2008 Qeeg Index

8/2/2019 2008 Qeeg Index

1/15

This article was downloaded by:[Universidad de Sevilla]On: 8 January 2008Access Details: [subscription number 786928174]Publisher: Informa HealthcareInforma Ltd Registered in England and Wales Registered Number: 1072954Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Brain InjuryPublication details, including instructions for authors and subscription information:http://www.informaworld.com/smpp/title~content=t713394000

A QEEG index of level of functional dependence forpeople sustaining acquired brain injury: The SevilleIndependence Index (SINDI)Jose Leon-Carrion ab; Juan Francisco Martin-Rodriguez ab; Jesus Damas-Lopez b;Juan Manuel Barroso Y. Martin a; Maria Del Rosario Dominguez-Morales ba Human Neuropsychology Laboratory, School of Psychology, Department ofExperimental Psychology, University of Seville, Seville, Spainb Center for Brain Injury Rehabilitation (CRECER), Seville, Spain

Online Publication Date: 01 January 2008

To cite this Article: Leon-Carrion, Jose, Martin-Rodriguez, Juan Francisco,Damas-Lopez, Jesus, Martin, Juan Manuel Barroso Y. and Dominguez-Morales,Maria Del Rosario (2008) 'A QEEG index of level of functional dependence for

people sustaining acquired brain injury: The Seville Independence Index (SINDI)', Brain Injury, 22:1, 61 - 74To link to this article: DOI: 10.1080/02699050701824143URL: http://dx.doi.org/10.1080/02699050701824143

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf

This article maybe used for research, teaching and private study purposes. Any substantial or systematic reproduction,re-distribution, re-selling, loan or sub-licensing, systematic supply or distribution in any form to anyone is expresslyforbidden.

The publisher does not give any warranty express or implied or make any representation that the contents will becomplete or accurate or up to date. The accuracy of any instructions, formulae and drug doses should beindependently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings,demand or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with orarising out of the use of this material.
http://www.informaworld.com/smpp/title~content=t713394000http://dx.doi.org/10.1080/02699050701824143http://www.informaworld.com/terms-and-conditions-of-access.pdfhttp://www.informaworld.com/terms-and-conditions-of-access.pdfhttp://dx.doi.org/10.1080/02699050701824143http://www.informaworld.com/smpp/title~content=t713394000

8/2/2019 2008 Qeeg Index

2/15

Brain Injury, January 2008; 22(1): 6174

A QEEG index of level of functional dependence for peoplesustaining acquired brain injury: The Seville Independence

Index (SINDI)

JOSE LEON-CARRION1,2, JUAN FRANCISCO MARTIN-RODRIGUEZ1,2,

JESUS DAMAS-LOPEZ2, JUAN MANUEL BARROSO Y. MARTIN1, &

MARIA DEL ROSARIO DOMINGUEZ-MORALES2

1Human Neuropsychology Laboratory, School of Psychology, Department of Experimental Psychology, University of

Seville, Seville, Spain and 2Center for Brain Injury Rehabilitation (CRECER), Seville, Spain

(Received 31 August 2007; accepted 23 November 2007)

AbstractPrimary objective: To find an easy-to-use, valid and reliable tool for evaluating the level of functional dependence ofan individual with brain damage who seeks a diagnosis of his/her functional dependence in daily activities.

Methods: Eighty-one patients with acquired brain injury (ABI) in post-acute phase, 40 traumatic brain injury (TBI) and 41cerebral vascular accident (CVA), were assessed using quantitative electroencephalography (QEEG) and grouped accordingto the FIM FAM scale. Discriminant analysis was performed on QEEG variables to obtain a discriminant function withthe best discriminative capacity between functionality groups.Results: Discriminant analysis showed classification accuracy of 100% in the training set sample and 75% in an externalcross-validation sample; 100% sensitivity and 100% specificity were reached. Coherence measures were the most numerous

variables in the function.Conclusions: These results point out that the discriminant function may be a useful tool in objective evaluations of patientsseeking a diagnosis of their level of dependence and that it could be included in current functionality assessment protocols.

Keywords: Traumatic brain injury, stroke, functional independence, QEEG, neuropsychological assessment, forensic assessment

Introduction

Neurological damage and acquired brain injury

affect a large portion of the population in western

countries. An estimated 1.5 million Americans

sustain traumatic brain injury (TBI) every year in

the US. Of these, 1.1 million are now living with

disabilities related to TBI [1]. In the European

Union, brain injury accounts for 1 million hospital

admissions per year. At discharge, most of these

patients show impairments in multiple areas, which

affect their ability to carry out daily life activities and

cause important legal, professional and personal

consequences.

