Machine learning for muscle activation patterns identification during cyclical movements

Topic:

Cyclical movements are characterized by different muscle activation patterns (onset-offset timings). To interpret correctly the electromyographic (EMG) data, it is important to group cycles sharing similar EMG activation patterns. We developed a method based on hierarchical clustering, able to group cycles showing homogeneous onset-offset activation intervals.

The method was applied to EMG acquired during gait. A by-product of the clustering procedure is the possibility to extract the principal activations of a muscle, that are those activations necessary for the specific muscle contribution to the biomechanical function of walking. Principal activations were also used to define a robust index to quantitative assess the asymmetry of muscle activations during locomotion.

People:

  • Gabriella Balestra (Confirmed Assistant Professor)
  • Samanta Rosati (Assistant Professor with time contract (RTD/b))
  • Gregorio Dotti (Ph.D. Student)
  • Marco Ghislieri (Assistant Professor with time contract (RTDa))
  • Valentina Agostini (Associate Professor)
  • Marco Knaflitz (Full Professor)

    • Recent publications:

  • Muscle activation patterns during gait: A hierarchical clustering analysis
      S. Rosati, V. Agostini, M. Knaflitz and G. Balestra
      Biomedical Signal Processing and Control
      10.1016/j.bspc.2016.09.017
  • Asymmetry Index in Muscle Activations
      Castagneri, C., Agostini, V., Rosati, S., Balestra, G., Knaflitz, M.
      IEEE Transactions on Neural Systems and Rehabilitation Engineering
      10.1109/TNSRE.2019.2903687
  • Evaluation of Muscle Function by Means of a Muscle-Specific and a Global Index
      S. Rosati, M. Ghislieri, G. Dotti, D. Fortunato, V. Agostini, M.Knaflitz, and G. Balestra
      Sensors
      10.3390/s21217186
  • Long short-term memory (LSTM) recurrent neural network for muscle activity detection
      M. Ghislieri, G. L. Cerone, M. Knaflitz, & V. Agostini
      Journal of NeuroEngineering and Rehabilitation
      10.1186/s12984-021-00945-w