Spinal Cord Injury Rehabilitation

Our lab is working on collaborative projects with researchers at the Institute Santos Dumont (ISD) in Brazil. Our collaborators at the ISD conduct research on rehabilitation methods to assist individuals with Spinal Cord Injury (SCI). Through the use of Brain-Computer Interfaces (BCI), robotic rehabilitation devices, and functional electrical stimulation, they seek to understand what methods can best assist in recovery from SCI.

Several of their rehabilitation protocols aim to reestablish or strengthen the connection from conscious thought to physical sensation and movement. Using non-invasive sensors placed on the scalp (electroencephalogram or EEG), the brain activity that is generated when a person thinks about moving one of their limbs, called motor imagery or motor intent, can be detected. These signals are then decoded through machine learning algorithms and converted into command signals for robot-assisted leg braces used for gait therapy or for muscle stimulation devices for the affected extremities.

An example of the type of system used for SCI rehabilitation. Adapated from N. A. Grigorev et al., IEEE Trans. Neural Syst. Rehabil. Eng., 2021.

One of the challenges encountered by our collaborators is suboptimal classification performance when decoding motor imagery. In the Parasol lab, we are investigating methods to assist with these rehabilitation protocols by addressing both sides of the BCI interface: first, generation of the motor imagery by the human (the person imagines moving their left foot) and second, the classification methods used to convert from motor imagery to commands for the robotic systems (muscle stimulation is provided to the left leg).

Immersive Visualizations

Motor imagery (MI), the mental simulation of a physical movement, can be used for brain-computer interface rehabilitation protocols to promote neuroplasticity and aid rehabilitation for spinal cord injury. However, individuals with long-term motor impairment or paralysis may struggle to imagine specific movements for a limb affected by the SCI, reducing treatment effectiveness. We developed a virtual reality (VR) simulation designed to demonstrate the foot movements that must be imagined during a rehabilitation protocol for individuals with SCI used by our collaborators at the Institute Santos Dumont. We plan to conduct a research study later this year to investigate if this VR environment can aid in the generation of more reliable patterns of MI brain activity.

Video 1: Actual protocol

Video 2: VR protocol

Undergraduates Natasha Bhatia and Aashna Anand presented a poster about the VR visualization at the 2026 Grainger College of Engineering Illinois Scholars Undergraduate Research (ISUR) Expo. They received the Audience Choice Poster Award for their poster, “Virtual Reality Simulation for Spinal Cord Injury Rehabilitation.” Congratulations Tasha and Aashna!

Tasha and Aashna's poster.

Tasha and Aashna receive their award.

Classification Methods

In addition to assisting individuals in generating stronger patterns of decodable brain activity, we are also exploring state-of-the-art and novel classification methods to improve the accuracy of models used during rehabilitation. We have begun a pre-registered literature review on classification methods which have been investigated in studies with participants who have a spinal cord injury. Once we have finished synthesizing the available work on this topic and comparing the performance of the available methods, we will test new models informed by an in-depth understanding of the most effective methods for classification of brain activity from individuals with spinal cord injury.