In Europe, new laws are being established by

national governments to assist people in accordance

with their level of dependency. In Spain, a new law

was approved on 15 December 2006 called

The Law to Promote Personal Autonomy and toAssist People in a State of Dependence. This law

has created a bureaucracy and a medical system that

regulate promoting the autonomy of dependent

individuals as well as the state-funded care they

receive. As a result, medical and social resources are

needed to implement this law, which will serve more

than 2 000 000 people in Spain alone. More than a

third of these individuals are dependent in their daily

Correspondence: Jose Leon-Carrion, PhD, Human Neuropsychology Laboratory, School of Psychology, Department of Experimental Psychology, C/Camilo

Jose Cela s/n. University of Seville, Seville-41018. Spain. Tel: 34 95 457 4137. Fax: 34 95 437 4588. E-mail: [email protected]

ISSN 02699052 print/ISSN 1362301X online

2008 Informa UK Ltd.DOI: 10.1080/02699050701824143

8/2/2019 2008 Qeeg Index

3/15

life activities due to neurological disorders, the more

prevalent being cerebrovascular disorders and TBIs.

Any individual in a state of dependency that asks to

be included in the new state-funded system must be

clinically evaluated to determine whether they are

dependent or not and, if so, at what level. If they are

clinically proven to be in a dependent state, then,

apart from the state-funded care, they may also

receive financial support and other benefits.

Under this new legal action, and in order to avoid

malingering, an objective evaluation of the func-

tional state of these patients is needed, with new

tools that offer high reliability and validity.

Nowadays, few objective instruments exist to eval-

uate independence in people seeking social and legal

assistance. Existent tools generally rely on the

subjective impressions of caregivers, whose inter-

pretations may distort results and thus provoke false

positives. Moreover, exaggerating and malingering

symptoms are very common among patients withacquired brain injury, especially when monetary

compensation is expected or in litigation, where

they make up 1520% of all cases [2]. Individuals

with mild brain damage and post-concussional

disorders constitute 40% of those seeking compen-

sation [3].

Functionality is considered a multidimensional

concept, where the whole cannot be explained by the

mere sum of its parts. Instruments that evaluate

functionality must be capable of detecting malinger-

ing. In current clinical practice, functionality assess-

ment evaluates both the level of independence in

basic tasks, such as eating or bathing, as well asin more complex and instrumental tasks, such as

getting on a bus. Thus, a full assessment has

multiple levels of complexity involving multiple

assessors. The need for objective functional assess-

ments would require summarizing this complex

concept in an index that informs on the impact of

certain medical conditions on the lives of patients.

In recent years, advanced technology has been

used to locate affected areas of the brain with greater

precision. The use of sophisticated imaging software

has made it possible to create defined cerebral maps

that can be adapted to a patients brain in order to

determine which functional areas are affected. In thefield of brain injury, this technology is needed in

order to make objective evaluations and determine

the severity of the condition. When applying

neuroimaging techniques to the assessment of brain

injury, the electroencephalography (EEG) plays a

crucial role, with its relatively low cost, simple

application procedure, high testre-test reliability

and inherent stability [47]. Moreover, the develop-

ment of mathematical tools and data visualization

has made it possible to quantitatively analyse the

human EEG, a technique known as quantitative

EEG (QEEG). Human QEEG measures have been

correlated with certain diagnostic categories, both in

healthy [8,9] and clinical populations [10]. By

meeting certain statistical requirements, these stu-

dies have obtained a set of QEEG variables, known

as discriminant functions, which can predict the

severity of a clinical condition. The advantages of

these measurement systems over others include their

low cost, speed, lack of cultural influence and

minimal human intervention in the analysis of the

results. These systems allow predictions to be made

on an individuals psychological and functional

characteristics based on specific physiological

variables.

In the field of brain injury, functional indepen-

dence is normally evaluated using clinical behaviour

scales, the most widely used being the FIM FAM

[11], the Glasgow Outcome Scale (GOS) [12] and

the latters extended version [13]. The first scale was

devised by adding FAM items to FIM in order toaddress functional assessment in the TBI popula-

tion; the second was designed to evaluate the general

state of patients after neurosurgery. These scales,

which proved reliable and easy-to-use, were later

applied to other areas of evaluation. This caused

serious drawbacks to their effectiveness, a common

problem being that both scales offered too few items

for evaluating higher psychological functions and the

possibility of malingering.

However, a recent study by van Baalen et al. [14]

found the FIM and FAM scales to be highly reliable,

sensitive tools for assessing TBI patients. The

authors stressed that both scales offered betterresults during the first phases of post-TBI recovery,

as their effectiveness diminished during post-acute

phases and rehabilitation (1 year after brain injury).

Studies that correlated sensitivity in long-term TBI

population (up to 10 year post-injury) reported low

contributions from FIM FAM and GOS in func-

tional status assessment [15]. A new tool based on

the neurophysiological profile of brain injury patients

could help overcome these limitations by establish-

ing a more robust measure for the diagnosis of

long-term brain injury population.

Studies on QEEG present a consistent and

common neurophysiological pattern associated withseverity of brain injury, which involves increased

slow band amplitudes, decreased fast band ampli-

tudes [16,17] and changes in EEG coherence [18].

