Objective: To develop and evaluate a non-invasive surgical assistance based on augmented reality (AR) in the detection of ureters on animal model.
Method: After an experimental prototyping step on two pigs to determine the optimal conditions for visualization of the ureter in AR, three pigs were operated three times at 1 week intervals. The intervention consisted of an identification of the ureter, with and without the assistance of AR. At the end of the intervention, a clip was placed on the AR-proposed ureter to evaluate its accuracy. By doing a cone beam computed tomography, we measured the distance between the contrasted ureter and the clips in the acquired volume. Thirteen videos were recorded, allowing subsequent evaluation of the clinical relevance of the device.
Results: The feasibility of the technique has been confirmed. The margin of error was 1.77 mm (± 1.56 mm) for ureter localization accuracy. In order to evaluate the perceived relevance and accuracy in the detection of AR-assisted ureter, 58 gynecological surgeons were shown the videos then questioned. Of the 754 responses obtained (13 videos × 58 surgeons), the ureter was identified in direct vision in 31.2% of cases versus 81.7% in AR (p value 3.62 × 10-7). When looking at pigs that had already had one or two operations, the ureter was identified in only 16% of cases with direct vision compared to 76.1% with AR (p-value 5.48 × 10-19). In addition, 67% of surgeons felt that AR allowed them to better identify the ureters and 61% that AR reconstruction was accurate.
Conclusion: This first AR device showed a satisfactory precision in the detection of ureters with a favorable opinion of surgeons. This surgical assistance system could be helpful in the performance of difficult procedures, for example in the case of patients, which have undergone multiple surgeries in the past.
Keywords: Augmented reality; Clinical relevance; Laparoscopy; Modeling of ureters; Precision.