Deep Dictionaries for Feature Extraction in Context of Sparse Data for Electroencephalographic Signals from Brain-computer Interfaces
Deep Dictionaries for Feature Extraction in Context of Sparse Data for Electroencephalographic Signals from Brain-computer Interfaces
The human brain, as a processing center, controls bodily, cognitive, emotional and social functions, enabling perception, signal analysis and decision making. However, these functions can be affected by acquired brain injury (ABI), resulting from traumatic (blows to the head) or non-traumatic factors (tumors, strokes, infections, among others). Annually, about 55 million new cases of ABI are reported, with sequelae that can affect the quality of life of patients and their families. This scenario has driven research into tools to mitigate and recover lost capabilities. The Center for Rehabilitation Engineering and Neuromuscular and Sensory Research (CIRINS) of the Faculty of Engineering of the National University of Entre Ríos in Argentina has developed neuromuscular and sensory rehabilitation systems, with a focus on the innovation of motor rehabilitation tools using EEG-based brain-computer interfaces (BCI). These BCIs stand out for their economy and versatility, showing significant effects in the rehabilitation of motor functions. Challenges in BCI include signal complexity, artifacts, and inter-person variability, making it difficult to estimate user intent and extending calibration time. To mitigate these problems, strategies based on Deep Learning and dictionary learning have been proposed, which allow for sparse representations of data, being robust to noise and missing data, but with challenges in classification. The study proposes to develop a database of electroencephalographic signals applicable in the development of new algorithms for processing and feature extraction of this type of signals, contributing to the development of technology that supports rehabilitation processes.
Inclusion Criteria:
Exclusion Criteria: