Inter-session Transfer Learning in MI Based BCI for Controlling a Lower-Limb Exoskeleton

• Laura Ferrero ^[1] ; Vicente Quiles ^[1] ; Mario Ortiz ^[1] ; Javier V. Juan ^[1] ; Eduardo Iáñez ^[1] ; José M. Azorín ^[1]
  1. [1] Miguel Hernández University of Elche, Spain
• Localización: Bio-inspired Systems and Applications: from Robotics to Ambient Intelligence: 9th International Work-Conference on the Interplay Between Natural and Artificial Computation, IWINAC 2022, Puerto de la Cruz, Tenerife, Spain, May 31 – June 3, 2022, Proceedings, Part II / José Manuel Ferrández Vicente (dir. congr.), José Ramón Álvarez Sánchez (dir. congr.), Félix de la Paz López (dir. congr.), Hojjat Adeli (aut.), 2022, ISBN 978-3-031-06527-9, págs. 243-252
• Idioma: inglés
• Texto completo no disponible (Saber más ...)
• Resumen

  • Motor imagery (MI) brain-computer interfaces (BCI) have a critical function in the neurological rehabilitation of people with motor impairment. BCI are systems that employ brain activity to control any external device and MI is a commonly used control paradigm based on the imagination of a movement without executing it. The main limitation of these systems is the time necessary for their calibration, before using them for rehabilitation. A shorter calibration scheme was proposed for a lower-limb MI based BCI for controlling an exoskeleton. Each subject participated in 5 experimental sessions. Before each session with the exoskeleton, users were guided to perform MI with visual feedback in a virtual reality scenario. Training with virtual reality involves less physical effort and users can have a previous practise on the MI mental task. In addition, transfer learning was employed, so information from previous training sessions was used for the new one. Results showed that the performance of BCI was superior in comparison with baseline methodologies when transfer learning was used.