Manual Registration in AR-Assisted Surgical Navigation: A Comparative Evaluation

Abstract

Purpose: This study evaluates two virtual auxiliary tools, degrees of freedom (DOF) Separation and PinNPivot, to address depth perception limitations and high error rates in manual registration for AR-assisted surgical navigation. Methods: DOF Separation decouples translation and rotation using six independent controls, minimizing cumulative errors. PinNPivot constrains object motion around virtual pins to stabilize rotation. Their effectiveness in AR remains underexplored. Using a hybrid evaluation system (Vuforia and NDI optical tracking), these tools were compared to unassisted manual registration on two patient-specific phantoms, assessing accuracy, task completion time, and NASA-TLX workload scores. Results: PinNPivot balanced efficiency and accuracy but was prone to initial pin placement errors. DOF Separation achieved the highest accuracy but required longer task times due to iterative adjustments. NASA-TLX results showed higher cognitive and physical workload for assisted methods. Conclusion: DOF Separation and PinNPivot improved registration accuracy and efficiency over unassisted manual registration. As software-based tools requiring no additional hardware, they hold promise for enhancing AR-assisted surgical navigation. Future work should validate their clinical applicability in diverse scenarios.