The LF-ASD Brain Computer Interface: On-Line Identification of Imagined Finger Flexions in Subjects with Spinal Cord Injuries
Steven G. Mason, Ziba Bozorgzadeh, Gary E. Birch · 2000 · Proceedings of the Fourth International ACM Conference on Assistive Technologies (Assets '00) · doi:10.1145/354324.354350
Summary
This paper presents the first evaluation of the Low-Frequency Asynchronous Switch Design (LF-ASD) brain-computer interface with individuals who have high-level spinal cord injuries. The LF-ASD is a brain-controlled switch designed specifically for asynchronous control applications — situations where the user issues commands infrequently and at their own pace, rather than in response to system-timed prompts. The system works by continuously sampling surface EEG electrodes positioned over the Supplementary Motor Area and Sensory-Motor Cortex, detecting imagined voluntary movement-related potentials (IVMRPs) when users imagine finger flexions. Prior work had validated the LF-ASD with able-bodied subjects performing both actual and imagined finger movements, achieving hit rates of 70-81% with low false positive rates. This study extended that work to two male subjects with high-level spinal cord injuries (C-5 level) who had no residual sensation or motor function in their hands. The subjects were trained for approximately two hours before completing two 1.5-hour test sessions, during which they imagined right index finger flexions to activate the switch. The research represents an important step in establishing that individuals with spinal cord injuries can produce detectable IVMRPs many years after their injury, opening the door to brain-controlled assistive devices for this population.
Key findings
The two subjects with spinal cord injuries achieved online hit rates between 36% and 48% while maintaining false positive rates below 1%. Specifically, Subject 1 demonstrated hit rates of 44% across both sessions with false positive rates of 0.2-0.3%, while Subject 2 achieved rates of 35-48% with false positives between 0.1-0.3%. These results confirm that individuals with high-level spinal cord injuries can produce IVMRPs detectable by the LF-ASD system, even years after injury. The study also found that the LF-ASD parameters did not need to be customized for the subjects — the same configuration used for able-bodied subjects worked effectively. Averaged EEG signals at electrode C1 showed strong IVMRP patterns consistent with those seen in able-bodied subjects, confirming the switch was responding to genuine motor imagery rather than artifacts. EOG contamination analysis showed minimal impact from eye movements, with contaminated hits ranging from 0-4% of total hits.
Relevance
This early research demonstrated the feasibility of brain-computer interfaces for people with severe motor disabilities, a population that stands to benefit most from BCI technology. The asynchronous design is particularly significant for real-world assistive technology applications — unlike synchronous BCIs that operate on system-defined schedules, the LF-ASD allows users to activate the switch whenever they choose, which is essential for controlling wheelchairs, communication devices, and environmental controls. Although the hit rates were modest by modern standards, the sub-1% false positive rates were promising for practical use. The work laid groundwork for subsequent BCI research focused on assistive applications and highlighted the importance of testing with the actual target user population rather than only able-bodied proxies.
Tags: brain-computer interface · spinal cord injury · EEG · switch access · motor imagery · assistive technology · asynchronous control