Hybrid Soft-Rigid Active Prosthetics Laboratory Exercise for Hands-On Biomechanical and Biomedical Engineering Education.

This paper introduces a hands-on laboratory exercise focused on assembling and testing a hybrid soft-rigid active finger prosthetic for biomechanical and biomedical engineering (BME) education. This hands-on laboratory activity focuses on the design of a myoelectric finger prosthesis, integrating mechanical, electrical, sensor (i.e., inertial measurement units (IMUs), electromyography (EMG)), pneumatics, and embedded software concepts. We expose students to a hybrid soft-rigid robotic system, offering a flexible, modifiable lab activity that can be tailored to instructors' needs and curriculum requirements. All necessary files are made available in an open-access format for implementation. Off-the-shelf components are all purchasable through global vendors (e.g., DigiKey Electronics, McMaster-Carr, Amazon), costing approximately USD 100 per kit, largely with reusable elements. We piloted this lab with 40 undergraduate engineering students in a neural and rehabilitation engineering upper year elective course, receiving excellent positive feedback. Rooted in real-world applications, the lab is an engaging pedagogical platform, as students are eager to learn about systems with tangible impacts. Extensions to the lab, such as follow-up clinical (e.g., prosthetist) and/or technical (e.g., user-device interface design) discussion, are a natural means to deepen and promote interdisciplinary hands-on learning experiences. In conclusion, the lab session provides an engaging journey through the lifecycle of the prosthetic finger research and design process, spanning conceptualization and creation to the final assembly and testing phases.

Enhanced System Robustness of Asynchronous BCI in Augmented Reality Using Steady-State Motion Visual Evoked Potential.

This study evaluated the effect of change in background on steady state visually evoked potentials (SSVEP) and steady state motion visually evoked potentials (SSMVEP) based brain computer interfaces (BCI) in a small-profile augmented reality (AR) headset. A four target SSVEP and SSMVEP BCI was implemented using the Cognixion AR headset prototype. An active (AB) and a non-active background (NB) were evaluated. The signal characteristics and classification performance of the two BCI paradigms were studied. Offline analysis was performed using canonical correlation analysis (CCA) and complex-spectrum based convolutional neural network (C-CNN). Finally, the asynchronous pseudo-online performance of the SSMVEP BCI was evaluated. Signal analysis revealed that the SSMVEP stimulus was more robust to change in background compared to SSVEP stimulus in AR. The decoding performance revealed that the C-CNN method outperformed CCA for both stimulus types and NB background, in agreement with results in the literature. The average offline accuracies for W = 1 s of C-CNN were (NB vs. AB): SSVEP: 82% ±15% vs. 60% ±21% and SSMVEP: 71.4% ± 22% vs. 63.5% ± 18%. Additionally, for W = 2 s, the AR-SSMVEP BCI with the C-CNN method was 83.3% ± 27% (NB) and 74.1% ±22% (AB). The results suggest that with the C-CNN method, the AR-SSMVEP BCI is both robust to change in background conditions and provides high decoding accuracy compared to the AR-SSVEP BCI. This study presents novel results that highlight the robustness and practical application of SSMVEP BCIs developed with a low-cost AR headset.

Can a highly accurate multi-class SSMVEP BCI induce sensory-motor rhythm in the sensorimotor area?

Objective. Different visual stimuli might have different effects on the brain, e.g. the change of brightness, non-biological movement and biological movement.Approach. In this study, flicker, checkerboard and gaiting stimuli were chosen as visual stimuli to investigate whether steady-state motion visual evoked potential (SSMVEP) effect on the sensorimotor area for rehabilitation. The gaiting stimulus was designed as the gaiting sequence of a human. The hypothesis is that only observing the designed gaiting stimulus would simultaneously induce: (1) SSMVEP in the occipital area, similarly to an SSVEP stimulus; and (2) sensorimotor rhythm (SMR) in the primary sensorimotor area, because such action observation could activate the mirror neuron system. Canonical correlation analysis was used to detect SSMVEP from occipital electroencephalograms (EEG), and event-related spectral perturbation was used to identify SMR in the EEG from the sensorimotor area.Main results. The results showed that the designed gaiting stimulus-induced SSMVEP, with classification accuracies of 88.9 ± 12.0% in a four-class scenario. More importantly, it induced clear and sustained event-related desynchronization/synchronization (ERD/ERS), while no ERD/ERS could be observed when the other two SSVEP stimuli were used. Further, for participants with a sufficiently clear SSMVEP pattern (classification accuracy >85%), the ERD index values in the mu-beta band induced by the proposed gaiting stimulus were statistically different from those of the other two types of stimulus.Significance. Therefore, a novel brain-computer interface (BCI) based on the designed stimulus has potential in neurorehabilitation applications because it simultaneously has the high accuracy of an SSMVEP (sim90% accuracy in a four-class setup) and the ability to activate the sensorimotor area.

Comparing user-dependent and user-independent training of CNN for SSVEP BCI.

