Neurophysiologic evidence of motor imagery in lower limb amputees: an event-related potential study

Abstract

Abstract Background This study aims to investigate w lower limb amputation affects the motor imagery (MI) process. In order to understand the potential neural mechanisms of phantom limb pain, we have measured the cerebral activation of event-related potential (ERP) in lower limb amputees and healthy controls with comparing the relationship between phantom limb pain (PLP) and cerebral activation. In addition, there is a model of motor imagery based on lower limb amputation by using deep learning techniques. Methods This study includes 18 lower limb amputees and 20 healthy controls, who performed a bilateral lower limb motor imagery task. A 256-channel electroencephalographic system has been recorded to capture cerebral activation. Electrodes C3 and C4 (corresponding to the sensorimotor area) were selected to analyze the cerebral activation of the ERP. Besides, the level of PLP in the lower limb amputees has been assessed using the visual analog scale (VAS), while the correlation between the level of PLP and cerebral activation has been computed. Lastly, we have decoded the post-amputation motor imagery using deep learning techniques. Results The cerebral activation degree has been calculated as the potentials of electrodes C3 and C4 at 0-800 ms. What is more, the ERP amplitudes are smaller in healthy controls compared with those in lower limb amputees, and the correlation analysis shows a significant positive correlation between the level of PLP and cerebral activation in the sensorimotor area (P < 0.05). Finally, the deep learning training accuracy is as high as 83.7%. Conclusion Lower limb amputees should activate more neural activity to perform MI tasks, and PLP is involved in cerebral activation processes which may influence neural plasticity in sensorimotor areas. Additionally, the classification of MI potentials can be used as a basis for brain-computer interface (BCI) control strategies aimed at achieving more natural control of neural prostheses or robotic arms.