Gaze Trajectory Index (GTI): A novel metric to quantify saccade trajectory deviation using eye tracking. (April 2019)

• Record Type: Journal Article
• Authors: Cercenelli, Laura; Tiberi, Guido; Bortolani, Barbara; Giannaccare, Giuseppe; Fresina, Michela; Campos, Emilio; Marcelli, Emanuela
• Is Part Of: Computers in biology and medicine. Volume 107(2019)
• Page Start: 86
• Page End: 96
• Publication Date: 2019-04
• Subjects: Eye tracking -- Saccades -- Gaze trajectories -- Saccade curvature -- Metrics; Medicine -- Data processing -- Periodicals; Biology -- Data processing -- Periodicals; 610.285
• DOI: 10.1016/j.compbiomed.2019.02.003
• Languages: English
• ISSNs: 0010-4825

Abstract

Abstract: Background: Many different indexes have been proposed to quantify saccade curvature based on geometric properties of the saccade trajectory projected on the 2D plane. We introduce the Gaze Trajectory Index (GTI), a novel metric to quantify saccade trajectory deviation based on calculation of the rotational eye movements performed in 3D space while following a 2D saccade trajectory recorded with eye tracking (ET). Methods: We provided a description of GTI calculation. In 13 subjects with normal binocular vision we assessed GTI in single-target tests, then we evaluated GTI against previously proposed metrics (Maximum Deviation, MD; Area Curvature, AC; Quadratic Curvature, QC; Initial Direction, ID) using a distractor paradigm that elicited two types of saccade deviations, i.e."inner-curved" and "outer-curved" saccades. Results: In single-target tests GTI showed that saccade curvature was significantly higher for oblique than for vertical saccades (0.86°±0.32 vs 0.55°±0.60, p < 0.05) and higher for vertical than for horizontal saccades (0.55°±0.60 vs 0.23°±0.17, p < 0.05), in accordance with previous studies. In distractor-based tests, for inner-curved saccades, GTI strongly correlated with MD (r = 0.965, p < 0.01), AC (r = 0.940, p < 0.01), QC (r = 0.866, p < 0.01), and Principal Component Analysis (PCA) confirmed that all these metrics reflect the same underlying phenomenon. For outer-curved trajectories, GTI showed poor correlation with MD and AC (r = 0.291 and 0.416, p < 0.01), however PCA included the three metrics in the same first component group. For outer-curved trajectories, GTI was the only metric showing strong correlation (r = 0.950, p < 0.05) with the overshoot degree of the trajectory. Conclusion: The novel GTI seems to have adjunctive potential, particularly for outer-curved trajectories, in the estimation of the absolute amount of saccade trajectory deviation. Highlights: A novel metric (Gaze Trajectory Index, GTI) for quantification of saccade trajectory deviation is presented. GTI is based on the assessment of rotational eye movements performed in 3D space while performing a 2D saccade trajectory. The novel GTI has great potential for measuring trajectory deviation in case of saccades with "outer-curved" patterns.