Improved accuracy of biomechanical motion data obtained during impacts using a time-frequency low-pass filter

Biomechanical motion data involving impacts are not adequately represented using conventional low-pass filters (CF). Time-frequency filters (TFF) are a viable alternative, but have been largely overlooked by movement scientists. We modified Georgakis and Subramaniam's (2009) fractional Fourier filter (MFrFF) and demonstrated it performed better than CFs for obtaining lower leg accelerations during football instep kicking. The MFrFF displayed peak marker accelerations comparable to a reference accelerometer during foot-to-ball impact (peak % error = -5.0 ± 11.4%), whereas CFs severely underestimated these peaks (30-70% error). During the non-impact phases, the MFrFF performed comparably to CFs using an appropriate (12-20 Hz) cut-off frequency (RMSE = 37.3 ± 7.6 m/s² vs. 42.1 ± 11.4 m/s², respectively). Since accuracy of segmental kinematics is fundamental for understanding human movement, the MFrFF should be applied to a range of biomechanical impact scenarios (e.g. locomotion, landing and striking motions) to enhance the efficacy of study in these areas.