Klinische Neurophysiologie 2010; 41 - ID182
DOI: 10.1055/s-0030-1251011

Intraoperative protection of the facial nerve by automated categorisation of EMG-activity

J Prell 1, C Strauss 1, S Rampp 1
  • 1Universität Halle-Wittenberg, Neurochirurgische Klinik, Halle, Deutschland

Background:

In intraoperative monitoring of the facial nerve during vestibular schwannoma surgery, the A-train is of paramount importance. This specific, rhythmic high-frequency pattern allows prognostic statements concerning postoperative facial nerve function; a statistic correlation of high significance between the amount of A-trains and the postoperative degree of functional impairment was demonstrated. However, false positive and false negative results still limit reliable application of the technique in individual cases.

Methods:

The project presented is based on a system developed for fully automated intraoperative quantification of A-trains in real-time. The cumulated amount of A-trains is expressed as „traintime“, which correlates closely with postoperative facial nerve function (sensitivity and specificity of 80% concerning prognosis of severe facial nerve paresis of House&Brackman grade 4+). In order to improve the method beyond this point, the number of analyzed channels was raised to 12 as the first step.

Results:

The raised number of channels leads to a significant improvement of the prognostic capabilities regarding indication of severe paresis. To show this, the contrast of measured, relative traintime between the groups of patients with, respectively without severe postoperative paresis (House&Brackmann grade 4+) was evaluated and correlated with the number of channels incorporated. A highly significant correlation between number of channels and discriminatory power concerning these two patient groups was demonstrated (Spearman's Rho 0.98, p<0.0001).

Discussion:

A raised number of channels allows for a broader and, thus, improved sample of assumed overall A-train activity. Consequently, this results in improved prognostic power. Further improvements, especially concerning specificity, may be achieved by categorisation of detected A-trains in terms of frequency, length and channel distribution. Specific algorithms fort his kind of analysis are currently implemented.