Predicting heat-stressed EEG spectra by self-organising feature map and learning vector quantizers——SOFM and LVQ based stress prediction

doi:10.4236/jbise.2010.35074

Paper Menu >>

Journal Menu >>

J. Biomedical Science and Engineering, 2010, 3, 529-537

doi:10.4236/jbise.2010.35074 Published Online May 2010 (http://www.SciRP.org/journal/jbise/

JBiSE

Published Online May 2010 in SciRes. http://www.scirp.org/journal/jbise

Predicting heat-stressed EEG spectra by self-organising

feature map and learning vector quantizers

——SOFM and LVQ based stress prediction

Prabhat Kumar Upadhyay1, Rakesh Kumar Sinha2, Bhuwan Mohan Karan3

1Department of Electrical and Electronics Engineering, Birla Institute of Technology, Ranchi, India;

2Department of Biomedical Instrumentation, Birla Institute of Technology, Ranchi, India;

3Department of Electrical and Electronics Engineering, Birla Institute of Technology, Ranchi, India.

Email: [email protected]; [email protected]

Received 9 December 2009; revised 28 December 2009; accepted 30 December 2009.

ABSTRACT

Self-Organising Feature Map (SOFM) along with

learning vector quantizers (LVQ) have been designed

to identify the alterations in brain electrical poten-

tials due to exposure to high environmental heat in

rats. Three groups of rats were considered—acute

heat stressed, chronic heat stressed and control

groups. After long EEG recordings following heat

exposure, EEG data representing three different vi-

gilance states such as slow wave sleep (SWS), rapid

eye movement (REM) sleep and AWAKE were visu-

ally selected and further subdivided into 2 seconds

long epoch. In order to evaluate the performance of

artificial neural network (ANN) in recognizing

chronic and acute effects of heat stress with respect to

the control subjects, unsupervised learning algorithm

was applied on EEG data. Mean performance of

SOFM with quadratic taper function was found to be

better (chronic-92.6%, acute-93.2%) over the other

two tapers. The effect of LVQ after the initial SOFM

training seems explicit giving rise to considerable

improvements in performance in terms of selectivity

and sensitivity. The percentage increase in selectivity

with uniform taper function is maximum for chronic

and its control group (4.01%) and minimum for

acute group (1.29%) whereas, with Gaussian it is al-

most identical (chronic-2.57%, acute-2.03%, control-

2.33%). Quadratic taper function gives rise to an in-

crease of 2.41% for chronic, 1.96% for acute and

2.91% for control patterns.

Keywords: ANN; EEG; Heat Stress; SOFM; LVQ

1. INTRODUCTION

The scientific interest of stress in relevance to health and

disease began to develop in the 20th century, when Selye

started his work on stress with more suited scientific

analysis, and now stress has been accepted to be a state,

comprised of certain psychophysiological reactions that

prepare an organism for action. It is usually described to

be an essential component that is enabling the organism

to survive in the hostile environment and to make an

effort to compensate with the altered situation of the

stressful conditions. Stress has been defined as nonspe-

cific responses of the body to any demand. Though in

some respect, every demand made on the body is unique

and specific, but all stress, however, have one thing in

common; they increase the demand for the readjustment

for performance of adaptive functions, which reestablish

normally. Generally, stress is meant to be acute or at

least of a limited duration. The time limited nature of

this process renders its accompanying antianabolic,

catabolic and immunosuppressive effects temporarily

beneficial and of no adverse consequences. Chroni-

cally and excessiveness of stress system activation, on

the other hand, would lead to the syndromal state that

severe chronic disease of any etiology could present

with anorexia, loss of weight, depression, and peptic

ulcers.

Although, the problems of heat-afflicted illness are

receiving increased importance in view of the current

estimates of global warming and its impact on biological

systems, the etiological factors that lead to heat exhaus-

tion and heat stroke have not been well established.

