In Response

    loading  Checking for direct PDF access through Ovid

Excerpt

The authors appreciate the constructive comments from Kreuzer, Schneider, and García1 in response to our article, “A Procedural Electroencephalogram Simulator for Evaluation of Anesthesia Monitors,”2 and welcome the opportunity to clarify the reasoning behind using a simple simulated signal to evaluate anesthesia depth of hypnosis monitors.
By way of example, the notion that anesthesia monitors can only be tested with an actual electroencephalogram signal is equivalent to thinking that professional audio-recording equipment for a concert hall can only be evaluated by performing a symphony. An audio engineer would find this position at odds with accepted practice. Indeed, it is well established to test high-fidelity audio equipment with the simplest possible signal: a pure constant tone.3 Nothing could be further removed from the rich polyphonic harmonies of a concert performance that it is designed to record. Yet, the very simplicity of the test signal exposes the inner workings of the equipment and allows the engineer to assess noise, linearity, and other characteristics. These findings apply equally to the equipment’s performance on a complex live signal. In a broader context, the notion of using simple probe signals to characterize unknown entities is a cornerstone to modern test and measurement engineering practice.4
Returning to the present case of evaluating electroencephalogram anesthesia monitors; a constant pure tone is too simplistic as a probe signal, because the monitors will reject it as an artifact. There are many ways to overcome this. We opted to use Perlin noise5 because this is a simple function widely used for generating realistic 1/f output common to a variety of critical physical systems that in certain ways may be similar to the brain (indeed recent work suggests that criticality may be a signature of the healthy brain6). It is very possible that this choice is not optimal; however, it does offer a simple test signal bearing some resemblance to what an anesthesia monitor is designed to operate on.
It is our belief that establishing an unbiased way to evaluate anesthesia monitors based on sound engineering principles is essential to develop and gain confidence in a quantified measure of anesthetic depth, which is necessary for the progress of anesthesia. Maintaining anesthesia monitors as undocumented black boxes, which do mysterious things only on live bioelectric signals, is not helpful in advancing the current state of the art. Our article is meant to raise debate around this issue, and encourage the community to look to other fields where solutions may already exist and readily be adopted.
Regarding the number of simulation channels, the presented audio-based implementation does produce a bilateral signal adequate for driving most anesthesia monitors. We chose to deliver the signal through audio because of the ubiquitous 2-channel audio output port on mobile devices. In special cases where more than 2 channels are needed, it is straightforward to design dedicated hardware based on the same principles presented in our article.
Finally, the authors agree that burst suppression should also be part of a comprehensive evaluation of anesthesia monitors. We continue to work on this, as mentioned in the future work section of our article. It is challenging, however, because burst suppression is poorly defined and difficult even for experts to accurately quantify. We are currently exploring possible new solutions to this problem within our research team.
    loading  Loading Related Articles