NCBS Lab

School of Electrical Engineering and Computer Science

Kyungpook National University



 Grant

Role

 Project Title

 Period

National Research Foundation

KRW 488,537,000

(450,000 USD approx.)

(In progress)

Principal Investigator

EEG based Brain Mapping Technology for Medication Response in ADHD 1 Mar 2017 - 28 Feb 2020
Korean Research Foundation,

Brain Korea 21 (BK21+)

KRW 80,000,000/yr

(In progress)

Co-Investigator Lab Funding Mar 2013 -

Feb 2020

National Research Foundation

KRW 125,000,000

Principal Investigator

Robotic based Microsurgical Instrument 1 Apr 2014 - 31 Mar 2017
National Research Foundation

KRW 180,000,000

Principal Investigator

Tremor Compensation Technology for Microsurgery Applications 1 Sep 2011 - 31 Aug 2014
KNU Global 100 Project

KRW 10,000,000

Principal Investigator

Real-time Time-frequency decomposition of EEG rhythms for Brain Computer interface (BCI) Applications 1 Sep 2011 - 31 Aug 2012
DGIST Grant

KRW 30,000,000

Collaborator Fault Reconstruction and Estimation in Electric Vehicles 1 Jun 2011 - 31 Dec 2011
Korean Research Foundation,

Brain Korea 21 (BK21)

KRW 30,000,000/yr 

Principal Investigator

Nonlinear Control and Bio Signal Processing Lab funding 1 OCT 2009 - Present
KNU-IT Convergence, Grant
KRW 50,000,000
(Completed)

Principal Investigator

Tremor Cancellation Technology for Surgical Robotics Applications
1 Aug 2010 – 31 March 2011
KNU Global 100 Project

KRW 10,000,000

(Completed)

Principal Investigator

Fault-detection and Isolation with Sliding Mode Observers 1 Sep 2009 - 31 Aug 2010
KNU Starup Grant

KRW 9,000,000

(Completed)

Principal Investigator

Robust sliding mode observers for state and unknown input estimations 01 Apr 2009 - 31 Mar 2010 

(Completed)

1 US $ (approx) = 1000 KRW.


Brain Signal Processing


EEG Classification for BCI Applications

The amplitude of EEG mu-rhythm varies when the subject is not moving or not imagining and attenuates when the subject is moving or imagines movement. The classification of events is generally performed in the frequency domain for a fixed band to compute the band power. Identification of subject-specific reactive band (optimal band) is paramount for accurate event classification. This work aims to develop a new time-frequency decomposition method for EEG rhythm by estimation of bandlimited signals through adaptive filtering. Multiple fourier series based filltering is adopted to estimate the individual components of frequency weights through LMS algorithm. The knowledge of individual frequency components in time-domain provides useful insight into the classification process of EEG. Instead of using the total band-power, this project analyzes the usage of individual frequency components to determine the optimal band for a subject. Study is conducted on 30 subjects for optimal band selection and event classification. Results show that over 93% the subjects have an optimal band and selection of this band improves the average power difference by a minimum of 70-110%.  This will improve the classification accuracy for BCI systems.

Brain Functional Network Analysis

Understanding the relation between structure and function of the brain is one of the basic questions of neuroscience. Although a large body of knowledge about both healthy and pathological brain structure and function has been gathered over the last decades, we still have a poor understanding of their exact relationship. Over the last decade, due to the development and interdisciplinary combination of techniques and methods, network analysis applied to biological research fields such as immunology, genetics and neuroscience has taken a great flight. A novel approach, applying concepts from graph theory (a branch of the mathematical field of complex network theory) to neurophysiological data, is a promising new way to characterize brain activity. A fundamental hypothesis is that cognitive dysfunction can be illustrated and/or explained by a disturbed functional organization. Applied to patient data, this technique might provide more insight in the pathophysiological processes underlying the various forms of ADHD and potentially lead to the development of new diagnostic or monitoring tools with wide range applicability.

A novel approach applying concepts from graph theory (a branch of the mathematical field of complex network theory) to neurophysiological data, is a promising new way to characterize brain activity. Graph theory provides a method to study the relation between network structure and function, concerning for example qualities such as network efficiency, robustness, cost, or growth.  Both anatomical and functional brain networks can be described by forming graphical network representations based on the measured (functional) connections.   It provides a method to evaluate whether the functional connectivity patterns between brain areas resemble the organization of theoretically efficient, flexible or robust networks (based on the strength of synchronization in the oscillatory electromagnetic activity of different brain regions as measured by EEG or MEG). These network quantification measures will determine the brain functional changes during the given period.


Bio-Signal Processing


Tremor filtering for Microsurgery

Our research interests lies in the area of signal processing/nonlinear systems with focus on adaptive filtering of tremor motion for active tremor compensation in surgical robotics applications. This project is conducted in collaboration with Biorobotics Group, Nanyang Technological University, Singapore.  The project is funded by National Research Foundation of Korea.

Physiological tremor is the main cause for human imprecision in microsurgery procedures. Humans have intrinsic limitations in manual positioning accuracy. These limitations are consequences of small involuntary movements that are inherent in normal hand motion. The most studied involuntary hand movement is physiological tremor, which is found to be about 50-100 µm rms under surgical conditions and in the frequency band of 6-14 Hz.

 

The main objective is to develop tremor cancellation technology that would distinguish between undesired and intended motion, and compensates the undesired motion accurately in real-time for bio-robotics applications. The project aims to develop an improved real-time tremor filtering methods/technology that would achieve 60-70% final tremor cancellation in handheld instruments. The  layout of the scheme is shown in the above figure.

 Our focus has been on designing algorithms for estimation/filtering of physiological tremor for surgical robotics applications. A new bandlimited multiple Fourier linear combiner (BMFLC) for estimation of unknown signals that lie within a specific band is developed.                                                  

Also to improve the performance of the multiple Fourier linear combiner, a double adaptive multiple Fourier combiner for accurate tremor estimation is also developed. Several new algorithms are being developed for real-time estimation and filtering. The developed methods are also applicable for automatic stabilization of images/videos in cameras and camcorders. Proposed algorithms will also be applied for optical image stabilization for robot stereo vision applications.

Prediction of Respiratory motion

for Radiotherapy Applications

RPM data   --> 1) Accelerometers 2)  IR imaging (VICON system) 3) Thermistor (Nasal Cannula Sensor)

Cardiac Motion --->  1) ECG (Biopac) 2) IR imaging 


Nonlinear Systems and Control


Focus of research is on estimation of states together with unknown inputs/disturbances through the new techniques developed via sliding mode theory. Concepts of discrete-time sliding modes for nonlinear systems  and the problems of state and unknown input estimations for nonlinear systems will be explored.  Several new estimation and control algorithms are being developed. The focus of current research is on establishing discrete-time sliding mode and to develop faults detection in nonlinear systems using sliding mode observers.

Sensor-less Speed Estimation  

Our current research is on developing new methods for sensor-less speed estimation using sliding mode control. The DSP for controller implementation is a TMS320F2812 micro-processor.               

Fault Reconstruction and Estimation

Our focus is on developing new methods for accurate fault estimation and reconstruction in Electric vehicles using sliding mode observers.