However, the failure of cardiovascular system had been

thought to be an important factor. Inadequate acclimati-

zation also appears to be a significant factor predispos-

ing to the onset of heat stroke. Review of literature re-

vealed that the afflictions and damages to the central

nervous system (CNS) imposed by high environmental

temperature have largely been ignored as the likely

cause of heat induced mortality, although it is well

known that neurochemical and cellular mechanisms of

neural issues are highly temperature sensitive [1]. The

conventional long term paper recording of EEG signals

530 P. K. Upadhyay et al. / J. Biomedical Science and Engineering 3 (2010) 529-537

following stress events such as heat stress is not of much

diagnostic value. So, computer and digital signal process-

ing tools have been used to quantify the EEG signals for

all three sleep-wake stages after acute and chronic heat

stress. Since, long-term EEG recordings, in addition to

other two channels of electrophysiological signals, EOG

(Electrooculogram) and EMG (Electromyogram), reflect

the variations in sleep-wake states, so the changes in sleep

parameters following acute as well as chronic states were

also observed. Further, to reduce the labor involved in the

manual sleep staging and to analyze psychophysiological

alterations, artificial neural network’s architectures were

designed. Present study signifies the computerized recogni-

tion of different sleep-wake states and the changes occurred

in EEG signals due to exposure to high environmental heat.

In the last decade, several works introduced the use of

artificial neural network (ANN) as a tool for automated

sleep scoring. Most of the system used spectral in-

formation of the signal using Fourier transformation

[2]. Computerized EEG and other electrophysiological

parameters monitoring reduces the problem of huge data

handling. Computerization has led to more sophisticated

use of EEG, even in effective disorders, where percep-

tual processes are significantly distorted [3-5]. Fourier

transformation, as a conventional method, has been

widely used for the standard quantitative analysis of the

spectral decomposition and the clinical application of

EEG signals [6]. The ANN programs were developed for

the analysis of most of the works that rely on spectral

analysis and power spectrum method to evaluate elec-

trographic data. In an attempt to classify sleep-wake

stages determined, power of FFT or power spectrum

band were used for better performance of the system

[3,7]. The numerical values of the power of different

frequency bands were used as inputs to ANN. As multi-

layer perceptron neural network (MLPNN) undergoes

some limitations, the performance of SOFM has also

been tested to solve the problem at hand. In the present

study, an effort has been made to exploit the inherent

qualities of SOFM. The results obtained from computer

simulations have been found to be very encouraging.

In addition to widespread application of artificial

neural networks (ANNs) in diagnosis, much develop-

mental work is being undertaken in signal processing

and analysis of bioelectric signals. ANNs are widely

used as to process raw electroencephalogram (EEG) data

or features. Recently, ANNs have been employed suc-

cessfully for many pattern recognition problems of elec-

troencephalogram (EEG) spectral component [8-11],

K-complex detection [12-13], event related potentials

(ERPs) detection [14-15], classification of the evoked

potentials [16-17] and the recognition of epileptic spike

patterns [18-19].

Review of literature reveals that supervised ANNs

have been used many times for sleep-wake state identi-

fication, but the literature on the methodology for use of

unsupervised ANNs are still obscure. The SOFM as

proposed by T. Kohonen [20-21] follows unsupervised

learning (competitive learning) and consists of a single

layer feed-forward network or lattice, the neurons of

which become specifically tuned to various input pat-

terns through a self-organizing process. The spatial loca-

tion of a neuron then corresponds to a particular feature,

or group of features, of the input patterns. Output neu-

rons of a topographic map are usually arranged such that

each neuron has a set of neighbours. Each node of the

output layer is connected with all other nodes of the

same layer with inhibitory weights and competitive

learning occurs among all the nodes of the output layer.

The most highly activated node which has least Euclid-

ean distance becomes the winner for a particular pattern. It

has been realized that the trained and calibrated Self-Or-

ganising Feature Map (SOFM) alone can not be used as a

classifier. Hence, to increase the performance of the pattern

classifier, Learning Vector Quantizers (LVQ) is applied in

the trained SOFM [22]. In the present study, unsupervised

ANN system using both SOFM and LVQ has been de-

signed for classification of heat stressed spectra.

2. MATERIALS AND METHODS

2.1. Subjects and Electrode Implantation

The experiments were carried out with male Charles

Foster rats of age 12-14 weeks and weight around 180-

200 grams. The rats were individually housed in poly-

propylene cages (30 cm × 20 cm × 15 cm) with drinking

water and food (Hindustan Liver Limited, India) ad libi-

tum. All rats were kept in an ambient environment tem-

perature of 23 ± 1°C from birth and the animal room was

artificially illuminated with 12 : 12 hours Light : Dark

cycle, changed at 7 o’clock and 19 o’clock Indian Stan-

dard Time (IST). The technique of electrode implanta-

tion for polygraphic sleep recordings have been used as

suggested earlier [7].

