Enhancing sensorimotor BCI performance with assistive afferent activity: An online evaluation

Area of Science:

• Neuroscience
• Biomedical Engineering
• Rehabilitation Technology

Background:

• Brain-computer interface (BCI) control accuracy is crucial for user efficacy.
• Existing BCI methods face challenges with users exhibiting insufficient control accuracy.
• Procedural strategies are needed to enhance BCI performance for specific user groups.

Purpose of the Study:

• To investigate if assistive muscular stimulation below the motor threshold can improve motor imagery classification for BCI.
• To evaluate the effectiveness of training classifiers on data combining motor imagery and sub-threshold stimulation (BOTH) versus motor imagery alone (MI).
• To explore potential applications for individuals with motor impairments, such as ALS and stroke patients.

Main Methods:

• Offline analysis and online experiments were conducted with healthy participants.
• Experimental conditions included: Motor Imagery (MI) alone, sensory threshold stimulation (STM) alone, and BOTH (MI with STM).
• Classifiers were trained on MI or BOTH data, and neurofeedback was provided during online MI tasks.

Main Results:

• Offline analysis demonstrated superior BCI accuracy when decoding MI using a classifier trained on BOTH data compared to MI data alone.
• Online experiments confirmed improved accuracy for MI decoding when using classifiers trained on BOTH data.
• Sensorimotor connectivity patterns in specific frequency bands during the BOTH condition predicted performance in the MI condition.

Conclusions:

• Training BCI classifiers with data from combined motor imagery and sub-threshold neuromuscular electrical stimulation (BOTH) enhances BCI accuracy.
• This approach offers a promising new avenue for training sensorimotor rhythm-based BCIs, especially for users with control difficulties.
• It presents a viable alternative for patients with motor impairments (e.g., ALS, stroke) who retain afferent pathway function.