Sixteen chronic stroke patients with homonymous hemianopia demonstrated significant improvements in motion discrimination and visual field boundaries following a novel brain stimulation protocol that synchronized oscillatory activity between the primary visual cortex (V1) and motion-sensitive medio-temporal area (MT), according to research published in Brain.
In the double-blind, randomized, cross-over trial, Estelle Raffin and Michele Bevilacqua of the Neuro-X Institute, École Polytechnique Fédérale de Lausanne in Geneva, Switzerland and colleagues investigated whether pathway-specific, cross-frequency transcranial alternating current stimulation (cf-tACS) could enhance visual rehabilitation outcomes when combined with perceptual training. Each patient received both forward cf-tACS (alpha-tACS over V1 and gamma-tACS over MT) and backward cf-tACS (gamma-tACS over V1 and alpha-tACS over MT) as a control condition in counterbalanced order. They completed 2 blocks of 10 daily training sessions separated by at least 1 month. One patient discontinued after the first block, and one subacute patient was excluded after baseline due to complete spontaneous recovery. Fifteen patients completed forward cf-tACS and 14 completed backward cf-tACS.
Kinetic Perimetry Shows Field Expansion
Forward cf-tACS produced greater improvements in kinetic visual field boundaries compared with backward cf-tACS . In most patients, the visual field extension corresponded to the trained retinotopic area, which confirmed spatial specificity of the training-induced improvements.
Static Humphrey perimetry revealed localized increases in visual sensitivity for both conditions , though the difference between conditions did not reach significance. Significant areas of improvement were defined as visual field locations that differed by at least 6 dB, representing twice the measurement error of the Humphrey test.
Motion Discrimination Learning Curves Diverge
The analysis of normalized direction range (NDR) thresholds—a measure of motion direction discrimination ability—demonstrated a Training Day × tACS condition interaction, with forward cf-tACS showing superior improvements from Training Day 6 onward. Both conditions improved NDR thresholds from baseline, but baseline performance did not differ between groups.
Patient performance began near chance level (54.3 ± 4.2% correct) at the start of training. The researchers used a 3:1 staircase procedure to adjust task difficulty, and NDR thresholds were calculated from Weibull psychometric functions using a 72% correct criterion.
These improvements were achieved after only 10 daily sessions over approximately 2 weeks, suggesting a several-fold faster rate of improvement compared with previous vision restoration protocols that typically require months of intensive training.
Phase-Amplitude Coupling Reveals Dynamic Network Changes
Source-reconstructed EEG data showed distinct temporal patterns of phase-amplitude coupling (PAC) between V1 and MT following forward cf-tACS. V1 alpha phase–MT gamma amplitude PAC increased at approximately 100 ms after stimulus onset, followed by a decrease at approximately 300 ms. The reverse direction (V1 gamma amplitude–MT alpha phase) showed a late increase at approximately 300 ms. Backward cf-tACS produced no PAC changes in either direction.
PAC was quantified to measure how gamma oscillations are temporally structured by alpha phase. This synchronization mode is hypothesized to enable efficient gating and filtering of sensory information, which is particularly relevant for visual retraining protocols where strong anticipatory processing of spatial attention is expected.
fMRI Reveals Network-Level Reactivation
Whole-brain fMRI analysis using a full factorial design (factors: time [pre/post] and tACS condition [forward/backward]) demonstrated a main effect of time in V1, in which both conditions showed increased BOLD activity postintervention. The Time × tACS condition interaction revealed activation in ipsilesional MT, bilateral frontal eye field, and ipsilateral prefrontal cortex. Beta weight extraction confirmed increased MT activity after forward cf-tACS and decreased activity after backward cf-tACS.
Structural and Functional Predictors Identified
A backward multiple regression model incorporating baseline predictors explained 72% of variance in NDR threshold improvements. The predictor analysis was conducted in 11 patients. One data set was excluded due to incorrect targeting, and 4 patients did not participate in this optional substudy.
Two independent variables emerged as key predictors: TMS-evoked fMRI activity in perilesional V1 and sum of weighted V1–MT diffusion tractography streamlines. These predictors showed no collinearity, and lesion volume did not contribute to the model.
The TMS-fMRI protocol used high-intensity vs low-intensity stimulation to probe perilesional V1 reactivity. Group-level analysis revealed activation in perilesional V1 and remote clusters in bilateral frontal eye field and cuneus. A covariate analysis showed that regions in ipsilesional V1 and contralesional frontal eye field increased with improved motion discrimination.
Diffusion-Weighted Imaging Confirms Pathway Integrity
Whole-brain probabilistic tractography using second-order integration over fiber orientation distribution (iFOD2) generated 10 million streamlines per patient, which were weighted using spherical deconvolution-informed filtering of tractograms (SIFT2). V1 and MT regions of interest were derived from individual functional localizers using thresholded activation maps.
Analysis of structural connectivity revealed a trend toward hemispheric differences in fractional anisotropy that reflected structural reorganization and loss of fibers in the lesioned hemisphere. The sum of weighted ipsilesional V1-MT tracts served as an indicator of cross-sectional tract area, while fractional anisotropy indexed structural integrity.
Patient Demographics and Study Design
The study enrolled 16 adult patients who were a median of 60 years old and were a median of 11 months since their last stroke. Twelve patients presented with homonymous hemianopia and 4 presented with homonymous quadrantanopia. Fifteen patients had posterior cerebral artery territory infarctions that caused occipital cortex lesions, while 1 patient had a carotid artery rupture. All but 1 patient had ipsilesional MT spared by the lesion.
Each 10-session block involved daily cf-tACS (3 mA constant current intensity, 0.18 mA/cm² current density) concurrent with coarse direction discrimination and integration task performance. Individual alpha (8 to 12 Hz) and gamma (30 to 45 Hz) peak frequencies were extracted from 5-minute resting-state EEG recordings. Training stimuli consisted of random dot patterns that were presented at individually mapped blind-field border locations where performance declined to 50% to 60% correct.
Clinical Trial Registration
This study was registered at ClinicalTrials.gov (NCT05220449), approved by the local Swiss Ethics Committee (2017-01761), and performed within Declaration of Helsinki guidelines.
Source: Brain