Feature domain-specific movement intention detection for stroke rehabilitation with brain-computer interfaces

Abstract

Brain-computer interface (BCI) driven electrical stimulation has been proposed for neuromodulation for stroke rehabilitation by pairing intentions to move with somatosensory feedback from electrical stimulation. Movement intentions have been detected in several studies using different techniques, with temporal and spectral features being the most common. A few studies have compared temporal and spectral features, but conflicting results have been reported. In this study, the aim was to investigate if complexity measures can be used for movement intention detection and to compare the detection performance based on features extracted from three different domains (time, frequency and complexity) from single-trial EEG. Two data sets were used where four different isometric palmar grasps or dorsiflexions were performed while continuous EEG was recorded. 39 healthy subjects performed or imagined these movements and 11 stroke patients attempted to perform the movements. The EEG was pre-processed and divided into two epoch classes: Background EEG (2 s) and movement intention (2 s). To obtain an estimated detection performance, temporal, spectral and complexity features were extracted and classified (linear discriminant analysis) after the feature vector was reduced using sequential forward selection. The results show that accuracies between 82-87% and 74-80% are obtained for foot and hand movements, respectively. The temporal feature domain was the most dominant for foot movement intention detection, while the spectral features contributed more to the hand movement detection. The complexity features could be used to detect movement intentions, but the performance was much lower compared to temporal and spectral features.

Low Latency Estimation of Motor Intentions to Assist Reaching Movements along Multiple Sessions in Chronic Stroke Patients: A Feasibility Study

“…Moreover, no interventions so far have tested BMI platforms decoding pre-movement BP and ERD patterns online in patients with brain damage due to a stroke. While BP detections online in healthy subjects doing ankle dorsiflexions have demonstrated to be reliable for BMI approaches, BPs in upper-limb movements (Hadsund et al, 2016 ; Martínez-Expósito et al, 2017 ), and specially in stroke patients (Daly et al, 2006 ) present particularities that make them less reliable for BMI applications, which may limit their usability in BMIs. Recently, it was demonstrated that an appropriate combination of BP- and ERD-based classifiers could lead to reliable and low-latency estimation of stroke patients' upper-limb motor intentions (Ibáñez et al, 2014a , b ).…”

Participant-specific classifier tuning increases the performance of hand movement detection from EEG in patients with amyotrophic lateral sclerosis

“…To obtain the spectral features, raw EEG signals were band-pass filtered in theta (4-8 Hz), alpha (8)(9)(10)(11)(12)(13), beta (13)(14)(15)(16)(17)(18)(19)(20)(21)(22)(23)(24)(25)(26)(27)(28)(29)(30) and gamma (30-80) bands using a zero phase second order Butterworth filter, and then squared. The filtered signals were then segmented into movement and rest epochs, and the mean was calculated over the same sliding windows as for the temporal features to estimate the average power.…”

“…This technique outperformed movement detection using template matching with raw EEG samples [21]. Another approach to constructing temporal features is to divide the signal segment into windows, with or without overlap, and then compute average parameters, such as the mean, slope and variability [23][24][25]. Temporal parameters extracted from an MRCP can be supplemented with spectral features capturing the change in power within relevant frequency bands [18,26].…”