Robustness Analysis of the CFB-MNE Approach

The aim of this study is to assess the robustness of a recent Motor Unit (MU) spatial localization method, namely, the Curve Fitting Based Minimum Norm Estimation (CFB-MNE) to several perturbations. This approach is based on using a MU dictionary for spatial localization. In other words, we want to evaluate if a unified dictionary works for any instrumental/physiological configuration or if there is a need for dictionary personalization. Firstly, to do that, we assumed that there was a difference between one of these parameters (Signal to Noise level (SNR), Conduction Velocity (CV), Muscle Fiber length) when constructing the dictionary and between the same parameter in the tested muscle configuration. Results showed that varying Signal to Noise Ratio (SNR) in a specific range, does not affect the performance of the algorithm. However, other parameters, like the CV mean value and the fiber length were influencing on the localization performance. Hence, we were convinced that personalized libraries within MU dictionaries must be saved ahead of time in a patient-specific way.