A Charge-Integrated Modular NIR Armband for Forearm Optical Myography

Optical myography is a promising wearable alternative to surface electromyography. This paper presents a modular near-infrared armband for forearm optical myography based on discrete 850 nm LEDs, silicon PIN photodiodes, and charge-integration digitization. The device performs a complete emitter-detector sweep to obtain a multi-channel measurement map at 5 Hz, with distance-dependent integration and illumination timing to extend dynamic range while limiting optical cross-talk. A dataset of 50,587 balanced windows from four subjects performing four gestures in two forearm postures across multiple sessions was collected. Mixed-subject evaluation achieved 92.5% ± 4.6% gesture accuracy with near-perfect posture recognition. Few-shot transfer learning mitigated session-to-session placement variability and enabled rapid adaptation with minimal calibration, including for previously unseen users.