2.2. Heat Stress Model

In order to produce the effects of heat stress, rats were

subjected to the Biological Oxygen Demand (BOD) in-

cubator at preset temperature of 38 ± 1℃ and relative

humidity 45-50%, simulated with the environmental

conditions of Varanasi (India) in the months of May and

June. For chronic heat stress, rats were subjected to the

incubator for one hour daily for 21 days of chronic heat

exposure from 8.00 a.m. to 9.00 a.m. and electrophysio-

logical signals were recorded on 22nd day whereas, acute

heat stressed rats were subjected to the incubator for

continuous four hours of heat exposure from 8.00 a.m. to

12.00 p.m. for a single day, just before the recording of

electrophysiological signals. Respective control groups

of rats were placed in the incubator at room temperature

P. K. Upadhyay et al. / J. Biomedical Science and Engineering 3 (2010) 529-537 531

(23 ± 1℃) and whole procedure was followed exactly

similar to that of their stressed groups.

2.3. EEG Data Acquisition

Four hours of continuous recordings of EEG, electro-

oculogram (EOG) and electromyogram (EMG) were

performed from 12 o’clock to 16 o’clock IST on the re-

cording day for chronic and acute heat stressed rats

through the 8 channels Electroencephalograph (EEG - 8,

Recorders & Medicare Systems, India). The paper re-

cordings were performed with standard amplifier setup

[23] and at the chart speed of 7.5 mm/sec. The digitized

data (at sampling frequency of 256 Hz) was collected,

stored and processed with the help of data acquisition

system (ADLiNK, 8112 HG, NuDAQ, Taiwan) and

processing software (Visual Lab.-M, Version 2.0 c,

Blue Pearl Laboratory, USA).

2.4. Self-Organising Feature Map and Learning

Vector Quantiser

With self-organised learning, no external teacher is re-

quired in order to adjust the weights of the ANN, but the

choice of the input data set will still reflect the general-

izing ability of the ANN (as for the supervised case). It

presents a method of identifying similarities (or features)

in a vast (unlabelled) training set. An important advan-

tage of self-organising ANNs over their supervised

counterparts is that they can be exposed to and make use

of vast quantities of input data for training purposes

without the need for assigning labels to each input for-

warded to the ANN. Let m, x, and α be the weight vector,

input vector and learning rate parameter respectively

then with the following rules the weights of the winning

neuron as well as its neighbours are updated.

)()];()([)()()1( tNifortwtxRttwtw ciiii 



)();()1( tNifortwtw cii 

where, Nc is a topological neighbourhood function

centered around the winning neuron c and Ri is the

neighbourhood taper function. Three different taper

function-uniform, Gaussian, and Quadratic have been

used for training the network, the mathematical expres-

sions of which are as followed:

Linear: Ri = 1, for i Nc,



Gaussian: Ri = exp2

)1( 



d for i Nc,



Ri = 1-1.052

)1( 



d for i



Nc.

where, di is the Euclidean distance between the weight

vector wi and the winning weight vector wc. Learning rate (η)

and neighbourhood size (Nc) are assumed to be time vary-

ing. Two types of monotonically decreasing functions are

considered here, a piecewise linear decreasing function and

an exponential decreasing function. The piecewise linear

function is given by:

,);()1()( max

max

min

max

0kkif











,;)( maxmin kkifk





,;)1()( max

max

min

max

0kkif

NkNc







.;)(maxmin kkifNkNc





The piecewise exponential function is given by:

.;)(

,;)(

maxmin

)1(

max

maxmax

kkifNkN

kkifNNkN

kkifk

































where, in both cases, η0 and N0 give the initial learning

rate and neighborhood radius respectively, whereas ηmi n

and Nmin give the respective minimum values. Here, k

and µ denote iteration number and fraction of total

number of iteration respectively.

To achieve improved classification performance,

training and calibration of the SOFM with a supervised

learning scheme is followed. Learning vector quantization

(LVQ) is one such technique, developed by Kohonen

to fine-tune the weights of the trained SOFM in a

supervised manner [22]. LVQ is a supervised technique

that uses class information to move Voronoi vectors

slightly in such a way as to improve the quality of the

classifier decision regions. The initial values of wi before

the fine-tuning with LVQ, must be such that wi represent

the overall statistical density function of the input. The

SOFM is suited to achieve this. Out of the three versions

of the LVQ algorithm, LVQ1 has been applied in the

present work, which follows the following weight update

rules:

)]()()[()()1( twtxttwtw iii 









if i = c and x is classified correctly.