The present study pays special attention to the

differences in these measures.

The aim of the present study is to find the linear

combination of QEEG variables, by means of the

discriminant analyses, with the best discriminative

capacity for functional states. These states include

complete dependence (total assistance in various

DLA), modified dependence and independence.

62 J. Leon-Carrion et al.

8/2/2019 2008 Qeeg Index

4/15

The discriminant analysis helps identify the char-

acteristics that differentiate (discriminate) two or

more groups. One will create a function capable of

distinguishing between the possible members of each

group with highly accurate precision. For classifica-

tion, the discriminant function must have linearity as

one of its characteristics. This feature exists when

the function follows the proportionality of magni-

tudes of the entry and exit variables. When this

happens, the scores in this function of the modified

dependence group members should fall between

those of the complete dependence and independence

group members.

Methods

Participants

The total sample included 81 patients with acquired

brain injury (ABI), 40 of which were traumatic braininjury (TBI) patients and 41 cerebral vascular

accidents (CVA). Of these, 48 made it into the

training set for the creation of the discriminant

function and its subsequent internal validation. The

remaining 33 patients participation served to

validate the function externally. Patients in the

training set sample ranged in age from 1675

(M 40); 27 were TBI patients and 21 CVA; 37

were male and 11 female. All subjects were recruited

from the Centre for Brain Injury Rehabilitation

(C.RE.CER) in Seville, Spain. Admission criteria

were age (over 16) and a clinical background of ABI

caused by TBI or AVC, confirmed by neuroimagingtests (CT or MRI). Figure 1 shows sample size and

representation of the clinical sub-groups.

All of the patients were in chronic or sub-acute

phase when examined. The average time period

between brain injury and QEEG evaluation was

22 months (range 0.5119 months). They were

also participants in a holistic, integral and

multidisciplinary rehabilitation programme which

treats neuropsychological, physical and functional

sequelae derived from brain injury.

Procedure

Functional assessment: FIMFAM. The

FIM FAM is a multidimensional scale for func-

tional assessments, widely used in evaluating the

impact of rehabilitation on the ABI population. The

current version of this scale (FIM FAM) stems

from a combination of the FIM (Functional

Independence Measure) and the FAM (Functional

Assessment Measure). It consists of 30 items, 18

from FIM and 12 from FAM. These items are

grouped into seven sub-scales: self-care (items 17),

sphincter control (89), mobility (type of transfer)

(1013), locomotion (1416), communication

(1721), psychological adjustment (2225) and

cognitive functions (2630).The FIM FAM scales have been widely studied

and are considered a valid assessment tool for ABI

patients. Statistical studies found a reliability index

of 0.860.97 [18,19]. They have also shown reason-

able validity, internal consistency and discriminative

capacity between brain damaged sub-populations

[20]. These characteristics render the FIM FAM

scale one of the most popular instruments for

evaluating the functional state of neurological

patients.

Three independent assessors (a physical therapist,

a speech therapist and a neuropsychologist) com-

pleted the FIM FAM sub-scales. All items werescored from 17 points, where 1 indexes extremely

functional dependence and 7 indexes total functional

independence. The average interval between func-

tional assessment and QEEG was 1 day (SD 1.32).

Averages were calculated for each FIM FAM

sub-test, as well as the average of the FAM items,

the FIM items and total FIM FAM. Based on

Figure 1. Sample sizes and clinical features.

A QEEG index of level of functional dependence for people sustaining ABI 63

8/2/2019 2008 Qeeg Index

5/15

these averages, three functional groups were created:

Complete Dependence (CD; range 12.99),

Modified Dependence (MD; range 35.99) and

Independence (I; range 67).

There were no significant differences between the

three functional groups in gender (p 0.448),

aetiology of ABITBI or CVA (p 0.111), age

(p 0.748) or time period from brain injury to

QEEG test (p 0.337).

EEG recordings. EEG recordings were carried out

in a softly-lit, sound-proof room, with room tem-

perature set at 23C. Before each recording, patients

were seated in a comfortable armchair or in his/her

wheelchair and asked to relax during the recording.

Impedance was kept below 10 k. Nineteen scalp

locations were taken into account, based on the

international 10/20 system [21], using linked ears

(A1 and A2) as a reference.The patient was asked (or helped) to close his/her

eyes (EC) and remain relaxed and alert while EEG

activity was recorded. In order to maintain vigilance,

a technician monitored each subject, inspecting the

EEG traces on-line and verbally alerting the

subject any time behavioural and/or EEG signs of

drowsiness appeared. Each recording lasted 3

minutes, with bandwidths of 0.1100 Hz and

256 Hz sampling frequency.

Data selection. Data pre-processing and filtering

were carried out offline in Matlab

(TheMathworks, MA) using EEGLAB (http://

www.sccn.ucsd.edu/eeglab/index.html) [22] and

custom scripts. Low-pass filters were located at

40 Hz and high-pass filters at 0.5 Hz. EEG record-

ings were visually edited to remove any visible

artifact. Continuous artifacts or artifacts present in

over 80% of overall time are generally due to the

presence of spasticity or any motor artifact in some

patients. In most cases, ocular movement in frontal

electrodes and muscular tension in temporal deriva-

tions caused these artifacts. The authors applied an

Independent Component Analysis (ICA), a proce-

dure which has proven reliability in removing signalartifacts [23]. A maximum of two components were

