Reverse Bi-orthogonal wavelets fuzzy classifiers for the automatic detection of spike waves in the EEG of the hypoxic ischemic pre-term fetal sheep

摘要

There exists a 6-8 hour window of opportunity for the treatment of perinatal Hypoxic-Ischemic Encephalopathy (HIE) following the original insult after which significant irreversible brain injury manifests leading to debilitating neurological conditions such as epilepsy and cerebral palsy. At present, there are no identified biomarkers in the electroencephalogram (EEG) that are currently being used to help classify if a HIE insult has occurred or not. However, high frequency micro-scale transients in the form of spikes, sharp waves and slow waves appear in the EEG, post insult, that could provide precursory information whether a HIE insult has occurred or not. This paper describes the superiority of using reverse bi-orthogonal wavelets (RBIO-WT), in the form of the rbio2.8 mother wavelet, in conjunction with a Type-1 Fuzzy Logic System (Type-I FLS) classifier for accurate micro-scale spike wave transient detection in the EEG of Pre-term Fetal Sheep. The algorithm performance for spike detection was assessed over the most critical time period of 25 minutes within the first 8 hours, post occlusion using an in utero fetal sheep model. Obtained results demonstrate that the suggested algorithm detected spikes with a considerably high overall performance of 99.25% using the developed RBIO-WT Type-I FLS.