Many preterm infants suffer from neural disorders caused by early birth\ncomplications. The detection of children with neurological risk is an important\nchallenge. The electroencephalogram is an important technique for establishing\nlong-term neurological prognosis. Within this scope, the goal of\nthis study is to propose an automatic detection of abnormal preterm babiesâ??\nelectroencephalograms (EEG). A corpus of 316 neonatal EEG recordings of\n100 infants born after less than 35 weeks of gestation were preprocessed and a\ntime series of standard deviation was computed. This time series was thresholded\nto detect Inter Burst Intervals (IBI). Temporal features were extracted\nfrom bursts and IBI. Feature selection was carried out with classification\nin one step so as to select the best combination of features in terms of\nclassification performance. Two classifiers were tested: Multiple Linear Regressions\nand Support Vector Machines (SVM). Performance was computed\nusing cross validations. Methods were validated on a corpus of 100 infants\nwith no serious brain damage. The Multiple Linear Regression method shows\nthe best results with a sensitivity of................
Loading....