Validity and Reliability of OpenCap: A low-cost markerless motion capture system for lower extremity kinematics in...

The 3D marker-based optical system (MBS) is the benchmark for biomechanical measurement, but it is limited by cost, time, and accessibility. OpenCap, a web-based 3D markerless motion capture system, has the potential to address these limitations, although its validity and reliability in transfemoral prosthesis users remain unclear. This study aimed to evaluate the validity and within-session reliability of OpenCap in 31 transfemoral prosthetic users (mean age 44.52 ± 14.90 years). Participants performed walking trials on a dual-belt treadmill at a self-selected comfortable speed, while kinematic data were collected simultaneously using OpenCap and MBS. Statistical analyses included Intraclass Correlation Coefficients (ICC), Statistical Parametric Mapping (SPM1d), and Root Mean Square Error (RMSE). OpenCap showed excellent agreement for spatiotemporal parameters (ICC >0.99 with RMSE <2). However, agreement across joint kinematics varied between parameters, with lower accuracy observed in peak joint angle estimation. RMSE values generally ranged from 0 to 10°, with higher errors observed in hip rotation (>10°). For time-series kinematics, SPM1d analysis identified significant differences (p < 0.05) in joint kinematics across the gait cycle, particularly at the knee and ankle. Within-session reliability was good to excellent (ICC >0.75) for spatiotemporal parameters and several sagittal- and frontal-plane kinematic variables. Lower reliability (ICC = 0.584-0.733) was observed in transverse-plane measures on both sides and in some ankle measures on the prosthetic side, particularly hip rotation ROM on the prosthetic side. OpenCap provides valid spatiotemporal data and may serve as an accessible tool for gait assessment. However, joint kinematic accuracy requires improvement. Future research should focus on optimizing tracking for the complex prosthetic movements.