05May 2020 — Jun 2023Lead researcher · IIT Madras
Automated Movement Evaluation System
Low-cost, computer-vision tools for evaluating a subject's motor performance — no specialized hardware required — validated on healthy subjects and several classes of patients.
MediaPipeOpenCVLSTMFlip-flop neurons
[ pose-estimation overlay / scoring demo ]
// approach
→Use computer-vision techniques to capture a subject's movement from low-cost cameras.
→Score motor performance with LSTM and flip-flop-neuron models from the pose estimates.
→Validate the system on healthy subjects and on different classes of patients.
→Package it to evaluate, monitor, and track improvements across rehabilitation.
// outcomes
●Automates the Fugl-Meyer motor assessment without clinician scoring overhead.
●Makes objective motor evaluation affordable enough for home and clinic.
// related publications
Automation of Fugl-Meyer Assessment
13th World Stroke Congress (WSO) · 2021
// related patents
FILED
A Method and System for Automated Assessment of Upper-Limb Motor Impairment
App. No. 2443 · under review