Regensburg 2016 – wissenschaftliches Programm

Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

BP: Fachverband Biologische Physik

BP 22: Posters - Neurosciences

BP 22.2: Poster

Montag, 7. März 2016, 17:30–19:30, Poster C

How to estimate a threshold: theoretical limitations and practical implications — •Achim Schilling, Patrick Krauss, Konstantin Tziridis, and Holger Schulze — Experimental Otolaryngology, ENT-Hospital, University of Erlangen

We present a novel and robust method to universally estimate physiological and behavioral thresholds using the example of measurements of auditory brainstem responses (ABR) and pre-pulse inhibition (PPI) of acoustic startle responses (ASR). By definition the threshold defines the weakest stimulus strength that evokes a response significantly different from the non-stimulus condition. It is common practice that for threshold estimation measurements of physiological or behavioral responses to stimulus intensities that are close to the putative threshold are carried out. Unfortunately, the signal-to-noise ratio (S/N) naturally is worst near the threshold, since the intensities of evoked responses are positively correlated with stimulus strength. Here we demonstrate that thresholds may be estimated without any near threshold measurements if data are fitted to a generalized logistic function and an additive term representing the measured signal amplitude to the non-stimulus condition is added. We demonstrate that the goodness of fit becomes best if the supporting points are located within the area of the logistic function with the highest gradients, also referred to as its dynamic range, i.e. in a range with good S/N. To become independent from the number of measurement repetitions and the absolute noise level we perform stepwise subsampling with increasing sample-size followed by extrapolation and estimation of the asymptote.