Single-trial classification of motor imagery differing in task complexity: a functional near-infrared spectroscopy study

摘要: For brain computer interfaces (BCIs), which may be valuable in neurorehabilitation, signals derived from mental activation can monitored by non-invasive methods, such as functional near-infrared spectroscopy (fNIRS). Single-trial classification is important for this purpose and was the aim of presented study. In particular, we aimed to investigate a combined approach: 1) offline single-trial novel wireless fNIRS instrument; 2) use motor imagery (MI) task thereby discriminating between MI response different tasks complexities, i.e. simple complex tasks. 12 subjects were asked imagine either finger-tapping using their right thumb or sequential all fingers hand. recorded over secondary areas contralateral hemisphere. Using Fisher's linear discriminant analysis (FLDA) cross validation, selected each subject best-performing feature combination consisting one out three channel, an time interval ranging 5-15 s after stimulation onset 3) up four Δ[O2Hb] signal features (Δ[O2Hb] mean amplitudes, variance, skewness kurtosis). The results our showed that set channels, intervals comprising kurtosis, it possible discriminate single-trials differing complexity, versus (inter-task paired t-test p ≤ 0.001), with average accuracy 81%. Although accuracies look promising they are nevertheless considerable subject-to-subject variability. discussion address these aspects, limitations future approaches relevance neurorehabilitation.