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

Area of Science:

• Medical Imaging
• Computer-Assisted Surgery
• Human-Computer Interaction

Background:

• Manual registration in AR-assisted surgery suffers from depth perception issues and high error rates.
• Virtual auxiliary tools are needed to improve the precision and efficiency of AR surgical navigation.

Purpose of the Study:

• To evaluate the effectiveness of two virtual tools, degrees of freedom (DOF) Separation and PinNPivot, in improving manual registration for AR-assisted surgical navigation.
• To compare these tools against unassisted manual registration in terms of accuracy, task completion time, and user workload.

Main Methods:

• A hybrid evaluation system combining Vuforia and NDI optical tracking was used.
• Two patient-specific phantoms were employed to test DOF Separation, PinNPivot, and unassisted manual registration.
• Key metrics included registration accuracy, task completion time, and NASA-TLX workload scores.

Main Results:

• DOF Separation yielded the highest accuracy but increased task completion time due to iterative adjustments.
• PinNPivot offered a balance between efficiency and accuracy, though initial pin placement errors were noted.
• Both assisted methods resulted in higher cognitive and physical workload scores compared to unassisted registration.

Conclusions:

• DOF Separation and PinNPivot significantly improve registration accuracy and efficiency in AR-assisted surgical navigation compared to manual methods.
• These software-based tools offer a promising, hardware-independent approach to enhance surgical navigation.
• Further validation in diverse clinical scenarios is recommended to assess their practical applicability.