)]()()[()()1( twtxttwtw iii 









if i = c and x is classified correctly.

)]()()[()()1( twtxttwtw iii 









for i ≠ c.

2.5. Data Preparation and Simulation

The EEG data set prepared for input space of the net-

work was divided into training data set and test data set.

The training data set consists of the raw EEG signals of

SWS, REM and AWAKE states, each having matrix size

of [1, 512]. So, the total size of the input vectors for one

532 P. K. Upadhyay et al. / J. Biomedical Science and Engineering 3 (2010) 529-537

presentation at the input layer of SOFM is [1, 1536], in

which all the three sleep states have been equally pre-

sented. This constitutes one training epoch. The training

set was used to repeatedly train a 2-D lattice of neurons

of different sizes and varying SOFM training parameters.

The input data was presented to the network 600 times

with different values of learning rate parameter (η),

topological neighborhood function (N) centered around the

winning neuron and neighborhood taper function (h). The

weights of the trained SOFM now get ordered in the input

space such that they represent the underlying densities of

the inputs.

JBiSE

The program searches the position of the weight vectors

in relation to the input vectors after different iterations dur-

ing training SOFM of different sizes such as 4 × 4, 8 × 8, 10

× 10, 14 × 14. The distribution of the model vectors in the

n-dimensional space will approximate the probability dis-

tribution of the input vectors. The topographic organization

of the map will also approximate the metric ordering rela-

tions in the input space. Thus similar inputs project near

each other onto the map. The map thus forms an “elastic

surface” in the input space, which approximates the prob-

ability density function of the input samples. Increasing the

number of locations (neurons in the lattice) increases the

accuracy of the approximation. After training a 2-D lattice

of neurons by SOFM, each trained map was calibrated and

learning vector quantizer (LVQ) algorithm (a supervised

learning technique) was applied in each case for fine tuning

the weights of the trained SOFM.

2.6. Body Temperature

Core body temperature was recorded as stress markers for

both acute and chronic stress group of rats through the

thermistor probe connected to 6-channel telethermometer.

The marked probe at 4 cm was inserted to the rectum of the

animal and kept static for one minute to record the body

temperature. For acute stress group, body temperature was

recorded before and after the heat exposure. While for

the chronic stress group, the body temperature was re-

corded on every third day just before putting them into

the incubator for chronic heat stress.

3. RESULTS

With the processed EEG data sets as shown in Figure 1,

the network was simulated number of times and the per-

formance was calculated for some of the simulations

employing two decay functions of learning rate and

neighborhood size, three neighborhood tapering schemes,

and different number of training iterations. Use of linear

decay and exponential decay for learning rate and

neighborhood size did not show any fixed trend such

that conclusion could be drawn precisely. A few graphs

showing the mean change in Euclidian distance between

weight vectors of SOFM of varying sizes at successive

epochs during the training process, have been presented

in Figure 2. The first subscript denotes decay mode of

learning rate and neighborhood distance whereas, letters

1 and 2 stand for linear and exponential decay respec-

tively. The second subscript denotes taper function,

where 1, 2 and 3 represent linear, Gaussian, and quad-

ratic functions respectively. Time span of the ordering

phase and fine adjustment phase can distinctly be seen in

the curves of Figure 2. The ordering phase occurs within

the first 10%-20% of the training process and is char-

acterized by large changes in the Euclidean distances.

The fine adjustment phase is characterized by smaller

changes in the distances. Substantial increase in per-

formance was seen in the simulation of 8 × 8 SOFM in

acute heat stress. Number of iterations of the input vec-

tors required for simulation was set to 500 and 200 times

the number of neurons of the Kohonen layer. The results

as shown in the tables (Tables 1 and 2) suggest that iter-

ating 200 × size of the SOFM produces almost identical

Figure 1. Processed sleep EEG of awake, slow save sleep and REM.

