Fortuna Galdieri

Background and Objectives: Sensorimotor rhythms (SMR) and their modulation are widely used in brain–computer interface (BCI) applications, but their reliable detection during overt upper-limb movements remains chal lenging due to motion-related artifacts and concurrent peripheral activity. This study investigates whether SMR modulation can be decoded during el bow flexion/extension performed under active and resistive conditions, with and without voluntary modulation. Methods: Nine healthy participants performed elbow flexion and ex tension across four run types: active (A), active with voluntary modulation (AM), resistive (R), and resistive with voluntary modulation (RM). Elec troencephalography (EEG), electromyography (EMG), and inertial measure ment unit (IMU) signals were synchronously acquired. EEG data were clas sified using a filter bank common spatial pattern (FBCSP)-based pipeline. EMG was analyzed to estimate the exerted force. Subjective workload was assessed using the NASA-TLX questionnaire. Statistical testing was con ducted to evaluate effects across experimental conditions. Results: Classification accuracies consistently exceeded chance level, with average values above (73 ± 14)% across participants and above (80 ± 12)% when discriminating SMR modulation regardless of movement type (A vs AM, R vs RM). EMG confirmed greater force output in resistive condi tions, while voluntary modulation was associated with increased muscular activation in most participants. Statistical testing highlighted effects mainly related to resistive conditions. In participants without significant force differ ences, classification remained above 70%, and event-related desynchroniza tion/synchronization analyses revealed distinct cortical modulations in the µ and β bands, supporting a neural origin of the decoded features. Workload ratings indicated higher cognitive demand during voluntary modulation and increased physical effort in resistive conditions, while combining resistance and voluntary modulation did not substantially increase perceived cognitive demand. Conclusions: The proposed framework enables detection of SMR modu lation during overt upper-limb movement, supporting rehabilitation scenarios in which movement resistance can be adaptively modulated based on real time neural activity.

This study focuses on leveraging deep learning and transfer learning techniques for classifying EEG signals recorded during motor imagery tasks. The primary objective is to develop a robust brain-computer interface (BCI) system capable of operating effectively in a zero-calibration scenario. Deep neural networks are investigated to improve classification accuracy. However, performance variability due to uncertainty sources is a critical challenge. To assess this variability, the effects of parameter initialization and pre-processing were analyzed. EEGNet and Sinc-EEGNet served as test models. Results indicate that network's weight initialization significantly impacts performance. Classification accuracy actually improves from 67 % ± 3 % to 73 % ± 3 % with adjusted initialization. Conversely, EEG pre-processing does not improve performance. As a result, pre-processing can be omitted to reduce computational effort. These findings support the development of real-time application scenarios.

The aim of this study is to characterize the acquisition delay in wireless EEG instruments and evaluate its impact on the detection of time-locked neural phenomena, such as P300 and movement-related cortical potentials (MRCP). Accurate timing is critical for both research and clinical applications, especially for real-time brain-computer interfaces (BCI). A measurement setup was thus developed to assess acquisition delays and their uncertainty. Delays were measured at both the start and stop of a reference signal generation to investigate the consistency and reliability of the devices. BCI experiments were also performed to evaluate the impact of the measured delay on the detection of the time-locked phenomena. Statistical tests confirmed significant differences in delays across devices and configurations (e.g., from few tens to a hundred ms). These delays directly impacted P300 and MRCP detection, raising concerns about potential misclassification. Nonetheless, the correction of the measured acquisition delay proved beneficial, especially with regard to the P300 latency measured through low-cost instrumentation.

Objective. A wearable brain–computer interface is proposed and validated experimentally in relation to the real-time control of an endless runner game by self-paced motor imagery(MI). Approach. Electroencephalographic signals were recorded via eight wet electrodes. The processing pipeline involved a filter-bank common spatial pattern approach and the combination of three binary classifiers exploiting linear discriminant analysis. This enabled the discrimination between imagining left-hand, right-hand, and no movement. Each mental task corresponded to an avatar horizontal motion within the game. Twenty-three healthy subjects participated to the experiments and their data are made publicly available. A custom metric was proposed to assess avatar control performance during the gaming phase. The game consisted of two levels, and after each, participants completed a questionnaire to self-assess their engagement and gaming experience. Main results. The mean classification accuracies resulted 73%, 73%, and 67% for left-rest, right-rest, and left-right discrimination, respectively. In the gaming phase, subjects with higher accuracies for left-rest and right-rest pair exhibited higher performance in terms of the custom metric. Correlation of the offline and real-time performance was investigated. The left-right MI did not correlate to the gaming phase performance due to the poor mean accuracy of the calibration. Finally, the engagement questionnaires revealed that level 1 and level 2 were not perceived as frustrating, despite the increasing difficulty. Significance. The work contributes to the development of wearable and self-paced interfaces for real-time control. These enhance user experience by guaranteeing a more natural interaction with respect to synchronous neural interfaces. Moving beyond benchmark datasets, the work paves the way to future applications on mobile devices for everyday use.