Vision‑aided brain–machine interface training system for robotic arm control and clinical application on two patients with cervical spinal cord injury

BioMed Eng OnLine, 2019 · DOI: https://doi.org/10.1186/s12938-019-0633-6 · Published: February 5, 2019

Neurology Neurorehabilitation Biomedical

Simple Explanation

This research introduces a new training system that combines visual information with brain signals to improve robotic arm control for individuals with spinal cord injuries. The system uses a camera to identify target objects and adjusts the robotic arm's movements based on both the user's brain signals and the visual data. The training system helps users learn to control the robotic arm more effectively by providing assistance through 'shared control,' where the system anticipates the user's intended target and makes the control easier. The system was tested on two patients with cervical spinal cord injuries, and brain scans showed that the training potentially helped to focus their brain activity in areas related to motor control.

Study Duration

Not specified

Participants

Two patients with cervical spinal cord injury

Evidence Level

Pilot clinical study

Key Findings

1
The proposed system significantly improved the distance error to the intended target compared to the unintended ones (p < 0.0001).
2
Functional magnetic resonance imaging after five sessions of observation-based training with the developed system showed brain activation patterns with tendency of focusing to ipsilateral primary motor and sensory cortex, posterior parietal cortex, and contralateral cerebellum.
3
Shared control with blending parameter α less than 1 was not successful and success rate for touching an instructed target was less than the chance level (= 50%).

Research Summary

This study introduces a vision-aided brain-machine interface (BMI) training system designed to enhance robotic arm control, particularly for individuals with cervical spinal cord injuries. The system integrates a Kinect camera for target object detection and employs an artificial potential approach for shared control, blending visual information with neural signals to facilitate more intuitive and accurate robotic arm movements. The system was evaluated in a pilot clinical study involving two patients with cervical spinal cord injury. Results indicated that the training system improved the accuracy of reaching intended targets and induced beneficial changes in brain activation patterns, as observed through functional magnetic resonance imaging (fMRI). Despite the potential benefits, the study also revealed limitations, including the unsuccessful implementation of shared control with certain blending parameters and the need for finer electrode spatial resolution to discern detailed brain activation patterns associated with specific motor imagery tasks.

Practical Implications

Rehabilitation potential

The vision-aided BMI system offers a non-invasive method to potentially improve motor skills and brain reorganization in patients with spinal cord injuries.

Robotics Advancement

The research contributes to the development of more intuitive and effective shared control strategies in robotic systems, combining visual and neural data.

BMI training optimization

Findings emphasize the importance of customizing blending parameters in shared control BMIs to maximize user success and motivation.

Study Limitations

1
The implemented target detection algorithm can automatically detect multiple targets; however, it is still limited in that target objects must be green and the performance can deteriorate for increasing numbers of target objects.
2
Using the proposed shared control strategy for the robot end-effector to reach the intended target, the success rate of reaching the instructed target did not exceed the chance level significantly.
3
So, as the future study, more invasive electrodes with finer spatial resolution can be considered.

Your Feedback

Was this summary helpful?

Back to Neurology