Purchase this article with an account.
H. Morori - Katz, A. Loewenstein, J. R. Ferencz, I. Yeshurun, Y. Lang, T. Lifshitz, J. Karp, A. Pollack; Threshold Finding Strategies in Sensitivity Calibration of Home Macular Perimetry Device. Invest. Ophthalmol. Vis. Sci. 2008;49(13):4218.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
The HMP is a home perimeter designed to monitor the macular visual field of intermediate AMD patients. The HMP runs a psychophysical test that comprises hyperacuity stimuli, to which the patient responds by marking their spatial location. Calibration of stimulus amplitude (inversely related to sensitivity level, SL) to just above threshold ensures that the sensitivity of the test is adapted to the patient’s capabilities. A SL level at which > 70% of the stimuli are perceived was arbitrarily chosen as the optimal value for HMP usage. A key question is how to obtain an accurate evaluation of this threshold, with a minimal number of stimuli.The objective of the study is with the aid of computer simulations, to determine the optimal threshold strategy for performing visual field testing in intermediate AMD patients using the Home Macular Perimeter.
Two different strategies for threshold finding were examined. In the Decrease-Until-Fail (DUF) strategy, SL was gradually incremented if in a block of M successive stimuli >70% of the stimuli were responded correctly. The estimated threshold was defined as the most sensitive SL level at which this criterion could be met. In the Decrease-Increase (DI) strategy, SL was gradually incremented after N correct responses, or decremented after 1 incorrect response (the N:1 regime). In this case the estimated threshold was defined as the most visited SL level. To decide which strategy yielded the best estimation, we run a set of stochastic computer simulations, using realistic psychophysical functions of 129 intermediate AMD patients.
In DUF, the estimated threshold was underestimated by -1.5 % ± 15.4 %: in only 36% of the subjects, the estimated threshold was on average above the true threshold. In DI 2:1 regime, the estimated threshold was overestimated by +12 % ± 15.1%: in 79% of the subjects, the estimated threshold was on average above the true threshold.
Although DUF gave a more accurate estimation of the threshold, our simulations show that DIis a better strategy because it allows a larger proportion of subjects to be tested in the supra-threshold regime. Hence, the DI strategy was adopted for personal calibration of the HMP.
Clinical Trial: :
This PDF is available to Subscribers Only