Comparison of algorithms of testing for use in automated evaluation of sensation

P. J. Dyck, J. L. Karnes, D. A. Gillen, P. C. O’Brien, I. R. Zimmerman, D. M. Johnson

Research output: Contribution to journalArticlepeer-review

54 Scopus citations


Estimates of vibratory detection threshold may be used to detect, characterize, and follow the course of sensory abnormality in neurologic disease. The approach is especially useful in epidemiologic and controlled clinical trials. We studied which algorithm of testing and finding threshold should be used in automatic systems by comparing among algorithms and stimulus conditions for the index finger of healthy subjects and for the great toe of patients with mild neuropathy. Appearance thresholds obtained by linear ramps increasing at a rate less than 4.15 µm/sec provided accurate and repeatable thresholds compared with thresholds obtained by forced-choice testing. These rates would be acceptable if only sensitive sites were studied, but they were too slow for use in automatic testing of insensitive parts. Appearance thresholds obtained by fast linear rates (4.15 or 16.6 µm/sec) overestimated threshold, especially for sensitive parts. Use of the mean of appearance and disappearance thresholds, with the stimulus increasing exponentially at rates of 0.5 or 1.0 just noticeable difference (JND) units per second, and interspersion of null stimuli, Bekesy with null stimuli, provided accurate, repeatable, and fast estimates of threshold for sensitive parts. Despite the good performance of Bekesy testing, we prefer forced choice for evaluation of the sensation of patients with neuropathy.

Original languageEnglish (US)
Pages (from-to)1607-1613
Number of pages7
Issue number10
StatePublished - Oct 1990

ASJC Scopus subject areas

  • Clinical Neurology


Dive into the research topics of 'Comparison of algorithms of testing for use in automated evaluation of sensation'. Together they form a unique fingerprint.

Cite this