Noninvasive bioelectrical neuromonitoring in anaesthesia and critical care

Abstract

In the past few years we have learned more about the brain in context with anaesthesia than ever before. But the most complex organ of all continues to hold immense mysteries. Shedding light on these will be crucial for monitoring purposes and for preventing, diagnosing and treating neurological and mental illnesses. Neuromonitoring is a discipline with enormous potential and should be an essential part of modern anaesthesia and critical care [1]. Compressed spectral array (CSA) is one of the bioelectrical methods often used today. From the raw electroencephalogram (EEG) it is possible to calculate an EEG power spectrum using fast Fourier analysis. If this is performed repeatedly CSA is obtained. From the EEG power spectrum calculation of different parameters such as the peak frequency (or peak-power frequency), the frequency at the centre of gravity, the mean dominant frequency or the well-known SMF (mean power frequency) can be performed. Last but not least, there is the important spectral edge frequency (SEF) parameter. SEF90 is the frequency below which 90% of the power in the spectrum resides. We would like to point out that a single feature such as mean power frequency or spectral edge frequency may not be sensitive to all possible changes in spectral distribution. However, there is no evidence, at present, suggesting that additional parameters describing a complex spectrum improve the clinical utility of simple univariate parameters. Different manufacturers define different spectral edge frequencies, for example SEF90 or SEF95. But even if the SEFs are defined at exactly the same level, the same bioelectrical signal can lead to different results. The reason for this is that we have to take into account the different technical parameters, such as the upper and especially the lower cut-off frequency. A comparison of the different SEFs in the literature is therefore not possible, if different systems are used. The number of publications on the topic of noninvasive neuromonitoring has increased exponentially (Figure 1). Note the number of papers published on the bispectral index (BIS), an EEG parameter used to monitor the depth of anesthesia. Results similar to those in Figure 1 are produced by the Internet search engine ISI Web of Science®. Of 1674 178 listed documents in the years 2000 and 2001, 37 matched the term 'neuromonitoring' and 136 the term 'bispectral index'.Figure 1 .: Numbers of scientific publications from January 1975 to September 2001. Source: MEDLINE; Search: 'Neuromonitoring' and 'Bispectral Index'.Multivariate statistical methods were used to combine different EEG features into an easy-to-use number called BIS, which ranges from 100 – indicating that the patient is awake – to zero, indicating a total lack of brain activity [2]. At induction of anaesthesia the BIS falls rapidly from an awake value (80–100) to below 40. Tracheal intubation and skin incision result in an increase in BIS. Bispectral analysis originally had nothing to do with anaesthesia. Bispectral analysis is a mathematical procedure used for describing ocean waves. This was described by Hasselman and his colleagues in 1963 [3]. Like almost all technical developments, this one was first used in the military area for the analysis of sonar and radar signals in military tracking systems. One of the first clinical uses of bispectral analysis was mentioned by Barnett and colleagues in 1971 [4]. His group used the bispectral analysis of EEG signals during waking and sleep. This brings us quite close to anaesthesia in terms of topic. In 1991, Kearse and colleagues published a contribution in Anesthesiology entitled 'Bispectral analysis may predict anesthetic depth during narcotic induction' [5]. As a consequence, multicentre studies such of that of Peter Sebel in 1997 were published [6]. The bispectrum is calculated in a two-dimensional space of frequency1 (f1) vs. frequency2 (f2). A strong relationship between f1 and f2 creates a large bispectral