1. Inicio
  2. Atrás
Título: An expert system for EEG monitoring in the pediatric ICU
Autores: Pasupathy, Anitha K.
Fecha: 1994
Publicador: McGill University - MCGILL
Fuente:
Tipo: Electronic Thesis or Dissertation
Tema: Biology, Neuroscience.
Engineering, Biomedical.
Engineering, Electronics and Electrical.
Artificial Intelligence.
Descripción: A knowledge-based expert system was developed to assess the level of abnormality in the brain electrical activity of pediatric patients monitored in the intensive care unit. Six hours of an 8-channel EEG record serves as the input to the monitoring device based on which the brain activity is classified as being normal, mildly abnormal, moderately abnormal or severely abnormal.
Spectral band activity is computed for each channel for every 30-second epoch. Artifact rejection is accomplished by a median filter with a hard-limiter thresholder. Quantitative variables reflecting possible abnormality: a measure of amplitude depression, a measure of assymmetry, a measure of anterio-posterior differentiation and a measure of EEG variability over time are extracted from each EEG record. Statistical distributions of these measures are established for a control "normal" population of about ten patients so classified by a neurologist on visual interpretation. New EEGs to be analysed are statistically compared with the control population and a probability measures of normality for the various measures are determined. The expert system learns from prior examples of classification done by the neurologist by a technique of inductive machine learning. The monitor is trained and tested using sixty examples using the rotation method of error estimation.
The monitor had a tendency to classify the EEGs with a higher level of abnormality than the expert. Possible reasons and potential solutions are discussed.
Idioma: en