REMoDNaV: Robust Eye-Movement Classification for Dynamic Stimulation

Abstract

Tracking of eye movements is an established measurement for many types of experimental paradigms. More complex and more prolonged visual stimuli have made algorithmic approaches to eye movement event classification the most pragmatic option. A recent analysis revealed that many current algorithms are lackluster when it comes to data from viewing dynamic stimuli such as video sequences. Here we present an event classification algorithm-built on an existing velocity-based approach-that is suitable for both static and dynamic stimulation, and is capable of classifying saccades, post-saccadic oscillations, fixations, and smooth pursuit events. We validated classification performance and robustness on three public datasets: 1) manually annotated, trial-based gaze trajectories for viewing static images, moving dots, and short video sequences, 2) lab-quality gaze recordings for a feature length movie, and 3) gaze recordings acquired under suboptimal lighting conditions inside the bore of a magnetic resonance imaging (MRI) scanner for the same full-length movie. We found that the proposed algorithm performs on par or better compared to state-of-the-art alternatives for static stimulation. Moreover, it yields eye movement events with biologically plausible characteristics on prolonged dynamic recordings. Lastly, algorithm performance is robust on data acquired under suboptimal conditions that exhibit a temporally varying noise level. These results indicate that the proposed algorithm is a robust tool with improved classification accuracy across a range of use cases. The algorithm is cross-platform compatible, implemented using the Python programming language, and readily available as free and open source software from public sources.

multimatch-gaze: The MultiMatch algorithm for gaze path comparison in Python

“…Data for scan path comparison can be supplied as nx3 fixation vectors with columns corresponding to x-coordinates, y-coordinates, and duration of the fixation in seconds (as for the original Matlab toolbox). Alternatively, multimatch-gaze can natively read in event detection output produced by REMoDNaV (Dar, Wagner, & Hanke, 2019), a velocity-based eye movement classification algorithm written in Python. For REMoDNaV-based input, users can additionally specify whether smooth pursuit events in the data should be kept in the scan path or discarded.…”

Strategies for enhancing automatic fixation detection in head-mounted eye tracking

“…The principal idea of this I-VT algorithm (Identification by Velocity Threshold) has been further elaborated in numerous studies. For example, variants have included noise-adaptive thresholds (Engbert & Mergenthaler, 2006;van der Lans et al, 2011), or multiple thresholds resolving additional types of eye movements, such as smooth pursuit (Larsson et al, 2015;Komogortsev & Karpov, 2012) and/or post-saccadic oscillations (Nyström & Holmqvist, 2010;Dar et al, 2020).…”

“…First, we used the REMoDNaV algorithm ("Robust Eye Movement Detection for Natural Viewing") (Dar et al, 2020). While REMoDNaV was originally developed for remote eye tracking, it was specifically tailored to enable more robust fixation detection for dynamic video stimuli.…”

“…Second, we coupled corresponding parameters for fixation detection and smooth pursuit detection to the same values. Third, we identified parameters which, after setting plausible initial values, would not alter detection performance on our dataset in explorative experiments (e.g., the maximum initial saccade frequency (Dar et al, 2020)).…”

“…As a first alternative method, we evaluated the fixationdetection algorithm ReMoDNaV (Dar et al, 2020), which has been originally developed for remote eye-tracking data. At its core, ReMoDNaV utilizes an adaptive velocity threshold and has been optimized for the task of noise-robust fixation detection with dynamic stimuli such as video sequences (see Methods).…”