Abstract
Purpose :
Individuals with Intermittent Exotropia (X(T)) often exhibit spontaneous and uncertain ocular motility behavior. Particularly, their eye movement patterns while perceiving 3D motion are not fully understood. Identifying visual system regions and underlying eye movement mechanisms by decoding frontoparallel (x & y) and depth (z) positions from fMRI response to a continuously moving 3D target would contribute to a deeper understanding of the neurological basis of X(T).
Methods :
We measured brain activity using fMRI in 6 X(T) & 7 age-matched healthy controls (HC) while they gazed at a target moving continuously in a 3D random walk. First, we binned target positions to match the time points (TP) in the fMRI signal. Next, we clustered (k-means) these positions to 10 major locations in the stimuli space. Then, we decoded these positions from fMRI (deconvolved) responses using one-vs-one Support Vector Classifiers. Positions were decoded for a range of 0 to 5 TP (lags) to account for delays in neural response at different levels of the visual system.
Results :
We found activations in several brain regions in response to 3D motion (using GLMsingle toolbox) – Right: R-V1, R-V2, R-FST/R-MT+, R-FEF (Frontal Eye Field), R-PEF (Precentral Eye Field), R-IP (Intraparietal) & R-TPO (Temporo-Parieto-Occipital junction), Left: L-SMA (Supplementary Motor Area), L-V3B, L-V4 and Bilateral: B-V3A, B-VIP (Ventral IP), B-PMd (Premotor-dorsal). In HC, x, y & z positions were decoded with above-chance (10 locations ~ 10%; one-sample t-test: p<0.05) accuracy in R-V3A, R-PEF, L-V4 & L-SMA for all lags. Particularly, z-positions were significantly decoded from R-FST/MT+ at a lag value of 3, affirming that depth-specific decoding is slower than that for frontoparallel positions (Accuracy – z>x: p=0.032; z>y: p=0.034). However, in X(T) we did not observe similar trends in aforementioned regions, except for R-V1, R-IP, B-V3A, R-TPO, L-SMA, R-PEF, B-PMd & B-VIP that may still be involved in processing frontoparallel motion alone. Specifically, we noticed poor decoding performance in R-FST/MT+, L-V4, L-VIP, L-V3B, R-FEF in X(T) indicating their inability to maintain stable fixation in response to the depth component of 3D Brownian motion.
Conclusions :
Our results show possibly compromised neural mechanisms for controlling eye movements to 3D motion in X(T) and understanding them may be useful for targeted interventions.
This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.