-
1
CT-guided interventions rely on precise needle placement, necessitating accurate real-time tracking to navigate complex anatomy.
-
2
Existing needle-tracking systems often lack dynamic uncertainty estimates, leading to potential procedural errors and unrecognized inaccuracies.
-
3
The proposed framework provides real-time uncertainty levels as a percentage, correlating with spatial tracking error for informed decision-making.
-
4
Three uncertainty-assessment strategies are compared: classic metrics, end-to-end CNN, and a hybrid CNN that balances interpretability and automation.
-
5
The framework enhances transparency and trust in needle tracking, supporting safer surgical workflows amid dynamic operating conditions.
