Title:

A Deep Learning Approach for Brain ActivityPattern Classification Using fNIRS Data

Functional near-infrared spectroscopy (fNIRS) is a non-invasive, low-cost method used to study the changes in hemoglobin (Hb) concentrations within the brain in response to neural activity. Such brain activity patterns are widely used by Brain-Computer Interfaces (BCIs) to classify tasks performed by a user. A BCI is a system that can convert acquired brain signals into commands that can control output devices to carry out the desired action. BCIs are commonly found in assistive devices, which facilitate restoring the movement ability for physically challenged users. Therefore, accurate brain signals classification is of great importance to determine the user's intention in BCI applications. Most existing task classification systems use conventional machine learning algorithms. Despite the ease of implementation, these traditional methods usually suffer from low accuracy and are time-consuming due to extensive pre-processing. The purpose of this study is to introduce a deep learning approach for task classification using fNIRS data. The tasks considered for this study include mental arithmetic, motor imagery, and idle state. The proposed model can improve the accuracy of task classification while minimizing the pre-processing time.

Name

Comment*

Comments are viewable only by submitter

Captcha*