Page 1

International Journal of Computer Science Engineering and Information Technology Research (IJCSEITR) ISSN 2249-6831 Vol. 3, Issue 2, Jun 2013, 19-26 Š TJPRC Pvt. Ltd.

ANALYZING EEG DATA FOR COGNITIVE BEHAVIOUR USING EEGLAB THROUGH DATA MINING PARNEET KAUR1 & SHEVETA VASHISHT2 1 2

Lovely Professional University, Punjab, India

Assistant Professor, Lovely Professional University, Phagwara, Punjab, India

ABSTRACT This paper is a theoretical analysis of human brain and its activity which is measured by EEG. Electroencephalography (EEG) is the process where electrodes are placed on scalp and brains activity is explored by EEG measurements. This measurement is done on basis of any condition like someone listening music, watching any type of video, while studying etc. These different conditions can explore different areas of brain where different information is stored in human brain memory. For an example human brain stores all information in different sections like visual, language related data differently. This data coming from EEG is feed in EEGLAB tool for further analysis and making sure that the particular person on which EEG test is done, is having average cognitive skills or not.

KEYWORDS: Data mining, Electroencephalography, EEGLAB, Database INTRODUCTION Application area of data mining is vast for example data mining is used in banks, hospitals, stores, insurance companies and also in biology. Analysing and storing data about brain in a database and then extracting hidden patterns to make decisions is also an application area of data mining. Some biomedical instruments are used for this work. Electroencephalography EEG is used to measure the cognitive load on human brain. EEG signals are composed of various signals which represent the information of neuronal assemblies. EEG can detect brain wave rhythms and these rhythms are used for sensory registration and tracking, perception, movement and cognitive processes which are related to attention, learning and memory [1]. EEG is identified as a physiological index that can serve as an online, continuous measure of cognitive load detecting subtle fluctuations in instantaneous load, which can help explain effects of instructional interventions when measures of overall cognitive load fail to reflect such differences in cognitive processing.

Figure 1: Human Brain Wave Rhythm (Wave Per Second) [1]

At a sufficient electrical distance from the cortex, e.g., on the scalp surface, the projection of a single cortical patch source strongly resembles the projection of a single cortical dipole termed its equivalent dipole [2], [3]. EEG allows tracking of changes in human brains activity within a reasonable time resolution.


20

Parneet Kaur & Sheveta Vashisht

Data Mining This is the process of extraction of hidden pattern of predictive information from a large database. This is done by data mining tools and this allows business to make proactive and knowledge driven decisions [9]. Nowadays more than 1,000,000,000,000,000 bytes of data can be ranges into some terabytes and within this vast amount of data hidden information of strategic importance lies [10]. Data mining is a process that uses a variety of data analysis tools to discover patterns and relationships in data that may be used to make valid predictions. Data mining is a process that uses a variety of data analysis tools to discover patterns and relationships in data that may be used to make valid predictions. Machine learning represents a sub-field of artificial intelligence and it was conceived in early 60s with the clear objective to design and develop algorithms and techniques that implement various types of learning, mechanisms capable of inducing knowledge from examples of data. Machine learning has a wide spectrum of applications including natural language processing, search engines, medical diagnosis, bio-informatics, speech and handwriting recognition, object recognition in computer vision, game playing and robot locomotion. Cognitive Skills Cognitive skills are human’s mental skills which help in processing and acquiring knowledge. These skills include reasoning, perception and intuition. Mid-continent Research for Education and Learning describes the importance of cognitive skills in acquiring literacy skills [11]. Cognitive skills are those learning skills by which human brain stores information like attentiveness of students in class or understanding behavior of persons. Auditory and visual processing and processing speed all comes under cognitive skills. Cognitive skills include the process of brain of fetching this information from memory where large amount of data is stored. For an example- If in any class if teacher is teaching and the listeners do not find that topic interesting they will obviously not pay attention to the teacher but the words teacher is speaking will strike in their mind and these will be stored in their short term memory which they will forget within minutes or seconds. But when the listeners find any topic interesting they pay attention and their mind keep on practicing and thinking about that topic and therefore that topic will be stored in long term memory. Data Mining Approach to Analyse Cognitive Skills Using EEGLAB Tool The EEGLAB signal processing environment for electrophysiology is being used in hundreds of laboratories around the world. EEGLAB allows convenient sharing, downloading, and re-analysis or meta-analysis of sophisticated human electroencephalography data sets [1]. Data mining is used in many application areas. Data mining can also be used in analyzing the cognitive skills of human brain that is mining or extracting data from human brain where human brain acts as database where metadata is present. Human brain contains large amount of information regarding different aspects like information may be emotional, educational and social or of many more types. By doing classification sections of human memory are created. Different sections stores related information. When extracting this information indexing from outer word to the database is done [4]. EEGLAB is a MATLAB tool which is designed especially for purpose to analyse data related to EEG, ECG and MRI etc.

Figure 2: Overview of Proposed Technique [4]


Analyzing EEG Data for Cognitive Behaviour Using EEGLAB through Data Mining

21

The figure 2 explains how data is collected from electroencephalography and is stored in system. This collected data is further pre-processed and after clustering is classified to explore different classes of brain.

