Voluntary and FES-induced finger movement estimation using muscle deformation features

Yu Zhou, Jia Zeng, Kairu Li, Yinfeng Fang, Honghai Liu

Research output: Contribution to journalArticlepeer-review


It is practically crucial to estimate dexterous hand movement for applications such as prosthetics and rehabilitation, especially scenarios actively assisted with functional electrical stimulation (FES). This paper attempts to estimate hand finger movement through forearm muscle deformation for both clinical scenarios with voluntary and FES-induced muscle contractions based on A-mode ultrasound sensing approaches. Six healthy subjects participated in the experiment of middle finger movement. In order to evaluate the sensing approach, a least-squares support-vector-machine (LS-SVM) model and a linear-ridge (LR) model were developed to estimate the hand finger movement for datasets captured from the flexor digitorum superficialis (FDS) muscle belly, extensor digitorum muscle belly with or without FES. The average results of using FDS datasets as inputs revealed excellent evaluation performance (R2 > 0.9 and RMSE < 10%) both for LR and LS-SVM regression technologies. Furthermore, finger-movements-related muscle deformation features corresponding to voluntary and FES-induced contraction were proved to be different. These results demonstrated the potential application of the forearm muscle deformation in wearable A-mode ultrasound-based prosthesis finger movements control and optimizing the closed-loop FES system for motor function rehabilitation.
Original languageEnglish
Pages (from-to)4002-4012
Number of pages11
JournalIEEE Transactions on Industrial Electronics
Issue number5
Early online date7 Jun 2019
Publication statusPublished - 1 May 2020


  • A-mode ultrasound
  • functional electrical stimulation (FES)
  • finger position
  • linear-ridge (LR) regression
  • least squares support vector machine (LS-SVM)
  • muscle deformation


Dive into the research topics of 'Voluntary and FES-induced finger movement estimation using muscle deformation features'. Together they form a unique fingerprint.

Cite this