No abstract

No abstract

Absrracf-We are attempting to evaluate somatosensory evoked potential (SEP) estimation methods in terms of their performance and reliability in addressing neurophysiology research questions. Standard signal analysis measures may describe improvements in such characteristics as signal-to-noise ratio without establishing whether that performance is adequate to serve as a tool in answering the question that will be investigated with the data. Therefore, testing the method's performance on control data would be useful. In this paper, we develop criteria for testing SEP estimation in order to study transient changes in short latency SEPs. The evaluations are based on latency stability of control data and the time resolution of the estimator.used in the rhesus studies, signal averaging.In order to evaluate SEP estimates for latency comparison studies, we selected two criteria: I ) latency stability in control data and 2) time resolution of the estimate. Latency stability was selected because latency shifts are measured in the impact acceleration studies and they should have little variability before impact. Any estimation method which resulted in highly variable pre-impact (control) latency measures would not be expected to be reliable for making pre-and post-impact comparisons.In addition, time resolution is important because of the transient nature of the response under investigation. Based on the rhesus monkey studies, we anticipate the need for individual estimates based on no more than several tens of seconds of data.I. ~NTRODUCTION 11. METHODS We are attempting to evaluate somatosensory evoked potential (SEP) estimation methods in terms of their performance and reliability in addressing neurophysiology research questions. In addition to signal averaging, several newer methods [1.2] have been developed and applied to the problem of evoked potential estimation. Typically, the performance of these approaches can be characterized according to standard statistical measures such as improvement of signal-to-noise ratio (SNR), error analysis, and selection of model order to meet information criteria. However, these measures alone may be insufficient to characterize the performance of an estimate in a particular application, such as monitoring latency changes.We require SEP estimates that can monitor transient latency changes over a short duration in order to study the effects of impact acceleration on the nervous system. Our prior rhesus monkey studies [3] suggest that SEPs could be used to monitor safe acceleration exposure. We found that transient latency changes in short latency SEPs occur at lower acceleration levels than those resulting in asymptomatic injury to the spinal cord. These transient changes occur and reverse in as few as tens of seconds. In order to pursue similar studies in humans, SEP estimates are needed that can resolve similarly short time spans despite a poorer SNR (due to the switch from implanted to surface electrodes). We begin our investigation of appropriate estimation methods by evaluating t...

No abstract

No abstract