LITERATURE REVIEW Recently interest in the field of human cognitive behaviour is increased so much. These researches give result in a way to improve the cognitive human cognitive behaviour. Rahul isola et al (2011) explored the use of knowledge discovery in medical fields by collecting patient’s data from hospitals. That data contains hidden patterns, and relationships that can lead to better diagnosis and better treatment [5]. This paper focuses on computing the probability of occurrence of ailments and by using algorithms like by combining key points of neural network, large memory storage and retrieval, k-NN and differential diagnosis onto a single algorithm the accuracy of diagnosis can be increased. Terry Peckham et al (2012) uses data mining technique to analyze positive and negative cognitive skills by using reading comprehension tasks. Students were interacted with a learning environment, comprehensions were provided to them and asked questions based on those comprehensions. Time stamp data was processed to calculate reading, scanning and scrolling navigation times. Llyod’s algorithm was used to find students who behave in similar manner for different levels of difficulties [6]. This experiment was conducted using multidimensional k clustering approach combined with Bloom’s Taxonomy. These clusters can be turned to metrics that can be used to discover the strategies the students are using and provide the cognitive skills set. Pavlo Antonenko et al (2010) in their research they gave some case studies which were about to explore the idea of analysing cognitive loads on human brain using EEG machine. According to their research EEG calculates and provides human brain’s activity readings in very less time [1]. Continuous measurement of instantaneous cognitive load allows looking at data for specific instances of time, which will allow a more detailed, and likely more accurate interpretation of the effects of instructional interventions on cognitive load and learning. Scott Makeig et al (2012) explored brain computer interface which uses the knowledge of electroencephalography machine to deliver relevant feedback to user and to increase safety and enhance overall performances. The BCI field is clearly observing an asymptotic trend in the accuracy of EEG-based BCI systems for cognitive state or intent estimation, a performance trend that does not appear to be converging to near-perfect estimation performance but rather to a still significant error rate [7]. For BCI technology to exploit information provided from expert systems modelling brain and behavioural data (such as EEG and motion capture and/or scene recording), in combination with rich if imperfect information about subject environment and intent, some modality independent representations of high-level concepts including human cognitive and affective processing states could be highly useful. Anna M. Bianchi et al (2010) explored that the functional connectivity of the brain is investigated through the study of multivariate autoregressive models (MVAR) applied to multi channel EEG recordings. After the identification of the model, different indices can be calculated that are able to quantify direct and indirect functional connections between cortical areas. These methodologies are used for the investigation of possible connectivity alterations in patients after Traumatic Brain Injury (TBI) who suffered from Diffuse Assonal Injury (DAI) [8]. As one of main consequences of DAI are cognitive and attention impairment, the subjects underwent an attention test (Conners CPT) during EEG recording.


22

Parneet Kaur & Sheveta Vashisht

PROPOSED WORK Proposed work is based on brain computer interface and mapping of information to human brain using data mining techniques. EEG is used to extract information from brain and this data is analysed using EEGLAB tool. Independent component analysis (ICA) algorithm is used as technique to separate independent sources mixed in several sources and data released by EEG is stored in database. Clustering algorithm is used to make clusters of given data. Kmeans clustering is used in this research. K-means clustering is based on defining centroids for each cluster. These clusters are placed as much as possible far away from each other. A loop is generated in which each point in the dataset is examined by comparing to the nearest centroid [12]. Here centroid changes its location step by step until no more changes are done. This algorithm is based on the following object function.

,

where

is a chosen distance measure between a data point

and the cluster centre

, is an indicator of

the distance of the n data points from their respective cluster centres. Different clusters form different classes of human brain like imagination, reasoning, intelligence, emotional, routine work and personal. Finally by applying fuzzy rules on the bases of result of clustering cognitive nature of human can be examined.

PROPOSED METHODOLOGY Data set- Primary data set is collected from doctor. Readings of EEG are the primary datasets and these readings are interpreted by doctor. Figure 3 shows the connectivity of electroencephalography machine. In this figure 01, 02, P1P10, T1-T10, A1-A10 all are the points according to which sensors are placed on human brain. EEG results in alpha, beta and gamma rays. EEG considers frequency of these rays and calculates the frequency per sensor’s activation. Whenever human brains activity starts EEG measures the electrical activity and gives some values.

Figure 3: EEG Connecting Settings


Analyzing EEG Data for Cognitive Behaviour Using EEGLAB through Data Mining

23

Secondary dataset- For secondary dataset Questionnaire is created in which different questions related to each section of brain are included and according to these questions brain’s activity is measured. This data set is in the form of values which are analysed from human brain during doing EEG. Figure 4 represents the total values which are measured by each electrode or sensor. These values are in numeric form. Electrodes represent the sensors on brain. Numeric values represents the time when brains activity occurs.