selected that isolated the artifacts, which were then

removed from the original recording. This process of

artifact removal was carried out on a total of 11

recordings (22%) out of 50. Subsequently, frag-

ments free of artifacts in the most representative

EEG sections were visually selected, using

Neuroguide 2.2.1 software (Applied Neuroscience,

Inc., St. Petersburg, FL), for a total of 120 seconds

of recordings. Only fragments with over 90%

reliability were used for the spectral analysis.

QEEG measures. The measures most widely used in

QEEG research are spectral patterns and connectiv-

ity patterns. The former are based on the analysis of

EEG signal frequency spectrum at a specific loca-

tion; they are independent of time and yield the

intensity of the electromagnetic field of that location.

The latter are more complex measures that involve

spatial-temporal characteristics and include various

locations, yielding the strength of the connections

between these brain regions.

Amplitude measures

The most basic amplitude measure is absolute

magnitude (amplitude), defined as the average

absolute magnitude (expressed in microvolts) of

the frequencies that make up a specific band, over

a given time period. The relative magnitude is the

average relative magnitude (expressed in%) of a

specific frequency band (the absolute magnitudedivided by total microvolt generated at a particular

location by all bands). The amplitude asymmetry is

defined as the amplitude difference between two

locations in a particular bandwidth. It is calculated

as follows: (A B)/(A B), where A and B are

different locations [24].

Connectivity measures

Connectivity measures (or amplitude independence)

are coherence and phase lag. Coherence indexes the

degree to which two areas of the cortex are

functionally linked. Statistically, this measure iscalculated as the likelihood that two random signals

will arise from a common generator process and the

frequency bands in which this occurs [25].

Coherence is calculated for each frequency band

and for each combination of the 19 electrodes.

Coherence is calculated as:

2xyf

Gxyf 2

GxxfGyyf

where Gxy(f) is the cross-power spectral density and

Gxx(f) and Gyy(f) are the autopower spectral

densities, respectively. Due to the complex analysesinvolved in this formula, the calculation is made with

the cospectrum (r for real) and quadspectrum (q for

imaginary). Thus, coherence is obtained using the

following formula [26]:

2xy

r2xy q2xy

GxxGyy

Phase lag is defined as the time it takes one

wavelength from a particular location to reach the

phase or maximum amplitude of another wavelength

originating from a different specific location.


8/2/2019 2008 Qeeg Index

6/15

The contribution of each frequency band to the

EEG signal was obtained by means of the Fast

Fourier Transformation, FFT. The frequency bands of

interest were delta (13.5 Hz), theta (47.5Hz),

alpha (812 Hz) and beta (1225 Hz), including high

beta (25.530 Hz). Coherence and phase lag were

computed using cross-spectra analysis.

Statistical analyses

Descriptive analyses were conducted on both demo-

graphic and behavioural variables. A factorial analy-

sis (varimax rotation) was applied to each of the

eight measures that make up the FIM FAM scale

(all quantitative variables). The highest load factor

explained the total variance of 85.52%. Correlation

analyses were performed to study the relation

between the different measures, with positive results

showing Pearson correlation coefficients (r) higherthan 0.8. These two results allowed one to select the

total FIM FAM measure (average of seven mea-

sures) as the dependent variable for the subsequent

discriminant analysis.

The aim of this experimental design was to obtain

a discriminant equation, capable of classifying and

predicting the functional state of each patient based

on his or her QEEG pattern. The first objective was

to differentiate between the extreme groups (inde-

pendence vs complete dependence). After classifying

the two groups, the equation was tested on an

intermediate functional group (modified depen-

dence) in order to evaluate the hypothesis on thelinearity of functionality. The final step was to

classify a new group of patients using this equation.

The procedure for data reduction and creation of the

discriminant function is summarized in Figure 2.

Power spectral analyses yielded a total of 2755

variables as candidates for the EEG discriminant

analysis. Student t-tests were conducted to deter-

mine the discriminant capacity of each variable,

using the functional group of the patient as grouping

factor. Of the 2755 variables, 368 (11 absolute

amplitude, 40 relative amplitude, 98 amplitude

asymmetry, 184 coherence and 35 phase lag) were

significant (taking into consideration the application

of the Levene test). No correction for multiple

comparisons was applied, given that the objective

of this analysis was data reduction, i.e. to distinguish

between significant and non-significant variables

without making inferences from the analysis.

