Michele Bevilacqua¹, Rebecca Frinco², Meghna Uzgare¹, Dean Thurston³, Peige Wang³, Aaron R. Seitz³, Lorella Battelli¹; ¹Berenson-Allen Center for Noninvasive Brain Stimulation, Department of Neurology, Beth Israel Deaconess Medical Center, Harvard Medical School. Boston, MA, USA, ²Department of Psychology, University of Turin, Torino, Italy, ³Department of Psychology, Northeastern University, Boston, MA

Stroke-related damage to primary visual cortex (V1) is one of the main causes of chronic homonymous hemianopia (“cortical blindness”, CB). Standard rehabilitation mainly relies on compensatory strategies, yet gaze-contingent visual perceptual learning can partially restore blindfield motion processing and expand perimetric sensitivity when training targets the blindfield border. Although effective, this approach often requires prolonged training, limiting compliance. More recently, pairing high-frequency transcranial random-noise stimulation (tRNS) with visual training has been shown to accelerate recovery in CB, likely by increasing cortical excitability and enhancing detection through noise-based (stochastic resonance) mechanisms. To extend such interventions to patients without clinic access, we piloted VR-based perimetry and motion-discrimination training paired with tRNS. Eight patients completed immersive, gaze contingent VR training with integrated eye tracking, using the new open science XRPSys platform, to ensure motion stimuli remained anchored to each patient’s blindfield border. Training used a coarse direction discrimination and integration task that adaptively varied motion coherence (normalized direction range, NDR) to estimate thresholds. Occipital tRNS (110-640 Hz) was delivered during training blocks. Thresholds were tracked across 10 days (Day1-Day10), with performance in each patient's intact visual field as an internal control. Visual-field extent was assessed pre- and post-training using VR-based perimetry, alongside Humphrey perimetry and contrast sensitivity (qCSF). Group-level blindfield performance improved by mid-training. Paired Wilcoxon tests showed significant baseline-to-day6 improvements in NDR (p=0.031), while the intact hemifield showed no significant change (p≥0.148). VR perimetry showed a mean seeingfield expansion of +5.73% (range -9.5% to +13.3%; 6/8 improved, 1 unchanged, 1 decreased). Precisely targeted VR-based perceptual learning paired with occipital tRNS produced concurrent gains in motion processing and measurable increases in visual-field extent, consistent with retained plastic potential at the blindfield border. This VR+tRNS approach is feasible in chronic hemianopia and yields promising functional and perimetric gains, supporting its potential for home-based rehabilitation.

Acknowledgements: This work was supported by the NC NM4R 2024-2025 Pilot Project Grant (NIH Grant UM1TR004406 for NC TraCS) and by NEI-R61EY037510-01.