Skip to Main Content
 

Global Search Box

 
 
 
 

Files

File List


case1301667321.pdf (2.87 MB)

ETD Abstract Container

Abstract Header

Improved decoding for brain-machine interfaces for continuous movement control

Marathe, Amar Ravindra
http://rave.ohiolink.edu/etdc/view?acc_num=case1301667321

Abstract Details

2011, Doctor of Philosophy, Case Western Reserve University, Biomedical Engineering.
People with severe paralysis have limited options for commanding assistive device movements. Accessing movement commands directly from the brain will increase the options available to this population and could enable them to control complex device movements. The studies discussed here focus on improving decoding algorithms and strategies commonly used in brain machine interfaces (BMIs) for continuous device movements. These studies focus on improving micro‐electrocorticography (μECoG) based BMIs that decode naturalistic arm movements, but many of the results shown here can be applied to all forms of continuous‐movement BMIs. In the first phase of the project, we evaluated how various spatial filtering methods affect decoding quality in μECoG‐based BMIs. Spatial filtering is the process of creating new signals from linear combinations of the raw signals in order to remove common noise and maximize the detection of unique information from the entire set of electrodes. Many novel variants of the common spatial pattern spatial filter were compared to three standard methods to determine which methods are most effective at improving decoding performance. Our results suggest that some novel variants of common spatial patterns developed here can dramatically improve field potential decoding. In the second phase of this project, we determined which movement parameters should be decoded from the brain to maximize BMI performance. Furthermore, in situations where the ideal parameter cannot be decoded, various options for using one decoded movement parameter to control another aspect of device movement may improve BMI performance. The final phase of the project evaluated how different characteristics of decoders vary as a consequence of using increased amounts of past data to predict the current arm state. We also quantified how subtle changes in offline decoders affect BMI performance during online use. This knowledge will enable people to develop BMIs that will be effective in real‐time. These studies taken together lay the foundation for developing a μECoG‐based BMI that can effectively control the movements of a device using neural signals associated with naturalistic arm movements.
Robert Kirsch (Committee Chair)
Dawn Taylor (Advisor)
Kenneth Gustafson (Committee Member)
Cameron McIntyre (Committee Member)
Lynn Landmesser (Committee Member)
104 p.
Biomedical Engineering
Brain-machine interfaces; electrocorticography; Brain-computer interfaces;

Recommended Citations

Citations

  • Marathe, A. R. (2011). Improved decoding for brain-machine interfaces for continuous movement control [Doctoral dissertation, Case Western Reserve University]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=case1301667321

    APA Style (7th edition)

  • Marathe, Amar. Improved decoding for brain-machine interfaces for continuous movement control. 2011. Case Western Reserve University, Doctoral dissertation. OhioLINK Electronic Theses and Dissertations Center, http://rave.ohiolink.edu/etdc/view?acc_num=case1301667321.

    MLA Style (8th edition)

  • Marathe, Amar. "Improved decoding for brain-machine interfaces for continuous movement control." Doctoral dissertation, Case Western Reserve University, 2011. http://rave.ohiolink.edu/etdc/view?acc_num=case1301667321

    Chicago Manual of Style (17th edition)

case1301667321
669
© 2011, all rights reserved.
This open access ETD is published by Case Western Reserve University School of Graduate Studies and OhioLINK.