Closed-Loop EEG BCI: VR and Electrical Stimulation to treat Neuropathic Pain

Abstract: Chronic pain is a life-threatening disease affecting over 20% of theU.S. adult population as of 2016. It impacts the physical as well as emotional components of a human being significantly affecting a person’s quality of life. This impact is due to the complex nature of pain which is a dynamic integration between cognitive and contextual processes. Persistent episodes of pain can have a psychological effect on the body causing distress, depression, anxiety, and mood fluctuations resulting in altered perceptions and cognition, and emotional instability. To prevent these implications of pain, there have been techniques to detect and treat pain. The detection of pain has been a research problem due to its multifaceted and subjective nature. There are pain detection therapies that rely on finding biomarkers of pain in physiological signals such asElectroencephalogram (EEG) and Magnetoencephalography. However, due to the lack of reliable and universal pain biomarkers, a framework forpain detection does not exist. Alongside, the present therapies are not wholesome as they are unable to target the emotional, cognitive, and physiological impact of pain. This makes chronic pain a challenge to society both economically and socially. In our research, we aim at developing a closed-loop Brain-Computer Interface (BCI) which detects pain and delivers a non-invasive therapy. Given the complexity of pain perception and its impact, we have designed a therapy that targets both the sensory and the emotional components of pain.   Therefore, we use Transcutaneous electrical nerve stimulation (TENS) to address the sensory and physical component and Virtual Reality (VR) to address attention andemotions. Moreover, the combination of TENS and VR allows the nerve stimulation sensation to be more realistic, providing a multisensory stimulation. This induces the illusion of ownership over a healthyvirtual body, modulating the subject’s body representation and having a beneficial impact on brain plasticity. We measure and analyse the brain’s response to pain by recording electroencephalogram signals. Due to the variable nature of EEG signals, we also use skin conductance along with EEG as a biosignal to detect and classify pain more accurately. For the system to work as a stand-alone BCI, the system is broken down into a training and a testing phase. In the training phase, we train a machine learning model on EEG and skin conductance (SC). This model is then used in the testing phase to detect pain in real-time. The systemis first implemented on healthy subjects by experimentally inducing painful and non-painful stimulation using TENS. The resultant system works as a closed-loop adaptive system which detects pain by continuously monitoring the EEG and skin Conductance of a person and delivers the therapy of VR and TENS. The system was successfully tested on 5 healthy subjects.