Recent technological advances show the feasibility of offline decoding speech from neuronal signals, paving the way to the development of chronically implanted speech brain computer interfaces (sBCI). Two key steps that still need to be addressed for the online deployment of sBCI are, on the one hand, the definition of relevant design parameters of the recording arrays, on the other hand, the identification of robust physiological markers of the patient’s intention to speak, which can be used to online trigger the decoding process. To address these issues, we acutely recorded speech-related signals from the frontal cortex of two human patients undergoing awake neurosurgery for brain tumors using three different micro-electrocorticographic ([Formula: see text]ECoG) devices. First, we observed that, at the smallest investigated pitch (600[Formula: see text][Formula: see text]m), neighboring channels are highly correlated, suggesting that more closely spaced electrodes would provide some redundant information. Second, we trained a classifier to recognize speech-related motor preparation from high-gamma oscillations (70–150[Formula: see text]Hz), demonstrating that these neuronal signals can be used to reliably predict speech onset. Notably, our model generalized both across subjects and recording devices showing the robustness of its performance. These findings provide crucial information for the design of future online sBCI.

No abstract

Robotics rehabilitation is a widely used approach for the treatment of patients with severe motor disabilities, such as stroke survivors. Robots can provide intense, controlled and repeatable rehabilitation and they can also provide different levels of assistance when patients are not able to initiate or complete a movement. Nevertheless, several studies proved that completely passive movements are not sufficient to stimulate neuro-motor recovery and patients' engagement is a key factor for an effective rehabilitation. For this reason it is important to combine techniques for detection of movement intention (MI) with rehabilitation robotics. In this study we developed an algorithm capable of detecting MI before the movement onset, in order to obtain a trigger signal for providing robotics assistance. The proposed algorithm automatically selects the channels used to extract MI based on the motor-information content of each channel. The developed algorithm was tested on data recorded on n = 8 healthy subjects performing 3D reaching movements with an exoskeleton in active and assisted conditions. MI was detected about 400 ms before the beginning of the movement and the performance of the proposed method were significantly higher than the one achieved when six preselected channels, located over motor areas, were used for MI decoding. MI was also detected during robot-assisted movements. Interestingly, in active movements the highest performance was achieved with electrodes over a well-localized cluster above the contralateral and central motor areas, while in passive executions, the areas with the best performances became more sparse.