In order to reduce the data, a factorial analysis was

applied to each EEG measure (absolute amplitude,

relative amplitude, amplitude asymmetry, coherence

and phase lag). A Varimax rotation was applied.Variables with the highest load for each factor were

identified (three absolute amplitude variables, five

relative amplitude, 14 amplitude asymmetry, 18

coherence and 11 phase lag). The result was a total

of 51 variables. By using this two-level procedure, the

initial set of 2755 variables was reduced by 98.15%.

The next step involved performing a discriminant

analysis, using the Bayesian criterion in order to

adjust for the n difference between groups. A step-

wise method was used to obtain information on the

individual significance of each variable in the

discriminant function. The Wilks Lambda statistic

was used to assess each of the 51 variables. In orderto demonstrate the linearity of the discriminant

equation, this study analysed the correlation between

Figure 2. Procedure used to reduce data and create discriminant function.


8/2/2019 2008 Qeeg Index

7/15

the FIM FAM scores and the corresponding

discriminant scores (DS). A one-way ANOVA for

independent measures was used to determine the

differences between the functional groups discrimi-

nant scores, with DS as the dependent variable and

functional group as grouping factor (three functional

groups: complete dependence, modified dependence

and independence). Post-hoc tests were applied to

analyse the differences between groups in pairs.

Results

Functional outcomes

The factorial analysis on the 11 variables of the

FIM FAM scale (eight measures, FIM items,

FAM items and overall FIM FAM) produced a

factor that explained the total variance of 85.52%.

Furthermore, correlations between these variables

reached significance (p

8/2/2019 2008 Qeeg Index

8/15

Since this procedure tends to over-estimate the

precision of the classification, we used the leave-one-

out method of classification. This study observed

that the global classification accuracy is near 100%

(see Table IV).

Figure 4 shows the distribution of each functional

group and lineal adjustment of the DS obtained with

the function. The lineal adjustment is correct and

there is distinction between the groups.

The clinical validity of the DS was tested by means

of the Pearson correlation for each FIM FAM

variable. Figure 5 shows the results of this correla-

tion analysis. All correlations reached significance

(all ps

8/2/2019 2008 Qeeg Index

9/15

showed statistically significant differences between

the Independence and Complete dependence groups,

using the t-test. A chi-squared analysis was performed

with adjusted standardized residual analysis to inter-

pret the signs of the correlations. This procedure was

applied only to the measures of absolute amplitude,

coherence and phase lag. It was not applied to the

measures of relative amplitude and asymmetrical

amplitude due to the difficulty of interpreting signs of

the correlations for these measures. Table V sum-

marizes the results of the correlations betweenabsolute amplitude, coherence and phase lag and

the FIM FAM scores. The correlations of the

coherence variable are significantly higher in

number than those of absolute amplitude and phase

lag (2 142; p

8/2/2019 2008 Qeeg Index

10/15

Figure 5. Diagram of distribution and lineal adjustment between DS (x-axis) and each FIM FAM sub-scale (y-axis). The colour code

represents the patients FIM FAM scores, red being the lowest score and blue the highest.


8/2/2019 2008 Qeeg Index

11/15

highest extreme. Positive correlation and multipleregression analyses supported the linearity hypoth-

esis. The significant positive correlations between the

DS and the other measures of the FIM FAM sub-

scales attested to the clinical validity of the function.

In addition, EEG frequency-based analysis

showed a pattern of the differences in EEGs between

dependence and independence patients. The pattern

for dependent individuals was characterized by an

increase in slow wave amplitude and a decrease in

fast wave amplitude. Although the slow wave

increase was generalized, there appeared to be a

Figure 6. Mean DS for the three functional groups: Complete dependence, Modified dependence and Independence. The mean DS for the

Modified dependence group falls between the DS of the other two groups. (Overlapped) Diagram of distribution and lineal adjustment

between DS (y-axis) and global FIM FAM scores (n 48). The colour code represents the patients FIM FAM scores, red being

the lowest score, green being the mid-range and blue the highest.

Figure 7. (Left) Predicted values for the FIM FAM variable for all patients (n 48) based on the multiple regression analysis (y-axis) and

the DS (x-axis). (Right) Predicted values for the FIM FAM variable for all patients (n 48) based on the multiple regression analysis

(x-axis) and the original FIM FAM scores (y-axis).

Table V. Discriminant analysis classification of Independence

and Complete Dependence groups, based on the leave-one-out

method.

Classification,% as

FIM FAM group N Independence

Complete

dependence

Independence 14 100 (n 14) 0

Complete dependence 15 6.7 (n 1) 93.3 (n 14)

Overall classification accuracy 96.65%.


8/2/2019 2008 Qeeg Index

12/15

greater presence of these waves in the left hemi-

sphere. The decrease of fast waves was generalized in

the group with the highest level of dependence. This

finding is in accordance with the literature on brain

injury EEGs. A recent study [27] found that TBI

patients with a lower response level showed a slower

EEG than patients with higher responsiveness.Moreover, a generalized presence of slow waves in

the EEG has been identified as a factor in negative

prognoses [17,28].

