A Novel Multi-scale Depth and Spatial-Energy joint Optimization Approach for External Marker Selection in Respiratory Motion Tracking

In medical image-guided percutaneous interventions, respiratory motion-induced target displacement challenges surgical precision. Emerging indirect respiratory motion tracking predicts internal motion by modeling respiratory dynamics and monitoring external surrogate signals, but the placement of external markers is a critical factor influencing model accuracy. This study proposed a novel external marker selection method based on multi-scale depth features and spatial-energy joint optimization. Using 4D-CT -derived depth images, the method quantified thoracoabdominal motion heterogeneity and identified key feature regions via a composite scoring function and recursive elimination. A spatial-energy joint optimization algorithm refined marker placement while ensuring clinical safety constraints. Additionally, an improved sliding average filtering algorithm with outlier removal enhanced the robustness of external respiratory signals. Experimental results showed that the proposed method significantly reduced target registration error (TRE) in respiratory motion models, achieved an average TRE reduction of 0.2201 mm compared to conventional marker placement. The optimized marker layout improves signal accuracy and reliability, offering a promising approach for precise respiratory motion tracking in clinical applications.