and placed a total of 110 EVDs under AR-guidance.Some weeks later, the procedure was repeated without AR (free hand procedure as per clinical standard).Four of the surgeons performed two additional AR-based series of VST to test for a possible learning curve.Each head model was equipped with an individually molded pair of ventricles before filling with agarose gel.Regular anatomy as well as anomalies such as midline shifts or deformities were included.The EVD catheters were left in place and CT scans were performed to verify the accuracy of placement. Results:The system functioned reliably and showed stable performance.The overall rate of successful freehand VST was 72.7% vs. 68.8%AR-based.The subgroup with repeated puncture series showed a significant improvement in precision and success rate, going from initially 65% to 95%.The surgeons rated the setup suitable for the test procedure and the overall feedback was positive, however some experienced visual issues when trying to keep hologram as well as head model fixated.General concerns were expressed regarding the reliability of the AR-headset hardware during surgical procedures.For most surgeons, it was the first encounter with AR-headset technology. Conclusion:AR-guided placement of EVDs in head models with different ventricular anatomy proved to be feasible.The precision of the AR-guided EVDs was satisfactory.Initial results did not differ significantly from success-rate of the free hand procedure.However, repeated use of the AR setup led to a significant improvement of results in subgroup.The results obtained in this project are promising.The learning curve with the proposed system appears to be steep.Several more developmental steps have to be taken before an application in humans can be considered safe.In the future, AR-headset based holograms have the potential to be well accepted in clinical application and serve as a compact navigational help inside and outside the operating room.