OBJECTIVE: We presented a comparative study on the training methodologies of a convolutional neural network (CNN) for the detection of steady-state visually evoked potentials (SSVEP). Two training scenarios were also compared: user-independent (UI) training and user-dependent (UD) training. APPROACH: The CNN was trained in both UD and UI scenarios on two types of features for SSVEP classification: magnitude spectrum features (M-CNN) and complex spectrum features (C-CNN). The canonical correlation analysis (CCA), widely used in SSVEP processing, was used as the baseline. Additional comparisons were performed with task-related components analysis (TRCA) and filter-bank canonical correlation analysis (FBCCA). The performance of the proposed CNN pipelines, CCA, FBCCA and TRCA were evaluated with two datasets: a seven-class SSVEP dataset collected on 21 healthy participants and a twelve-class publicly available SSVEP dataset collected on ten healthy participants. MAIN RESULTS: The UD based training methods consistently outperformed the UI methods when all other conditions were the same, as one would expect. However, the proposed UI-C-CNN approach performed similarly to the UD-M-CNN across all cases investigated on both datasets. On Dataset 1, the average accuracies of the different methods for 1 s window length were: CCA: 69.1% ± 10.8%, TRCA: 13.4% ± 1.5%, FBCCA: 64.8% ± 15.6%, UI-M-CNN: 73.5% ± 16.1%, UI-C-CNN: 81.6% ± 12.3%, UD-M-CNN: 87.8% ± 7.6% and UD-C-CNN: 92.5% ± 5%. On Dataset 2, the average accuracies of the different methods for data length of 1 s were: UD-C-CNN: 92.33% ± 11.1%, UD-M-CNN: 82.77% ± 16.7%, UI-C-CNN: 81.6% ± 18%, UI-M-CNN: 70.5% ± 22%, FBCCA: 67.1% ± 21%, CCA: 62.7% ± 21.5%, TRCA: 40.4% ± 14%. Using t-SNE, visualizing the features extracted by the CNN pipelines further revealed that the C-CNN method likely learned both the amplitude and phase related information from the SSVEP data for classification, resulting in superior performance than the M-CNN methods. The results suggested that UI-C-CNN method proposed in this study offers a good balance between performance and cost of training data. SIGNIFICANCE: The proposed C-CNN based method is a suitable candidate for SSVEP-based BCIs and provides an improved performance in both UD and UI training scenarios.

A Convolutional Neural Network for the Detection of Asynchronous Steady State Motion Visual Evoked Potential.

A key issue in brain-computer interface (BCI) is the detection of intentional control (IC) states and non-intentional control (NC) states in an asynchronous manner. Furthermore, for steady-state visual evoked potential (SSVEP) BCI systems, multiple states (sub-states) exist within the IC state. Existing recognition methods rely on a threshold technique, which is difficult to realize high accuracy, i.e., simultaneously high true positive rate and low false positive rate. To address this issue, we proposed a novel convolutional neural network (CNN) to detect IC and NC states in a SSVEP-BCI system for the first time. Specifically, the steady-state motion visual evoked potentials (SSMVEP) paradigm, which has been shown to induce less visual discomfort, was chosen as the experimental paradigm. Two processing pipelines were proposed for the detection of IC and NC states. The first one was using CNN as a multi-class classifier to discriminate between all the states in IC and NC state (FFT-CNN). The second one was using CNN to discriminate between IC and NC states, and using canonical correlation analysis (CCA) to perform classification tasks within the IC (FFT-CNN-CCA). We demonstrated that both pipelines achieved a significant increase in accuracy for low-performance healthy participants when traditional algorithms such as CCA threshold were used. Furthermore, the FFT-CNN-CCA pipeline achieved better performance than the FFT-CNN pipeline based on the stroke patients' data. In summary, we showed that CNN can be used for robust detection in an asynchronous SSMVEP-BCI with great potential for out-of-lab BCI applications.

A Convolutional Neural Network for Enhancing the Detection of SSVEP in the Presence of Competing Stimuli.

Stimulus proximity has been shown to have an influence on the classification performance of a steady-state visual evoked potential based brain-computer interface (SSVEP-BCI). Multiple visual stimuli placed close to each other compete for neural representations leading to the effect of competing stimuli. In this study, we propose a convolutional neural network (CNN) based classification method to enhance the detection accuracy of SSVEP in the presence of competing stimuli. A seven-class SSVEP dataset from ten healthy participants was used for evaluating the performance of the proposed method. The results were compared with the classic canonical correlation analysis (CCA) detection algorithm. We investigated whether the CNN parameters learned on one inter-stimulus distance (ISD) can generalize across to other ISDs and sessions. The proposed CNN obtained a significantly higher classification accuracy than CCA in both the offline (75.3% vs. 67.9%, (p <; 10-3)) and the simulated online (71.3% vs. 60.7%, (p <; 10-3)) conditions for the closest ISD. The results suggest the following: the CNN is robust in decoding SSVEP across different ISDs, and can be trained independent of the ISD resulting in a model that generalizes to other ISDs.