Utilizing microstimulation and local field potentials in the primary somatosensory and motor cortexGodlove, Jason (2014) Utilizing microstimulation and local field potentials in the primary somatosensory and motor cortex. Doctoral Dissertation, University of Pittsburgh. (Unpublished)
AbstractBrain-computer interfaces (BCIs) have advanced considerably from simple target detection by recording from a single neuron, to accomplishments like controlling a computer cursor accurately with neural activity from hundreds of neurons or providing instruction directly to the brain via microstimulation. However as BCIs continue to evolve, so do the challenges they face. Most BCIs rely on visual feedback, requiring sustained visual attention to use the device. As the role of BCIs expands beyond cursors moving on a computer screen to robotic hands picking up objects, there is increased need for an effective way to provide quick feedback independent of vision. Another challenge is utilizing all the signals available to produce the best decoding of movement possible. Local field potentials (LFPs) can be recorded at the same time as multi-unit activity (MUA) from multielectrode arrays but little is known in the area of what kind of information it possess, especially in relation to MUA. To tackle these issues, we preformed the following experiments. First, we examined the effectiveness of alternative forms of feedback applicable to BCIs, tactile stimuli delivered on the skin surface and microstimulation applied directly to the brain via the somatosensory cortex. To gauge effectiveness, we used a paradigm that captured a fundamental element of feedback: the ability to react to a stimulus while already in action. By measuring the response time to that stimulus, we were able to compare how well each modality could perform as a feedback stimulus. Second, we use regression and mutual information analyses to study how MUA, low-frequency LFP (15-40Hz, LFPL ), and high-frequency LFP (100-300Hz, LFPH) encoded reaching movements. The representation of kinematic parameters for direction, speed, velocity, and position were quantified and compared across these signals to be better applied in decoding models. Lastly, the results from these experiments could not have been accurately obtained without keeping careful account of the mechanical lags involved. Each of the stimuli affecting behavior had onset lags, which in some cases, varied greatly from trial to trial and could easily distorted timing effects if not accounted for. Special adaptations were constructed to precisely pinpoint display, system, and device onset lags. Share
Details
MetricsMonthly Views for the past 3 yearsPlum AnalyticsActions (login required)
|
![[feed]](/images/feed-icon-32x32.png){kind=icon}