Empirical Mode Decomposition-based filtering for fatigue induced hand tremor in laparoscopic manipulation

Fatigue induced hand tremor (FIT) is an unavoidable phenomenon, which substantially limits the accuracy of the surgical manipulation for long duration laparoscopic surgeries. Filtering intended motion from tremor is a challenging task as the properties of tremor change with increasing muscle fatigue levels. Muscle fatigue induced hand tremor has highly nonlinear and nonstationary characteristics that need a filtering strategy different from the conventional filters. Empirical Mode Decomposition (EMD) based filters have become popular in the recent past for its enhanced nonlinear signal handling capability. EMD based filtering strategy is case specific in nature as the EMD does not have any general analytical formulation unlike other (Kernel based) popular filtering techniques. In this work, we have addressed the tremor filtering issue with the help of EMD and the probability distribution characteristics analysis of Intrinsic Mode Functions (IMF) of the tremulous laparoscopic tool trajectory. A modified distribution asymmetry measure was employed to find out the threshold IMF for reconstruction of tremor free motion at different fatigue levels. In order to find the robustness of the proposed technique, the compensation strategy has been tested extensively on synthetic signal and experimentally acquired signals. Filtering threshold at different fatigue levels was also demonstrated for various subjects. Despite the time-varying properties of tremor, the proposed filtering strategy substantiates its efficacy to diminish the effect of tremor which was not possible by the conventional fixed cut-off filtering techniques.