P. K. Upadhyay et al. / J. Biomedical Science and Engineering 3 (2010) 529-537 533

020 40 60 80100120140 160 180 20

0.02

2,3

020 40 60 80100120140 160 180 20

0.02

0.04

2,3

020 40 60 80100120140 160 180 20

0.02

0.05

2,3

020 40 60 80100120140 160 180200

0. 02

0. 04

2,3

020 406080100 120 140 160 180200

0. 02

0. 04

2,2

020 40 6080100 120 140160 180200

0. 0 2

0. 0 4

2,2

020 406080100 120 140160 180200

0. 02

0. 04

1,1

020 406080100 120 140160 180200

0. 02

0. 04

2,2

020 4060 80100 120 140160 180200

0.02

0.04

2,2

Figure 2. The mean change in euclidean distance between weight vectors during training of SOFM of varying

sizes. The first subscript corresponds to decay mode of learning rate and neighbourhood distance whereas the

second subscript denotes the type of taper function.

results as iterating 500 × size of SOFM. In some simula-

tions, for both stress conditions, better network per-

formance has been obtained by iterating 200 × size of

the SOFM. For 14 × 14 SOFM used in chronic heat

stress, it was found to be 94.6% whereas, mean per-

formance of only 91.1% was obtained when the network

was iterated by 500 × size of the SOFM. All simulations

were performed on two different values of ηmin, where

comparatively, ηmin = 0.001 offered better performance.

As the SOFM size decreases, the mean changes in-

crease and the SOFM become more settled. 14 × 14

SOFM acquires the optimum set of weight vectors al-

most after 20% of the training whereas for 8 × 8 SOFM

it is 60%. It is also obvious that with the change of net-

work size, learning rate and neighborhood decay func-

tion, training time of the network as well as its conver-

gence are being directly affected. Having trained the-

network, calibration of SOFM is accomplished, in which

a class label has been assigned to each neuron ac cording

to the maximum voting criteria. Each neuron is also as-

signed a value, which provides the confidence level with

which each neuron represents that class. The boundaries

between the three patterns of heat stress-chronic, acute

and respective control groups (symbolized by-‘ch’, ‘ac’

534 P. K. Upadhyay et al. / J. Biomedical Science and Engineering 3 (2010) 529-537

Table 1. Mean performance of SOFM in chronic heat stress under varying conditions after simulation.

Table 2. Mean performance of SOFM in acute heat stress under varying conditions after simulation.

and ‘co’ respectively) as shown in Figures 3–5, are well

demarcated showing topological ordering of the neurons.

It is important to note that topological ordering is ob-

served, both in the plots of the weight vectors in the in-

put space and in the assigning of labels during calibra-

tion. Labels form neat clusters with easily distinguish-

able boundaries. It is at the boundary that the confidence

labels assigned to neurons are at the weakest, due to

overlapping of the input distributions. Another important

point is that the SOFM assigns equal number of neurons

to each class, if the three classes were equally repre-

sented in the training set. When a particular class is un-

der represented in the input set, it will also be under rep-

resented in the SOFM, because fewer weight vectors

will be assigned to represent that class of input data.

Results (Figure 3) also indicate that to some data set no

decision has been taken by the output layer nodes and

hence tie between the adjacent nodes seems to appear.

