Study of motor imagery for multiclass brain system interface with a special focus in the same limb movement

The presented study focuses on simple upper limb movements; lifting and clenching for use in Motor Imagery (MI) based Brain System Interface (BSI). Furthermore, we analyzed the same limb movement imagery and its validity in developing multi-class BSI. 15 subject datasets were analyzed using feature extraction and classification methods from regularized common spatial pattern toolbox. The results for offline analysis arc presented in this study. The multiclass analysis was done using One-Versus-One (OVO) and One-Versus-Rest (OVR) approaches using pre-selected feature extraction and classification. OVO and OVR binary classifiers were converted into 4 class classifiers using a novel score based method. Neither lifting nor clenching actions showed any substantial advantage over the other for binary classification. Furthermore, within the same limb, two kinesthetically close actions (lifting vs. clenching) show good differentiability even with the same shared neural pathway. However, further analysis of the performance of the same limb actions in the 4 class environment is required. The same limb movements, if successfully incorporated, shows potential in increasing the number of differentiable classes in a multiclass BSI.