Cerebral palsy classification using heuristics and belief decision tree: A preliminary study

Abstract

The objective of this present study was to classify two types of spastic cerebral palsy pathology such as diplegia and hemiplegia with belief labels. The discrimination process aims to predict new issues of cerebral palsy in case of uncertain classification. The Belief theory was applied to perform uncertain and imprecise classification. Following data were provided by the Bois Larris Center: Clinical kinematics, kinetics, and muscle activity parameters of the cerebral palsy were acquired motion analysis, forces platforms and Electromyography (EMG) system. An extraction parameter process was developed to reveal significant characters from kinematics, kinetics, and EMG curves. A heuristics-based belief assignment method was developed to distribute the mass function of each subject based on theirs extracted parameters. Belief decision tree method was used to develop uncertain classification model. A preliminary clinical application of 10 subjects (5 diplegias and 5 hemiplegias) was performed. Significant kinematics and kinetics parameters of the cerebral palsy such as ground reaction forces, contact time between the ground and the foot, maximal tension muscle and EMG-based average rectified voltage of the muscles were reported and discussed. Preliminary clinical findings of cerebral palsy classification were addressed in order to help clinicians in their diagnosis, decision-makings, and communications. We showed that Belief formalism was a universal formalism to classify cerebral palsy subjects with belief level. © Springer-Verlag 2010.