Specific weight of QEEG variables (type, location

and band)

To create the discriminant function an absolute

amplitude variable and an amplitude asymmetry

variable were used. No relative amplitude variable

was used. The connectivity measures (coherence and

phase lag) did, however, take on an important role in

obtaining the function. These measures illustrate the

connections between different areas of the cerebral

cortex. They also quantify the cortico-cortical

coupling, which indicates the level of functional

connection between two areas.

Analyses on the variables which best distinguished

between the two functional states identified those

associated with coherence as the most effective.

Consequently, this study will now concentrate onthe significance of these variables.

It is known that areas of the brain are connected to

one another at certain levels. When two areas show

low coherence, they are functionally disconnected;

when they show very high coherence (hypercoher-

ence), they are excessively connected. In both

situations, a cerebral connectivity deficit is implied.

In the discriminant function, four coherence vari-

ables are included in three different frequency

bands. These four variables, all of which are long

distance connections, involve various cerebral lobes.

The discriminant function demonstrates the impor-

tance of long distance fibres that connect distantregions of the brain. Since level of dependence/

independence is a global measure, one would not

expect one single location in the brain to be respon-

sible for this function. It is more feasible to suggest

that there are wide, interconnected brain circuits

that account for this complex function. This finding

corresponds with the models of a functional brain

proposed by Hebb [29], Luria [30] and Fuster [31].

Another interesting result regarding localization is

that middle line locations (Fz, Cz, Pz) intervene in

both coherence variables. Studies on responses to

Figure 8. Interpolation of 19 t-scores for five frequency bands on a cortical topographic map. (*) indicates significant t-values (p

8/2/2019 2008 Qeeg Index

13/15

diverse paradigms in these locations, by means of

other neurophysiological techniques (see evoked

cognitive potentials, e.g. P300 or N400), have

shown their importance in processing information,

as well as their sensibility to diverse states of

pathological consciousness due to traumatic brain

injury (TBI) [32].

Measures involving frontal locations are also of

particular importance to the function. At least twofactors can help explain this. First, the proportion of

frontal lobe cortex is comparatively higher than that

of the other lobes. Therefore, it is probable that any

discriminant analysis will have more locations in this

area. Secondly, the frontal lobe is particularly

vulnerable to injury, especially in TBI cases, and

thus shows a higher number of irregularities in the

QEEG pattern. Neurophysiological studies have

found that the frontal regions of the brain show

greater measures of coherence than the posterior

regions [33, 34]. This could be due to the fact that

the frontal cortex favours long distance connections,

while the posterior cortex participates more in localprocesses.

In delta band coherence measures, a positive

correlation was found with the FIM FAM scores,

i.e. the higher the coherence in this band, the greater

the functionality measure. Beta and high beta

frequency bands correlated negatively with the

FIM FAM scores, i.e. the lower the coherence in

these bands, the lower the functionality. Both results

confirm previous findings related to the present EEG

pattern. They also support the idea that both

slow and fast frequencies bands must be considered

as a set in order to correctly interpret QEEG as a

measure of functionality.

Sensitivity, specificity and cross-validation of the

discriminant function

Results show that the discriminant function is

capable of clearly discriminating between different

functional states of dependence in ABI patientsduring post-acute phase and rehabilitation.

Moreover, the resulting function can classify these

patients with a high level of efficacy. It also has an

effective predictive capacity, as shown by its highly

accurate cross-validation. Results also confirm the

linearity of the discriminant function, which, accord-

ing to its indexes, classifies patients on a dimension

where the two extremes represent complete

dependence and complete independence and

whose mid-range values correspond to patients

with intermediate levels of functionality.

The obtained discriminant function in this study

offered 100% sensibility (i.e. [true positives (15)]/[false negatives (0)] [true positives (15)] (14/

15)*100 100%). Specificity also reached 100%

(i.e. [true negatives (14)]/[false positives (0)] [true

negatives (14)] (14/14)*100 100%). The loga-

rithm had a Positive Predictive Value (PPV) of

100%, meaning that all patients whose FIM FAM

scores indicated complete dependency were categor-

ized as such by the logarithm. In this case, the

logarithm had a Negative Predictive Value (NPV) of

100%, indicating that no patient was identified as

complete dependence when the QEEG is negative

Figure 9. (Left) Scatterplot of distribution and lineal adjustment between DS (y-axis) and global FIM FAM scores (n 80; training

sample external sample). The colour code represents the patients FIM FAM scores, red being the lowest score, green being the mid-

range and blue the highest. The correlation is R 0.826 (p

8/2/2019 2008 Qeeg Index

14/15

(the patient is not diagnosed as complete

dependence).

Cross-validation of the discriminant function was

done using the leaving-one-out method of classifica-

tion, with 95% accuracy. Moreover, all correlations

between discriminant scores and FIM FAM vari-

ables were significant, as were those between

discriminant scores and the predicted scores from

the multiple regression analysis. Finally, discrimi-

nant indexes were compared to the group of patients

with moderate dependence. Results showed, as

predicted, that this group was situated between the

independence and complete dependence scores.

