Evaluation and comparison of three popular computer-based contrast sensitivity measurement techniques.

Abstract

Purpose: To compare three computer-based methods for measuring the contrast sensitivity function (CSF): the method of limits (LMT), the method of adjustment (ADJ) and a four-alternative forced-choice (4-AFC) paradigm using the quick CSF (qCSF) algorithm.

Methods: Contrast sensitivity was assessed monocularly in 47 participants (mean age: 26.2 ± 5.6 years) using the LMT and ADJ methods with ascending and descending series (five spatial frequencies, three trials) and the qCSF method (up to 150 trials across 21 frequencies). Accuracy, reliability and precision were evaluated using paired t-tests, intraclass correlation coefficients (ICC), coefficient of repeatability (CoR) and coefficient of variation (CoV). Method durations were also compared.

Results: Ascending series outperformed descending series in accuracy and reliability (p < 0.001), though descending series were 24-33% faster. The ADJ method was significantly faster than the LMT method (≤4 vs. ≥8 min for three trials) and preferred by participants. Two trials provided reasonable estimates, while a third offered marginal improvement. The qCSF method offered full CSF estimation but required ≥75 trials (~5 min) for stable precision and reliability (ICC ≈ 0.85 for the area under the log CSF). Pronounced learning effects in the 4-AFC method reduced reliability at low spatial frequencies (ICC < 0.50 at 1.5 cycles per degree), whereas subjective methods showed minimal learning.

Conclusions: Ascending series should be preferred over descending series for accuracy and reliability. The ADJ method offers speed and simplicity for clinical use, while the LMT method is less efficient if linear grating orientations are used and is more demanding. The qCSF method is best suited for research, providing comprehensive CSF characterisation when ≥75 trials and practice sessions are included.