Hardware Analysis of an Asynchronous Brain Computer Interface for Motor Imagery Detection based on Non-Invasive Electroencephalography

  • Ljubomir Radakovic

Student thesis: Master's Thesis


Due to a severe damage of the spinal cord patients have difficulties when performing routine tasks. Injured peripheral nerves are unable to transfer the information from the brain to muscles. Therefore locomotion of the particular body parts becomes distorted. The goal of this thesis is to find a solution to bypass damaged nerves and create a new way of communication. With Brain-Computer Interface (BCI) based on a motor imagery, it is possible to establish easier and more reliable connection between brain and an external device. However, those systems are usually bulky, not comfortable for the patients and unable to continuously monitor patient state outside the laboratory. There is a need to develop a portable hardware device that will record and decode brain signals with non-invasive Electroencephalography (EEG) tools. First of all, this thesis investigates software solutions that apply to the BCI system design on an ASIC hardware platform. Subject specific patterns extraction was performed with two filters (temporal band pass filter and spatial filter based on CSP algorithm) and energy estimation block. Extracted features were classified with LDA algorithm. False detection rate of motor imagery events was increased along with overall accuracy for more than 5% when a postprocessing block was enabled. A second part of the thesis focuses on a hardware analysis of the proposed system and implementation methods. Area and power reduction techniques were applied when designed blocks. It is concluded that feature extraction engine, particularity energy estimation block consumes more than 75% of a total estimated chip area. This study was developed to present a proof of concept for designing an online asynchronous BCI system on a hardware platform. Moreover, this thesis may bring future ideas for developing a wearable device, that is also patient comfortable and can continuously monitor and detect motor imagery events.
Date of AwardDec 2016
Original languageAmerican English
SupervisorJerald Yoo (Supervisor)


  • Spinal Chord Injuries
  • Brain-Computer Interface
  • Wearable Devices
  • Motor Imagery Detection
  • Brain Signals
  • ASIC Hardware.

Cite this