Hand motion analysis associated with ultrasound-guided procedures
Purpose: To evaluate the ability of hand motion analysis using conventional and new motion metrics to differentiate between operators of varying levels of experience for central venous access (CVA) and liver biopsy (LB).
Methods: CVA task: 7 Interventional Radiologists (experts), 10 senior trainees, and 5 junior trainees performed ultrasound-guided CVA on a standardized manikin; 5 trainees were retested after one year. LB task: 4 Radiologists (experts) and 7 trainees biopsied a lesion on a manikin. Conventional motion metrics (path length and task time), a refined metric (translational movements), and new metrics (rotational sum, rotational movements), were calculated.
Results: CVA: experts outperformed trainees on all metrics (p ≤ 0.02). Senior trainees required fewer rotational movements (p = 0.02), translational movements (p = 0.045), and time (p = 0.001) than junior trainees. Similarly, on one-year follow-up, trainees had fewer translational (p=0.02) and rotational movements (p=0.003) with less task time (p=0.003). Path length and rotational sum were not different between junior and senior trainees or for trainees on follow-up. Rotational and translational movement had a greater area-under-the-curve (0.91 and 0.86 respectively) compared to rotational sum (0.73) and path length (0.61). LB: experts performed the task with a shorter path length (p=0.04), fewer translational movements (p=0.04), rotational movements (p=0.02), and less time (p<0.001) relative to the trainees.
Conclusion: Hand motion analysis using translational and rotational movements was better at differentiating levels of experience and improvement with training, as compared to the conventional metric of path length.
Weinstein JL, Ali H, Sarwar A, Dadour JR, Brook OR, Mitchell JD, Matyal R, Palmer MR, MacLellan C, Ahmed M. Evaluation of Technical Performance of Ultrasound-Guided Procedures through Hand Motion Analysis: an Exploration of Motion Metrics. J Vasc Interv Radiol. 2023 May 13:S1051-0443(23)00347-0. doi: 10.1016/j.jvir.2023.05.015. Epub ahead of print. PMID: 37187437.