Figure 4: EEG Data Set

ICA Independent Component Analysis (ICA) is a quite powerful technique and is able to separate independent sources linearly mixed in several sensors. While recording readings of electroencephalograms (EEG) on the scalp, there are mixed signals which can be separated by using ICA algorithm. Independent Component Analysis (ICA) is a signal-processing method to extract independent sources given only observed data that are mixtures of the unknown sources. Recently, blind source separation by ICA has received considerable attention because there are many potential applications of ICA which have brings it attention such as signal processing in speech recognition, telecommunication and medical signal processing [13]. The goal of ICA is to recover independent sources given only sensor observations that are unknown linear mixtures of the unobserved independent source signals. k-means Clustering ICA separates the mixed signals. After applying ICA k-means clustering is used as clustering technique which creates separate clusters for each section. In k-means all objects are represented by numeric features. User itself specifies the number of groups. Number of groups represents number of clusters. This algorithm randomly chooses k points in the vector space these are the centres of the clusters [14]. All other objects are assigned to the cluster which is nearest to it. This looping is done recursively until we get final clusters and values under it.


24

Parneet Kaur & Sheveta Vashisht

Figure 5: Result of Clustering on EEG Data Set

EEGLAB Knowledge of human brain connectivity will transform human neuroscience by providing not only a qualitatively novel class of data, but also by providing the basic framework necessary to synthesize diverse data and, ultimately, elucidate how our brains work in health, illness, youth, and old age. The study of human brain connectivity generally falls under one or more of three categories: structural, functional, and effective (Bullmore and Sporns, 2009) [15]. EEGLAB signal processing environment for electrophysiology is being used in hundreds of laboratories around the world. EEGLAB allows convenient sharing, downloading, and re-analysis or meta-analysis of sophisticated human electroencephalography data sets [16]. EEGLAB is a MATLAB tool which is designed especially for purpose to analyse data related to EEG, ECG and MRI etc.

CONCLUSIONS This research explored the idea to use independent component analysis algorithm for analysing electroencephalography readings and interpreting the different classes of human brain. Different clusters introduce different classes of brain. Clustering is done by using k-means algorithm and a rule base is designed to analyse whether the human brain which is under observation is responding within the threshold or not and If the brain works timely it results in strong cognitive skills and if response time is less than the threshold that results in weak cognitive skills. This whole technology comes under brain computer interface. Brain computer interface relies on the knowledge of EEG.

REFERENCES Researches 1.

Pavlo Antonenko & Fred Paas & Roland Grabner & Tamara van Gog (2010) : Using Electroencephalography to MeasureCognitive Load. Springer Science+Business Media, LLC 2010


25

Analyzing EEG Data for Cognitive Behaviour Using EEGLAB through Data Mining

2.

J. de Munck, B. Van Duk, and H. Spekreijse, BMathematical dipoles are adequate to describe realistic generators of human brain activity,[ IEEE Trans. Biomed. Eng., vol. BME-35, no. 11, pp. 960–966, Nov. 1988.

3.

M. Scherg, BFundamentals of dipole source potential analysis,[ Adv. Audiolog., vol. 6, pp. 40–69, 1990.

4.

Parneet kaur, sheveta vashisht (2012) “Analysing cognitive skills through data mining” Int.J.Computer Technology & Applications,Vol 3 (6), 1959-1962

5.

Rahul Isola, Rebeck Carvalho and Amiya Kumar Tripathy (2011) “Knowledge Discovery in Medical Systems Using Differential Diagnosis, LAMSTAR and k-NN”, IEEE TRANSACTIONS ON INFORMATION TECHNOLOGY IN BIOMEDICINE - TITB- 00346-2011.R4.

6.

Terry Peckham and Gord McCalla, (2012) “Mining Student Behavior Patterns in Reading. Comprehension Tasks”.Scott Makeig, Christian Kothe, Tim Mullen, Nima Bigdely-Shamlo, Zhilin Zhang, and Kenneth Kreutz-

7.

Delgado (2012) “Evolving Signal Processing for Brain–Computer Interfaces”, Vol. 100, May 13th, 2012 |Proceedings of the IEEE.

8.

Anna M. Bianchi (2010) “Measures of connectivity among the different brain areas during an attention task”,32nd Annual International Conference of the IEEE EMBS.

Websites 1.

KurtThearling(2012),Informationaboutanalyticsanddatasciencefrom http://www.thearling.com/text/dmwhite/dmwhite.htm

2. Two crowns corporations, Introduction to Data Mining and Knowledge Discovery, Third Edition retrieved from http://www.ncrel.org/sdrs/areas/issues/content/cntareas/reading/li1lk23.htmw.stat.ucla.edu/~hqxu/stat19/introdm.pdf 3.

ArnaudDelorme,ICA(Independent

Component

Analysis)

for

dummies

retrieved

from

http://sccn.ucsd.edu/~arno/indexica.html 4.

San Diego (2001), International Workshop on Independent Component Analysis and Blind Signal Separation http://cnl.salk.edu/~tewon/ica_cnl.html

5.

Apache Mahout , K-Means Clustering

retrieved from https://cwiki.apache.org/MAHOUT/k-means-

clustering.html 6.

http://sccn.ucsd.edu/wiki/Chapter_2._Introduction

7.

Swartz center for computational neuroscience from http://sccn.ucsd.edu/eeglab/data/headit.html


3.Analysing EEG .FULL  
Advertisement