This data shows a lineal relationship between

QEEG variables and the functional capacity of

patients with acquired brain injury.

Validity of QEEG as a tool for evaluating the functional

state of a patient with acquired brain injury

Finally, the aim of this study was to obtain an index

of the functionality of patients in post-acute phase.

Contrary to other studies, where functionality is

assessed in the acute phase [9, 35], this study focuses

on the post-acute phase, considering that brain

injury is a dynamic process which entails cerebral

restructuring, progress and deterioration. The focus

is also on rehabilitation and the potential for

recovery of each patient with acquired brain injury.

Consequently, the development of a discriminant

function was seen, with data on patients from the

post-acute phase, as more sensible and stable:

sensitive, because it reflects the current functionalstate of the patient with greater clarity and precision;

stable, because non-treated deficits that persist 6

months post-TBI normally are considered possible

sequelae. These assumptions are based on the big

bump theory of brain injury, which hypothesizes

that residual pathology or compensation after brain

injury could be detected months and years later

using QEEG [10].

Limitations of the discriminant function

Certain considerations should be taken into account

when using the discriminant function. First of all, itcan only be used on patients similar to those used

to design the function, that is, TBI or CVA patients

in post-acute phase (over 6 months post-injury).

Secondly, and no less important, the SINDI is for

use as a complement, not as a substitute, to

functional assessment.

Concluding remarks

The data clearly shows that the discriminant func-

tion obtained in this study is a tool capable of

discriminating between different functional states

of dependence in ABI patients during post-acute

phase and classifying them with a high level of

accuracy. The function offers 100% sensibility and

100% specificity, a PPV of 100%, a NPV of 100%

and a cross-validation of 95% accuracy. These

results attest to the functions usefulness in providinga QEEG index for the assessment of patients seeking

a diagnosis of their dependence state, which in turn

could be included in current functionality assess-

ment protocols.

Acknowledgements

Presented as an oral communication to the Society

of Applied Neuroscience Inaugural Meeting, 1319

September 2006, Swansea, Wales, UK.

This study was conducted in the Centre for

Brain Injury Rehabilitation C.RE.CER. in collabora-tion with the University of Seville.

Supported by the Ministry of Science and

Education as part of the National Plan for Scientific

Research, Development and Technological

Innovation (20042007) and co-funded by the

European Regional Development Fund (ERDF):

FIT-300100-2006-77.

References

1. Thurman DJ, Alverson C, Dunn KA, Guerrero J, Sniezek JE.

Traumatic brain injury in the United States: A public healthperspective. The Journal of Head Trauma Rehabilitation

1999;14:602616.

2. Mittenberg W, Patton C, Canyock EM, Condit DC. Base

rates of malingering and symptom exaggeration. Journal of

Clinical and Experimental Neuropsychology 2002;24:

10941102.

3. Larrabee GJ. Detection of malingering using atypical perfor-

mance patterns on standard neuropsychological tests. Clinical

Neuropsychology 2003;17:410425.

4. Lund TR, Sponheim SR, Iacono WG, Clementz BA. Internal

consistency reliability of resting EEG power spectra in

schizophrenic and normal subjects. Psychophysiology

1995;32:6671.

5. Arruda JE, Weiler MD, Valentino D, Willis WG, Rossi JS,

Stern RA, Gold SM, Costa L. A guide for applying principal-

components analysis and confirmatory factor analysis to

quantitative electroencephalogram data. International Journal

of Psychophysiology 1996;23:6381.

6. Corsi-Cabrera M, Solis-Ortiz S, Guevara MA. Stability of

EEG inter- and intrahemispheric correlation in women.

Electroencephalography and Clinical Neurophysiology

1997;102:248255.

7. Hughe JR, John ER. Conventional and quantitative

electroencephalography in psychiatry. The Journal of

Neuropsychiatry and Clinical Neurosciences 1999;11:190208.

8. John ER, Karmel BZ, Corning WC, Easton P, Brown D,

Ahn H, John M, Harmony T, Prichep L, Toro A, Gerson I,

Bartlett F, Thatcher F, Kaye H, Valdes P, Schwartz E.

Neurometrics. Science 1977;196:13931410.


8/2/2019 2008 Qeeg Index

15/15

9. Thatcher RW, North D, Biver C. EEG and intelligence:

relations between EEG coherence, EEG phase delay and

power. Clinical Neurophysiology 2005;116:21292141.

10. Thatcher RW, North DM, Curtin RT, Walker RA, Biver CJ,

Gomez JF, Salazar AM. An EEG severity index of traumatic

brain injury. The Journal of Neuropsychiatry and Clinical

Neurosciences 2001;13:7787.

11. Hall KM, Bushnik T, Lakisic-Kazazic B, Wright J,

Cantagallo A. Assessing traumatic brain injury outcomemeasures for long-term follow-up of community-based

individuals. Archives of Physical Medicine and

Rehabilitation 2001;82:367374.

