Towards quantitative ergonomic assessment in robotic surgery using depth imaging

Abstract

Robot-assisted surgery lowers postural load for surgeons, yet desk-like ergonomic challenges persist. We evaluate posture monitoring using the Azure Kinect. Two surgeons performed console training tasks in deliberately “good” and “poor” postures. Using Kinect’s body tracking, we estimated the sagittal plane from spinal landmarks and reoriented the skeleton. Three angles were extracted: Lower-Spine (LS), Upper-Spine (US) and Arm-Spine (AS). Surgeon-independent thresholds for LS and US separated favorable from unfavorable postures for the interquartile range, whereas AS was less discriminative. These preliminary results support the feasibility of an automated posture feedback. Validation on a more diverse cohort will refine proposed metrics.