ObjectiveTo verify the reliability and clinical feasibility of a self-developed navigation system based on an augmented reality technique for endoscopic sinus and skull base surgery.Materials and MethodsIn this study we performed a head phantom and cadaver experiment to determine the display effect and accuracy of our navigational system. We compared cadaver head-based simulated operations, the target registration error, operation time, and National Aeronautics and Space Administration Task Load Index scores of our navigation system to conventional navigation systems.ResultsThe navigation system developed in this study has a novel display mode capable of fusing endoscopic images to three-dimensional (3-D) virtual images. In the cadaver head experiment, the target registration error was 1.28 ± 0.45 mm, which met the accepted standards of a navigation system used for nasal endoscopic surgery. Compared with conventional navigation systems, the new system was more effective in terms of operation time and the mental workload of surgeons, which is especially important for less experienced surgeons.ConclusionThe self-developed augmented reality navigation system for endoscopic sinus and skull base surgery appears to have advantages that outweigh those of conventional navigation systems. We conclude that this navigational system will provide rhinologists with more intuitive and more detailed imaging information, thus reducing the judgment time and mental workload of surgeons when performing complex sinus and skull base surgeries. Ultimately, this new navigational system has potential to increase the quality of surgeries. In addition, the augmented reality navigational system could be of interest to junior doctors being trained in endoscopic techniques because it could speed up their learning. However, it should be noted that the navigation system serves as an adjunct to a surgeon’s skills and knowledge, not as a substitute.

No abstract

No abstract

Misleading depth perception may greatly affect the correct identification of complex structures in image-guided surgery. In this study, we propose a novel importancedriven hybrid rendering method to enhance perception for navigated endoscopic surgery. First, the volume structures are enhanced using gradient-based shading to reduce the color information in low-priority regions and improve the distinctions between complicated structures. Second, an importance sorting method based on the order-independent transparency rendering is introduced to intensify the perception of multiple surfaces. Third, volume data are adaptively truncated and emphasized with respect to the perspective orientation and the illustration of critical information for viewing range extension. Various experimental results prove that with the combination of volume and surface rendering, our method can effectively improve the depth distinction of multiple objects both in simulated and clinical scenes. Our importance-driven surface rendering method demonstrates improved average performance and statistical significance as rated by 15 participants (five clinicians and ten non-clinicians) on a five-point Likert scale. Further, the average frame rate of hybrid rendering with thin-layer sectioning reaches 42 fps. Given that the process of the hybrid rendering is fully automatic, it can be utilized in real-time surgical navigation to improve the rendering efficiency and information validity.

Identification of surgical instruments is crucial in understanding surgical scenarios and providing an assistive process in endoscopic image-guided surgery. This study proposes a novel multilevel feature-aggregated deep convolutional neural network (MLFA-Net) for identifying surgical instruments in endoscopic images. First, a global feature augmentation layer is created on the top layer of the backbone to improve the localization ability of object identification by boosting the high-level semantic information to the feature flow network. Second, a modified interaction path of cross-channel features is proposed to increase the nonlinear combination of features in the same level and improve the efficiency of information propagation. Third, a multiview fusion branch of features is built to aggregate the location-sensitive information of the same level in different views, increase the information diversity of features, and enhance the localization ability of objects. By utilizing the latent information, the proposed network of multilevel feature aggregation can accomplish multitask instrument identification with a single network. Three tasks are handled by the proposed network, including object detection, which classifies the type of instrument and locates its border; mask segmentation, which detects the instrument shape; and pose estimation, which detects the keypoint of instrument parts. The experiments are performed on laparoscopic images from MICCAI 2017 Endoscopic Vision Challenge, and the mean average precision (AP) and average recall (AR) are utilized to quantify the segmentation and pose estimation results. For the bounding box regression, the AP and AR are 79.1% and 63.2%, respectively, while the AP and AR of mask segmentation are 78.1% and 62.1%, and the AP and AR of the pose estimation achieve 67.1% and 55.7%, respectively. The experiments demonstrate that our method efficiently improves the recognition accuracy of the instrument in endoscopic images, and outperforms the other state-of-the-art methods.