Parameter Mean performance (%) 8 × 8 SOFM

Decay Linear 90.1 Exponential 92.3 -

Iterations 500 × (SOFM Size) 91.4 200 × (SOFM Size) 92.3 -



min 0.01, 92.4 0.001, 92.8 -

Nc tapper Uniform 90.73 Gaussian 91.38 Quadratic 92.6

Algorithm SOFM 90.6 LVQ1 92.8 -

Parameter Mean performance (%) 10 × 10 SOFM

Decay Linear 80.2 Exponential 79.4 -

Iterations 500 × (SOFM Size) 83.2 200 × (SOFM Size) 83.0 -



min 0.01, 80.2 0.001, 79.1 -

Nc tapper Uniform 80.4 Gaussian 80.3 Quadratic 82.6

Algorithm SOFM 80.1 LVQ1 92.8 -

Parameter Mean performance (%) 14 × 14 SOFM

Decay Linear 93.1 Exponential 94.6 -

Iterations 500 × (SOFM Size) 91.1 200 × (SOFM Size) 94.6 -



min 0.01, 93.4 0.001, 93.7 -

Nc tapper Uniform 92.9 Gaussian 94.1 Quadratic 94.3

Algorithm SOFM 93.6 LVQ1 92.9 -

Parameter Mean performance (%) 8 × 8 SOFM

Decay Linear 84.4 Exponential 88.3 -

Iterations 500 × (SOFM Size) 88.1 200 × (SOFM Size) 87.8 -



min 0.01, 91.3 0.001, 92.45 -

Nc tapper Uniform 90.73 Gaussian 92.3 Quadratic 93.2

Algorithm SOFM 90.8 LVQ1 93.6 -

Parameter Mean performance (%) 10 × 10 SOFM

Decay Linear 91.5 Exponential 91.7 -

Iterations 500 × (SOFM Size) 91.0 200 × (SOFM Size) 92.3 -



min 0.01, 92.4 0.001, 92.2 -

Nc tapper Uniform 91.1 Gaussian 91.5 Quadratic 90.6

Algorithm SOFM 91.4 LVQ1 92.8 -

Parameter Mean performance (%) 14 × 14 SOFM

Decay Linear 94.1 Exponential 94.8 -

Iterations 500 × (SOFM Size) 93.1 200 × (SOFM Size) 94.2 -



min 0.01, 93.6 0.001, 92.1 -

Nc tapper Uniform 91.8 Gaussian 92.1 Quadratic 94.3

Algorithm SOFM 90.7 LVQ1 93.5 -

P. K. Upadhyay et al. / J. Biomedical Science and Engineering 3 (2010) 529-537 535

crcr

Figure 3. Topological ordering of the output neurons of 8 × 8 SOFM

representing three patterns of heat stress-chronic, acute, and control (rep-

resented by class labels-cr, ac and co respectively).

Figure 4. Topological ordering of the output neurons of 4 × 4 SOFM

representing three patterns of heat stress-chronic, acute, and control (rep-

resented by class labels-cr, ac and co respectively).

co co

coco co

acac

Figure 5. Topological ordering of the output neurons of 10 × 10 SOFM

representing three patterns of heat stress-chronic, acute, and control (rep-

resented by class labels-cr, ac and co respectively).

536 P. K. Upadhyay et al. / J. Biomedical Science and Engineering 3 (2010) 529-537

Depending on the individual performance of several

SOFM networks, 10 × 10 SOFM was selected and

trained for recognizing three clusters. In heat stress de-

tection, using 10 × 10 SOFM, average selectivity table

(Table 3) confirms the presence of three separate pat-

terns-chronic, acute and control groups. For acute and

control patterns, Gaussian taper function finds advantage

over the other two functions. Again training of SOFM

followed by fine-tuning with LVQ increases the per-

formance. This percentage increase in selectivity with

uniform taper function is maximum for chronic and its

control group (4.01%) and minimum for acute group

(1.29%) whereas, with Gaussian it is almost identical

(chronic-2.57%, acute-2.03%, control-2.33%). Quadratic

taper function gives rise to an increase of 2.41% for

chronic, 1.96% for acute and 2.91% for control patterns.

The study of average sensitivity of 10 × 10 SOFM (Ta-

ble 4) ascertains the appearance of three distinct patterns

in which all three taper functions were found to offer

identical results. However, small variation in the average

sensitivity can be realized for LVQ.

4. DISCUSSION

The changes in body temperature of the present rat mod-

el of heat stress due of acute or chronic heat stress is the

sterotype phenomena of heat stress and thus confirms the

stressful events along with EEG variations in these ex-

Table 3. Average selectivity of 10 × 10 SOFM for heat stress

detection.

Table 4. Average sensitivity of 10 × 10 SOFM for heat stress

detection.

periments. Special features provided by unsupervised

networks of labeling large quantities of data and training

them in a self-organized manner have been successfully

utilized in the present work. Since a number of training

and testing parameters play important roles in the clus-

tering problems, the effect of the various parameters on

the training of a SOFM has been studied by performing

simulation in Matlab. To achieve better classification

result, each parameter was varied one by one during si-

mulation and has been tabulated. Performance of differ-

ent SOFM was finally shown in terms of selectivity and

sensitivity. SOFM and LVQ were primarily used for de-

tecting three different patterns-chronic, acute, and con-

trol. EEG data of control subjects of chronic group and

acute group have been mixed together for training, as

variation with respect to stressed subjects were not found

to be much. Each simulation was performed on a differ-

ent input data set. After SOFM training was complete,

LVQ was used to calibrate SOFM.

