+1 Recommend
0 collections
      • Record: found
      • Abstract: found
      • Article: not found

      Separation of spinal cord motor signals using the FastICA method.

      Journal of neural engineering

      physiology, Spinal Cord, Sensitivity and Specificity, Reproducibility of Results, Pyramidal Tracts, Principal Component Analysis, methods, Pattern Recognition, Automated, Motor Neurons, Motor Cortex, Evoked Potentials, Motor, Electric Stimulation, Diagnosis, Computer-Assisted, Cats, Animals, Algorithms, Action Potentials

      Read this article at

          There is no author summary for this article yet. Authors can add summaries to their articles on ScienceOpen to make them more accessible to a non-specialist audience.


          Evoked motor signals descending down the corticospinal tract can be recorded selectively with multi-contact electrodes from the spinal cord surface. This method of extracting motor signals from the spinal cord may provide a means of communication for people with spinal cord injury. The information rate obtained with such an interface will improve if the separation of neural channels can be increased. In this study, the feasibility of increasing the channel separation was investigated using the blind source separation (BSS) technique. Neural signals recorded with multi-contact surface electrodes were treated as a linear mixture of independent neural sources located inside the spinal cord. Principal component analysis (PCA) was applied to estimate the dimensionality of the raw signals, and then the fixed-point FastICA algorithm was used to separate the primary neural sources from the secondary (smaller) ones. In all trials but one, the separation between the neural channels has increased by eliminating the secondary sources. These results suggest that the information rate of a spinal cord interface can be improved by separating the neural recordings into their independent components and selecting the ones with the largest distance between them. Comparison of independent component analysis (ICA) and PCA reveals that ICA performs better in this application.

          Related collections

          Author and article information



          Comment on this article