Tractography-guided EMG Biofeedback in Stroke: An Exploratory Study of Plasticity and Motor Outcomes

Background: Upper limb dysfunction in stroke reflects layered impairments — from impaired motor execution to maladaptive plasticity and disrupted white matter architecture. Unlike the more resilient recovery of lower limbs, fine motor restoration remains limited. Although electromyographic (EMG) biofeedback is clinically validated, its influence on central structural substrates of motor recovery remains poorly understood. Objectives: This exploratory feasibility study examined EMG biofeedback as a potential neuromodulatory input influencing white matter architecture, assessed with diffusion tensor imaging (DTI). The findings are hypothesis-generating and provide preliminary indications rather than definitive evidence of structural change. Methods: Twelve individuals (aged 50 - 70) with ischemic stroke (6 - 22 months post-onset) and upper limb spasticity [Modified Ashworth Scale (MAS) ≥ 3] underwent an 8-week EMG biofeedback intervention targeting distal extensor activation. In this single-arm, exploratory assessment, functional outcomes were measured using the Fugl-Meyer Assessment for Upper Extremity (FMA-UE), MAS, and Barthel Index. Microstructural plasticity was assessed via DTI, with extraction of fractional anisotropy (FA), mean diffusivity (MD), and radial diffusivity (RD) across motor pathways. Due to the limited sample size and exploratory nature of the study, we employed both frequentist and Bayesian methods to assess pre-post changes and brain-behavior associations, allowing for prior-informed estimation and improved uncertainty characterization. Results: Motor performance showed improvement (FMA-UE: +12.4; P < 0.01), accompanied by increases in EMG tension amplitudes (P < 0.001), consistent with enhanced motor unit activation. The DTI analyses indicated a marginal FA increase in the corticospinal tract [CST, P = 0.05; posterior mean: 0.045 (CI: 0.0039 - 0.0877)] and an RD increase in the internal capsule [P < 0.01; posterior mean: 0.027 (CI: -0.0034 - 0.0584)], which appeared to correlate with motor gains. By contrast, functional independence measured by the Barthel Index showed no clear improvements in this cohort. Conclusions: The current preliminary findings suggest that EMG biofeedback may modulate motor function and corticospinal microstructure in stroke. Observed clinical, electrophysiological, and imaging trends warrant cautious interpretation given the pilot design. Larger, controlled studies are required to validate efficacy and clarify underlying neuroplastic mechanisms.