Embedded EMG-Controlled Gamification System for Rehabilitation of Hand Function after Stroke
Authors
Lau, Signe Meiners ; Garner, Natasja Toftgaard
Term
4. term
Publication year
2026
Abstract
Stroke often leads to impaired hand and arm function that limits daily activities. Intensive, task-oriented practice is effective, and game-based training can boost motivation. EMG-controlled systems can support adherence, but setup and usability remain barriers. NeuraLoop, a wearable high-density surface EMG armband, removes manual electrode placement and enables real-time recording of forearm muscle activity. This thesis aimed to develop a microcontroller-based EMG-controlled gamification system with NeuraLoop for rehabilitation of hand function after stroke. A qualitative interview with a therapist was used to align the design with patients’ functional needs and cognitive abilities. A threshold-based algorithm for gesture detection was implemented on the NeuraLoop microcontroller, and detected gestures were mapped to rocket actions in a simple game. Microcontroller state transitions matched expectations (accuracy = 100%), overall gesture classification was reliable (accuracy = 73%), and the mean reaction time from gesture completion (−6.36 ± 194.51 ms) was within 300 ms. The work demonstrates the feasibility of an embedded EMG-controlled system for interactive, game-based hand rehabilitation.
Apopleksi medfører ofte nedsat hånd- og armfunktion, som begrænser daglige aktiviteter. Intensiv, opgaveorienteret træning er effektiv, og spilbaseret træning kan øge motivationen. EMG-styrede systemer kan støtte vedholdelse, men opsætning og brugervenlighed er udfordringer. NeuraLoop, et bærbart high-density overflade-EMG armbånd, fjerner manuel elektrodeplacering og muliggør realtidsmåling af underarmens muskelaktivitet. Dette speciale havde til formål at udvikle et mikrocontrollerbaseret EMG-styret gamification-system med NeuraLoop til rehabilitering af håndfunktion efter apopleksi. Et kvalitativt interview med en terapeut blev brugt til at sikre, at designet matchede patienters funktionelle behov og kognitive forudsætninger. En tærskelbaseret algoritme til gestusdetektion blev implementeret på NeuraLoops mikrocontroller, og registrerede gestus blev koblet til raket-handlinger i et simpelt spil. Systemets tilstandsovergange på mikrocontrolleren svarede fuldt ud til forventningerne (accuracy = 100 %), den overordnede gestusklassifikation var pålidelig (accuracy = 73 %), og den gennemsnitlige reaktionstid målt fra gestusafslutning (−6,36 ± 194,51 ms) lå inden for 300 ms. Arbejdet demonstrerer gennemførligheden af et indlejret EMG-styret system til interaktiv, spilbaseret håndrehabilitering.
[This abstract has been generated with the help of AI directly from the project full text]