Article
Mobile, real-time and point-of-care augmented reality: A prospective pilot study with a phantom, animal and human models
Search Medline for
Authors
Published: | April 21, 2016 |
---|
Outline
Text
Background: Medical imaging is essential for the diagnosis and therapy of patients across a broad spectrum of medical disciplines. Imaging data is required at the patient’s bedside, but is usually accessed at desktop workstations. Advances in mobile technology make real-time, point-of-care medical applications possible. Traditional ways of viewing medical imaging data (i.e., Picture Archiving and Communication System (PACS) workstations) in a stacked fashion removed from the patient may change with augmented reality. Mobile augmented reality can assist surgeons in many clinical tasks that rely on anatomical knowledge, such as the visualization of adequate resection lines, identification of structures at risk, facilitation of trocar placement, and visual assistance in punctures. Our aim was to evaluate a mobile, real-time and point-of-care augmented reality system for medical diagnosis and therapy with regard to feasibility and accuracy in a pilot study involving phantom, animal, and human models.
Materials and methods: After computed tomography imaging a tablet computer was positioned above the patient and a semi-transparent 3D-representation of structures of interest were superimposed on top of the patient’s image resulting in augmented reality. Live camera images and the three-dimensional volume were registered by fiducial markers. Feasibility and accuracy were evaluated in a static model using the open source Heidelberg Laparoscopy Phantom (openHELP). The system was further analyzed in a porcine animal study. The reprojection error for both phantom and animal studies was defined as the average offset of the back-projected two-dimensional image points and the manually defined points in the three-dimensional volume. Finally the setup was tested with a human volunteer to prove basic feasibility for clinical application.
Results: In the phantom model of the 1380 analyzed AR-positions 83.9% could be successfully realized. The reprojection error was 2.83 ± 2.68 mm. 95% of the measurements were below 6.71 mm. In the animal model 79.3% of the 690 analyzed AR-positions could be successfully realized. In the animal study the reprojection error was 3.52 ± 3.00 mm. 95% of the measurements were below 9.49 mm. The reprojection error was significantly lower in the phantom model compared to the porcine model (P < 0.001). At last augmented reality was successfully realized in clinical case.
Conclusion: Mobile, real-time and point-of-care augmented reality systems for clinical purposes are feasible and accurate in a realistic experimental setting. Further research needs to be done in order to evaluate the implementation in a clinical setting.
Acknowledgement: The current study was conducted within the setting of Research Training Group 1126: “Development of New Computer-Based Methods for the Future Workplace in Surgery” and the Collaborative Research Center 125: Cognition Guided Surgery, both funded by the German Research Foundation (DFG).
Figure 1 [Fig. 1]