Supplementary Figure from Spatio-temporal feature extraction in sensory electroneurographic signals

Confusion matrices obtained with an LDA classifier when using the STW framework to extract features from the sensory ENG data. The confusion matrices are shown for the six datasets tested in this study. The classes correspond to the six proprioception angles used to provide mechanical stimulation to the rat's hindlimb. Most of the misclassified instances occurred for neighbouring angle values such as 10° and -10°. The lowest classification accuracy scores were obtained for Datasets 4 and 6. These datasets correspond to nerve cuffs implanted more proximally in the nerve, i.e., further from the site where the sciatic nerve divides into the tibial, peroneal, and sural branches.