Preliminary feasibility study of anesthetic agent identification from ion mobility spectra using a neural network

Anesthetic agent monitoring and identification have become accepted standard practice, to improve safety in the operating room. A simple and cost effective device called an aspiration condenser has been used for ion mobility analysis in a variety of gas monitoring applications. This device has the potential to become a competitive alternative to existing anesthesia monitoring technologies. This paper is a preliminary feasibility study of anesthetic agent identification, using an Artificial Neural Network (ANN), from ion mobility spectra obtained using an aspiration condenser. Spectra of the five commonly used volatile anesthetic agents (Halothane, Isoflurane, Enflurane, Sevoflurane, and Desflurane), in varying concentrations, were used to train and test an ANN. Preliminary results show an overall agent identification accuracy was 75% (concentration⇐1.0%) and 98.21% (concentration>1.0%). With further refinement of the measurement and analysis approach, a low-cost clinical anesthesia gas monitor can be developed