12. Jennet B, Bond M. Assessment of outcome after severe brain

damage. Lancet 1975;1:480487.

13. Wilson JTL, Pettigrew LEL, Teasdale GM. Structured

interviews for the Glasgow Outcome Scale and the

Extended Glasgow Outcome Scale: guidelines for their use.

Journal of Neurotrauma 1998;15:573585.

14. van Baalen B, Odding E, van Woensel MP, Roebroeck ME.

Reliability and sensitivity to change of measurement instru-

ments used in a traumatic brain injury population. Clinical


15. Hall KM, Bushnik T, Lakisic-Kazazic B, Wright J,

Cantagallo A. Assessing traumatic brain injury outcomemeasures for long-term follow-up of community-based

individuals. Archives of Physical Medicine and


16. Ruijs MB, Gabreels FJ, Thijssen HM. The utility of

electroencephalography and cerebral computed tomography

in children with mild and moderately severe closed head

injuries. Neuropediatrics 1994;25:7377.

17. Kotchoubey B, Lang S, Mezger G, Schmalohr D,

Schneck M, Semmler A, Bostanov V, Birbaumer N.

Information processing in severe disorders of consciousness:

Vegetative state and minimally conscious state. Clinical

Neurophysiology 2005;116:24412453.

18. Hamilton BB, Laughlin JA, Granger CV, Kayton RM.

Interrater agreement of the seven-level Functional

Independence Measure (FIM). Archives of Physical

Medicine and Rehabilitation 1991;72:. p 790.

19. Linacre JM, Heinemann AW, Wright BD, Granger CV,

Hamilton BB. The structure and stability of the Functional

Independence Measure. Archives of Physical Medicine and


20. Doods TA, Matrin DP, Stolov WC, Deyo RA.

A validation of the Functional Independence Measurement

and its performance among rehabilitation inpatients.

Archives of Physical Medicine and Rehabilitation

1993;74:531536.

21. Jasper HH. The ten-twenty electrode system of the

International Federation. Electroencephalography and

Clinical Neurophysiology 1958;10:371375.

22. Delorme A, Makeig S. EEGLAB: an open source toolbox

for analysis of single-trial EEG dynamics. Journal of

Neuroscience Methods 2004;134:921.

23. Srivastava G, Crottaz-Herbette S, Lau KM, Glover GH,

Menon V. ICA-based procedures for removing ballistocar-

diogram artifacts from EEG data acquired in the MRI

scanner. Neuro Image 2005;24:5060.

24. Thornton K. The electrophysiological effects of a brain injury

on auditory memory functioning. The QEEG correlates ofimpaired memory. Archives of Clinical Neuropsychology

2003;18:363378.

25. Lubar JF. Neocortical dynamics: Implications for under-

standing the role of neurofeedback and related techniques for

the enhancement of attention. Applied Psychophysiology and

Biofeedback 1997;22:111126.

26. Thatcher RW, Biver C, McAlaster R, Salazar A. Biophysical

linkage between MRI and EEG coherence in traumatic brain

injury. NeuroImage 1998;8:307326.

27. Leon-Carrion J, Martin-Rodriguez JF, Damas-Lopez J,

Barroso y Martin JM, Dominguez-Morales R Brain func-

tion in minimally conscious state: A qEEG study. Clinical

Neurophysiology.

28. Bricolo A, Turella G. Electroencephalographic patterns of

acute traumatic coma: Diagnostic and prognostic value.Journal of Neurosurgical Sciences 1973;17:278285.

29. Hebb DO. The organization of behaviour. New York: Wiley;

1949. pp 6370.

30. Luria AR. Human brain and psychological processes.

New York: Harper and Row; 1966.

31. Fuster JM. The cognit: A network model of cortical

representation. International Journal of Psychophysiology

2006;60:125132.

32. Neumann N, Kotchoubey B. Assessment of cognitive

functions in severely paralysed and severely brain-damaged

patients: neuropsychological and electrophysiological meth-

ods. Brain Research. Brain Research Protocols

2004;14:2536.

33. Thatcher RW, Krause PJ, Hrybyk M. Cortico-cortical

associations and EEG coherence: A two-compartmental model. Electroencephalography and Clinical

Neurophysiology 1986;64:123143.

34. Tucker DM, Roth DL, Bair TB. Functional connections

among cortical regions: Topography of EEG coherence.

Electroencephalography and Clinical Neurophysiology

1986;63:242250.

35. Finnigan SP, Rose SE, Walsh M, Griffin M, Janke AL,

McMahon KL, Gillies R, Strudwick MW, Pettigrew CM,

Semple J, Brown J, Brown P, Chalk JB. Correlation of

quantitative EEG in acute ischemic stroke with 30-day

NIHSS score: Comparison with diffusion and perfusion

MRI. Stroke 2004;35:899903.


Documents

2008 Qeeg Index