Elsevier

Clinical Neurophysiology

Volume 118, Issue 11, November 2007, Pages 2362-2367
Clinical Neurophysiology

The revised brain symmetry index

https://doi.org/10.1016/j.clinph.2007.07.019Get rights and content

Abstract

Objective

Recently, the extended brain symmetry index (BSI) was introduced to assist the visual interpretation of the EEG, in particular to quantify both the spatial (left–right) and the temporal spectral characteristics. The BSI has found application in monitoring during carotid endarterectomy, acute stroke and focal seizure detection. Here, we present additional relevant characteristics and a slightly modified version of this index, simulating its behavior as may occur in various clinical conditions, with an emphasis on the detection of cerebral ischaemia.

Methods

The behavior of the revised and standard sBSI and tBSI is illustrated using random noise signals to simulate various changes in the EEG. The indices are evaluated as a function of spatial and temporal changes, and as a function of the number of channels.

Results

The r-sBSI and the r-tBSI are normalized in the range [0–1] with sensitivities of about 0.05 for a 10% difference in signal amplitude, either spatial or temporal. The baseline value of the sBSI shows a modest dependence on the number of channels used.

Conclusions

The revised BSI has an improved sensitivity (about two times) to detect interhemispheric asymmetry and diffuse changes. The modified expression of the tBSI is more compact and allows a more intuitive understanding than previously proposed.

Significance

qEEG assists in a more objective interpretation of the EEG, and is relevant in neuromonitoring.

Introduction

Symmetry is ubiquitous in nature and an important property in the evaluation of various clinical conditions. For example, in patients with one-sided complaints, the clinical examination is typically facilitated by an explicit comparison of the left and right side (line symmetry). In clinical neurophysiology, symmetry plays a significant role, as well. It is rather common to compare if significant differences of a measurand exist between the symptomatic and the asymptomatic side, e.g. in the estimation of motor or sensory nerve action potentials (Levin and Lüders, 2000). Line symmetry exists as well between the cerebral hemispheres. This holds not only for the gross anatomy, but also applies to particular properties of the human electroencephalogram (EEG). Although various EEG rhythms may show a modest physiological asymmetry (Niedermeyer and Lopes da Silva, 1999, Weber, 2005), the mean spectral characteristics of the left and right hemispheric EEG, as estimated during a particular time frame, are nearly symmetrical (van Putten et al., 2004, van Putten and Tavy, 2004, Niedermeyer and Lopes da Silva, 1999), notwithstanding various significant differences in functional lateralization.

Evaluation of left–right symmetry, therefore, is an important characteristic of the EEG, for instance in the decision for selective shunting during carotid endarterectomy. Several EEG features have been proposed, including changes in relative power or spectral edge frequency (Hanowell et al., 1992, Minicucci et al., 2000, Laman et al., 2001, Laman et al., 2005, Cursi et al., 2005). Recently, the brain symmetry index (BSI) was introduced as a measure to quantify the interhemispheric spectral symmetry of the EEG (van Putten et al., 2004). This feature was originally designed to assist in the visual interpretation of the EEG during carotid endarterectomies, but has found applications in monitoring stroke patients, and detection of focal seizure activity, as well (van Putten and Tavy, 2004, van Putten et al., 2005). An extension to this index included a measure to capture diffuse changes (which could be regarded as differences in time symmetry), effectively resulting in two features, the standard BSI (sBSI), quantifying interhemispheric asymmetry, and the temporal BSI (tBSI), quantifying diffuse changes (van Putten, 2006).

Recent insights and improvements in the algorithm, leaving the original concept unchanged, motivatedc us to report in more detail on various characteristics of these two indices. Using artificially generated EEG signals, we simulate various conditions as may occur during ischaemia, for instance in carotid surgery.

Section snippets

Quantifying left–right symmetry

The (standard) brain symmetry index (sBSI) quantifies the difference between the spectral characteristics of the left and right hemisphere. For each discrete time series Xch,t with t discrete time, t = 1, 2,  , N recorded from a bipolar derivation or channel ch = 1, 2,  , M, we write for the discrete Fourier transformXch,t=i=1KAicos(ωit)+Bisin(ωit)with K the number of discrete frequencies, ωi. This is the usual model for a process with a purely discrete spectrum (harmonic model) (Priestley, 1981).

The

Amplitude or power spectrum

In Fig. 1a we illustrate the difference in sensitivity between the sBSI as based on the ratio of the spectral amplitudes and the r-sBSI as based on the ratio of the spectral power, both as a function of the amplitude ratio β = Ri/Li. In a typical range of interest, where the amplitude ratio Ri/Li  2, we find that this ratio is ∼2, as well. The general expression of the differences in sensitivity between the sBSI and its revision is given by (1+β)21+β2 (Eq. (9)).

If symmetry is present, within the

Discussion and conclusions

The BSI was originally proposed to quantify hemispheric asymmetry, as may occur during carotid surgery (van Putten et al., 2004). By now, various additional applications include monitoring of stroke patients (van Putten and Tavy, 2004), detection of focal seizure activity (van Putten et al., 2005) and the detection of diffuse EEG changes in ischaemia (van Putten, 2006). The primary motivation for the current paper was to present a detailed description of a modified and improved algorithm, with

Note added in the proof

If γ is erroneously estimated too large, an increase in asymmetry may result in a reduction of the value of the r-tBSI. It is recommended, therefore, to use a conservative value, and accept a small residual.

References (16)

There are more references available in the full text version of this article.

Cited by (99)

  • Perioperative monitoring of brain function, much more than “hypnometry”

    2023, Revista Espanola de Anestesiologia y Reanimacion
  • In vivo stiffness of multiple sclerosis lesions is similar to that of normal-appearing white matter

    2022, Acta Biomaterialia
    Citation Excerpt :

    While 35% of the lesions exceeded the visual delineation limit of 10%, none was apparent to the eye compared to the contralateral site. The symmetry of brain hemispheres is typically used as a criterion for the detection of focal brain abnormalities in addition to the contrast provided by surrounding tissue [55,56]. However, the symmetry of brain MRE maps is frequently compromised by the presence of hotspots, which are often due to wave scattering at tissue boundaries and pose another challenge for lesion detection based on WM stiffness.

  • Stroke identification using a portable EEG device – A pilot study

    2020, Neurophysiologie Clinique
    Citation Excerpt :

    With a Hamming window of 256 samples (at 220 Hz recording rate) and a slide window of 22 samples over (1/10th of a second) we allowed a 90% overlap from one window to the next. This data is transmitted by the MUSE device at 10 Hz to be used by the rsBSI calculation algorithm suggested by Van Putten [10]. The Revised Brain Symmetry Index (rsBSI) was introduced as an improved version of the Brain Symmetry Index for continuous quantitative EEG monitoring in hemispheric stroke patients [10].

View all citing articles on Scopus
View full text