Intracranial pressure processing with artificial neural networks: Classification of signal properties

Intracranial pressure (ICP) is commonly used by neurosurgeons as a source o f valuable information about the current condition of the neurosurgical pat ient. Nevertheless, despite years of effort, extracting clinically valuable information from the ICP signal is still problematical. Approaches, using current values of ICP, may fail to disclose imminent risk, because unpredic table factors can rapidly change the properties of the signal. An alternati ve approach is to determine some global characteristics of the signal withi n a longer time interval and such statistical analyses have been proposed b y several authors. A further, rarely considered, problem is assessment of t he results obtained from the point of view of their practical utility and/o r such classification of the obtained properties of the signal that they co rrespond to certain clinical stales of the patient. While this might be a t ypical task for discriminant analysis, we approached the analysis using an alternative methodology, that of computational intelligence, implemented in artificial neural networks (ANN). We tested two variants of the ANN algorithms for classification and discrim ination of global properties of the ICP signal. In a "dynamic pattern class ification" the network was presented with several sections of ICP records t ogether with information from the expert-neurosurgeon, classifying 4 risk g roups. In this mode no data preprocessing was carried out, in contrast to o ur second approach, in which the signal had been pre-processed using publis hed statistical analyses and only these intermediate coefficients were fed into the ANN classifier. The results obtained with both classification methods at their current stag e of training were similar and approximated to a 70% rate of judgements con sistent with the expert scoring. Nevertheless, the method based on the asse ssment of global parameters from the ICP record looks more promising, becau se it leaves the possibility for modification of the set of parameters anal ysed. The new parameters may include information extracted not only from th e ICP signal, but also from other diagnostic modalities, like colour coded Doppler ultrasonography. The ultimate goal of this work is to build up a pseudo-intelligent computer expert system, which would be able to reason from a reduced set of input i nformation, available from a standard monitoring modality, because it had b een taught salient links between these data and higher-order data, upon whi ch expert scoring was based.