Towards a Mixed-Reality First Person Point of View Needle Navigation System
Ultrasound-guidance has reduced complications, such as carotid artery punctures, during central venous catherization (CVC). The continued prevalence of these complications has promoted the use of mixed-reality systems for surgical needle navigation. We have developed a surgical navigation system that renders the calibrated ultrasound (US) image and tracked models of the probe, needle and needle-trajectory. We compared the effectiveness of this guidance system on a desktop monitor or within a head-mounted display (HMD) to the US-only approach in a phantom-based user study with 33 expert clinical practitioners. These users performed one needle insertion on each of the two vessel sets within the phantom, where the first insertion was used as training and the second was used for analysis. The guidance system rendered within the HMD significantly improved the safety margin, defined as number of successful needle insertions, where the final needle position was within the lumen of the vessel, as 31 users performed successful insertions with the HMD system compared to 21 successful insertions under US-only guidance. Furthermore, the HMD system significantly improved the distance from the final needle tip to the vessel wall, as clinicians more consistently position the needle such that it was within the vessel lumen but far from the vessel wall when using the HMD system. The clinicians’ performance using the monitor system was comparable to the US-only guidance. Therefore, using a HMD to align the visual and motor fields of the clinician is imperative to successful needle guidance, promoting the continued pursuit of HMD guidance research.
KeywordsSurgical navigation Mixed reality Needle guidance Tracking Calibration Perception User performance
- 3.Chen, E.C.S., Peters, T.M., Ma, B.: Which point-line registration? In: Webster, R.J., Fei, B. (eds.) Proceedings of the SPIE 10135, Medical Imaging 2017: Image-Guided Procedures, Robotic Interventions, and Modeling, p. 1013509 (2017)Google Scholar