Results obtained from the mean changes in Euclidian

distance between weight vectors during training of

SOFM of varying size and varying parameters indicated

the time of commencement of ordering phase and fine

tuning phase. The importance of appropriate selection of

taper functions and decay functions has also been well

studied by comparative analysis. After the LVQ was ap-

plied, all nodes in the output layer were found to be so

tuned that three classes of heat stress become apparently

separated. Observations suggest that the reason behind

the largest change encountered in performance of many

simulations, might have been due to the use of different

neighborhood taper functions. Overall, performance was

found to be better for quadratic taper over the other two

tapers. In some simulations, it was witnessed that the

increase in performance due to LVQ was negligible,

which might have occurred owing to the fact that the

input clusters were already reasonably well defined and

with a minimal overlap.

In the present work, an attempt has been made to

classify stressful conditions by means of changes in

EEG signals induced by high environmental tempera-

tures. The review of literature suggests that no work

has been reported that classifies heat stressed condi-

tions from normal candidates with help of SOFM and

LVQ. The ANN provides reliable information about

the stressed and normal artifact-free EEG power spec-

tra. However, in practical applications, EEG artifacts

can influence the sensitivity of the network. EEG pat-

tern recognition by ANN and clinical skill, are, how-

ever, not mutually exclusive but even reinforce each

other, and it is believed that a human clinician must

remain a necessary component of computerized diag-

nostic procedures to ensure a significant, high level of

diagnostic validity [24].

Uniform taper

Chronic Acute Control

SOFM 88.42 88.91 87.86

LVQ1 92.43 90.2 91.64

Gaussian taper

SOFM 87.62 89.50 88.02

LVQ1 90.19 91.53 90.35

Quadratic taper

SOFM 87.09 88.87 87.09

LVQ1 89.23 90.83 90

Uniform taper

Chronic Acute Control

SOFM 88.30 87.54 87.81

LVQ1 91.16 90.70 89.80

Gaussian taper

SOFM 87.64 88.70 87.64

LVQ1 89.97 90.65 90.23

Quadratic taper

SOFM 86.89 88.58 86.98

LVQ1 89.25 91.06 89.95

P. K. Upadhyay et al. / J. Biomedical Science and Engineering 3 (2010) 529-537 537

Studies indicate that, usually only 1-5% of the EEG

record is in clinical interest [25]; neural networks can

become useful for the on-line classification of EEG

waves. Since, exposure to high environmental heat has

significant effects on brain signal, SOFM with LVQ can

be used efficiently to identify the changes in brain sig-

nals, occurred due the stressful events and can also be

used further to develop an automated detection system

for psychophysiological analysis. Although, in the pre-

sent study, on-line classification was not carried out, it

may be possible with the help of fast computers and spe-

cific software. Furthermore, EEG technicians can easily

be trained for the manual selection of already detected

events whereas; recognition of abnormal patterns in the

background of the ongoing EEG requires substantial

experience.

JBiSE

REFERENCES

[1] Sinha, R.K. (2007) Study of changes in some patho-

physiological stress arkers in different age groups of an

animal model of acute and chronic heat stress. Iranian

Biomedical Journal, 11(2), 101-111.

[2] Claude, R., Cristian, G. and Limoge, A. (1998) Review of

neural network application in sleep research. Journal of

Neuroscience methods, 79(2), 187-193.

[3] Morstyn, R., Duffy, F.H. and McCarley, R.W. (1983)

Altered topography of EEG spectral content in schizo-

phrenia. Electroencephalography and Clinical Neuro-

physiology, 56(4), 263-271.

[4] Lin, S.L., Tsai, Y.J. and Liou, C.Y. (1993) Conscious

mental tasks and their EEG signal. Medical and Biologi-

cal Engineering and Computing, 31(4), 421-426.

[5] Sarbadhikari, S.N., Dey, S. and Ray, A.K. (1996) Chronic

exercise alters EEG power spectra in an animal model of

depression. Indian Journal of Physiology and Pharma-

cology, 40(1), 47-57.

[6] Cesarelli, M., Clemente, F. and Bracale, M. (1990) A

flexible FFT algorithm for processing biomedical signals

using a PC. Journal of Biomedical Engineering, 12(6),

527-530.

