Accelerometric decoding of upper extremity movement intention in children with and without cerebral palsy:
Marcela Correa1, Ali Samadani1, Ledycnarf J de Holanda1
1Bloorview Research Institute, Holland Bloorview Kids Rehabilitation Hospital, 150 Kilgour Road, Toronto M4G 1R8, Ontario, Canada; Institute of Biomedical Engineering, University of Toronto, 164 College Street, Toronto M5S 3G9, Ontario, Canada.
Insights
Researchers explored using muscle activity (mechanomyography) and motion sensors to decode movement intentions in children. Combining these signals significantly improved accuracy, offering potential for new assistive technologies for those with movement disorders.
Area of Science:
- Biomedical Engineering
- Neuroscience
- Rehabilitation Technology
Background:
- Children with movement disorders like cerebral palsy (CP) exhibit unpredictable upper extremity movements, complicating interaction and assistive technology use.
- Decoding movement intention in these children is challenging due to motor variability, such as athetosis and dystonia.
- Previous accelerometry studies for hyperkinetic movement analysis in CP have limitations in intention decoding.
Purpose of the Study:
- To investigate the potential of mechanomyography (MMG) and motion signals for detecting movement intention in children.
- To evaluate the effectiveness of integrating MMG with limb motion data for improved intention recognition.
- To assess the feasibility of this combined approach for children with and without movement disorders.
Main Methods:
- A multiple case study involving fifteen typically developing (TD) children and three children with CP.
- Placement of six tri-axial accelerometers on specific upper extremity muscles (biceps brachii, triceps brachii, flexor carpi radialis, extensor carpi radialis longus, anterior deltoid, pectoralis major).
- Utilized a tablet-based drawing task with discrete, self-paced linear motions and Hidden Markov Model (HMM) based classification.
Main Results:
- Integrating MMG activity with motion signals enhanced movement intention recognition accuracy.
- The combined approach achieved 88.0 ± 5.6% quaternary classification accuracy in TD children.
- Accuracies exceeding 70% for differentiating movement pairs were observed in all three CP participants using combined signals.
Conclusions:
- Mechanomyography holds significant potential for decoding movement intention in children, including those with movement disorders.
- Combining MMG and motion data offers a promising avenue for developing more effective assistive technologies.
- This approach could lead to improved human-computer interaction and environmental control for individuals with motor impairments.
Abstract:
Children with movement disorders often interact with their environment via targeted upper extremity movement. However, due to athetosis or dystonia, these gestural intentions can become variable and unpredictable. While accelerometry has been used for hyperkinetic movement analysis in cerebral palsy (CP), decoding movement intention remains challenging due to variability. In this multiple case study, we investigated the potential of detecting movement intention using six tri-axial accelerometers positioned over the biceps brachii long head, triceps brachii long head, flexor carpi radialis, extensor carpi radialis longus, anterior deltoid, and pectoralis major muscles to capture both limb motion and mechanomyographic (MMG) activity. Fifteen typically developing (TD) children and three children with CP performed a tablet-based drawing task involving discrete, self-paced linear motions. Hidden Markov model-based classification revealed that integrating MMG with motion signals improved recognition accuracy. The combined approach achieved 88.0 ± 5.6% quaternary classification accuracy in TD children. A similar quaternary recognition rate was achieved for one of the participants with movement disorders. Average participant-specific accuracies exceeded 70% for all three participants with CP when considering the differentiation between pairs of movements based on combined motion and MMG signals. These findings underscore the potential of MMG for movement intention decoding, informing assistive technology development for motor-impaired individuals.
