Comparison of Standard of Care Imaging vs Augmented Reality to Visualize Temporal Bone Structures

Introduction: Skull base surgery surrounding and involving the temporal bone requires intimate knowledge of the osseous anatomical landmarks. However, lack of reliable landmarks can be an immense challenge to various surgical procedures, for example the middle fossa approach. The purpose of this study is to evaluate the utility of augmented reality using the Microsoft HoloLens (AR-MH) to visualize structures in the temporal bone and plan the surgical approach.

Methods: Prior to the study, 18 temporal bone structures were segmented on 9 CT scans of different patients and 3D surface mesh models were generated on 3D Slicer, an open-source image processing software. Unique colors were randomly assigned to the segmented models, and the models were transmitted to the AR-MH over OpenIGTLink using a custom-built module on Slicer. Subjects were then asked to identify 16 structures in each of those 9 cases. For each case, each subject was randomly assigned to identify the structures with either 2D CT imaging (standard of care), 3D model visualization on a monitor, or 3D AR-MH, as shown in [Fig. 1]. While the modality for each case was randomly chosen, it was ensured that each subject would use each modality three times. For a 2D case, they were asked to simply mark the named structure with a fiducial, while for 3D and AR-MH cases, they were told to call out the color of the named structure according to a sheet provided (see [Fig. 1D]). Data recorded for each case included time to completion, accuracy, and the NASA Task Load Index: mental demand, physical demand, temporal demand, performance, effort and frustration.

Results: Seven radiologists (3 experts, 4 novices) and 7 otolaryngologists (3 experts, 4 novices) participated in the experiment. The mean time to identify 16 structures was 3:04 on 2D, 2:02 on 3D, and 2:09 on AR-MH. The mean accuracy was 89.0% on 2D, 91.9% on 3D, and 87.9% on AR-MH. However, to account for confusion with similarities between colors, some errors were negated to yield mean adjusted accuracies of 93.2% on 3D and 91.6% on AR-MH. Mean NASA Task Load Index values showed no significant difference in mental demand (2D: 4.5; 3D: 3.8; AR-MH: 4.1; p = 0.53), temporal demand (2D: 3.8; 3D: 3.5; AR-MH: 3.3; p = 0.65), performance (2D: 3.5; 3D: 4.0; AR-MH: 3.6; p = 0.68), effort (2D: 4.5; 3D: 3.5; AR-MH: 3.4; p = 0.39), or frustration (2D: 4.1; 3D: 3.1; AR-MH: 3.9; p = 0.49). However, in terms physical demand (2D: 2.6; 3D: 1.6; AR-MH: 2.7; p = 0.063), lower numbers were consistently selected for 3D modalities, indicating lower physical demand.

Conclusion: The results from these trials prove that AR-MH is successful in temporal bone visualization. Using AR-MH is more efficient than using 2D scans for rapid identification of the structures, while AR-MH and 3D models are comparable to one another. Although requiring greater physical demand, AR-MH has the potential to enhance 3D depth perception and real-time surgical navigation. Moreover, this platform may be used in education to train medical professionals.