[7] Sinha, R.K. (2003) Artificial Neural Network detects

changes in electro-encephalogram power spectrum of

different sleep-wakes in an animal model of heat stress.

Medical and Biological Engineering and Computing,

41(5), 595-600.

[8] Veselis, R.A., Reinsel, R., Sommer, S. and Carlon, G.

(1991) Use of neural network analysis to classify elec-

troencephalographic patterns against depth of midazolam

sedation in intensive care unit patients. Journal of Clini-

cal Monitoring, 7(3), 259-267.

[9] Doering, A., Galicki, M., Witte, H. and Krajeca, V. (1995)

Structure optimization of neural networks with A*-algo-

rithm application in EEG pattern analysis. Medical In-

formation for Patients, 8(1), 814-817.

[10] Witte, H., Doering, A., Galicki, M., Dorschel, J., Kra-

jeca, V. and Eiselt, M. (1995) Application of opti-

mized pattern recognition units in EEG analysis:

Common optimization of preprocessing and weights of

neural networks as well as structure optimization. Medi-

cal Information for Patients, 8(1), 833-837.

[11] Sinha, R.K., Aggarwal, Y. and Das, B.N. (2007)

Backpropagation artificial neural network classifier to

detect changes in heart sound due to mitral valve re-

gurgitation. Journal of Medical Systems, 31(3), 205-

209.

[12] Jansen, B.H. (1990) Artificial neural nets for K-

Complex detection. IEEE Engineering in Medicine

and Biology Magazine, 9(3), 50-52.

[13] Bankman, I.N., Sigillito, V.G., Wise, R.A. and Smith,

P.L. (1992) Feature-based detection of the K-complex

wave in the human electroencephalogram using neural

networks. IEEE Transactions on Biomedical Engi-

neering, 39(12), 1305-1310.

[14] Wu, F.Y., Slater, J.D., Honing, L.S. and Ramsay, R.E.

(1993) A neural network design for event-related po-

tential diagnosis. Computers in Biology and Medicine,

23(33), 251-264.

[15] Gupta, L., Molfese, D.L. and Tammana, R. (1995) An

artificial neural network approach to ERP classifica-

tion. Brain and Cognition, 27(3), 311-330.

[16] Bruha, I. and Madhvan, G.P. (1989) Need for a know-

ledge based subsystem in evoked potential neural net

recognition system. Proceedings of the 11 th Annual Interna-

tional Conference on IEEE-EMBS, 6, 2042-2043.

[17] Holdaway, R.M., White, M.W. and Marmarou, A.

(1990) Classification of somatosensory evoked poten-

tials recorded from patients with severe head injuries.

IEEE Engineering in Medicine and Biology Magazine,

9(3), 43-49.

[18] James, C.J., Jones, R.D., Bones, P.J. and Carrol, G.J.

(1996) The self-organizing feature map in the detec-

tion of epileptiform transients in the EEG. Proceed-

ings of the 18th international Conference of the IEEE

Engineering in Medicine and Biology Society, Am-

sterdam, 1996, 913-914.

[19] Adeli, H., Zhou, Z. and Dadmehr, N. (2003) Analysis

of EEG records in an epileptic patient using wavelet

transform. Journal of Neuroscience Methods, 123(1),

69-87.

[20] Kohonen, T. (1990) The self-organizing map. Pro-

ceedings of the IEEE, 73, 1464-1480.

[21] Kohonen, T. (1989) Self organization and associative

memory. 3rd Edition, Springer-Verlag, Berlin.

[22] Kohonen, T. (1988) Learning vector quantization.

Neural Networks, 1, 303.

[23] Dubois, M., Sato, S., Lees, D.E., Bull, J.M., Smith, R.,

White, B.G., Moore, H. and Macnamara, T. (1980)

Electroencephalographic changes during whole body

hyperthermia in humans. Electroencephalography and

Clinical Neurophysiology, 50(5-6), 486-495.

[24] Sarbadhikari, S.N. (1995) A neural network confirms

that physical exercise reverses EEG changes in de-

pressed rats. Medical Engineering & Physics, 17(8),

579-582.

[25] Jandó, G., Siegel, R.M., Horvàth, Z. and Buzsàki, G.

(1993) Pattern recognition of the electroencephalo-

gram by artificial neural networks. Electroencephalo-

graphy and Clinical Neurophysiology, 86(2), 100-109.