Automatic detection of artifact in neonatal ECG

Heart Rate Variability derived from the Neonatal Electrocardiogram has been found to be associated with the Electroencephalography grade of Hypoxic Ischemic Encephalopathy and neurodevelopmental outcome. This association has been established for clean or artifact free ECG. However, it was shown that the Electrocardiogram and subsequently Heart Rate Variability features can be heavily corrupted by artifacts which have to be manually removed. This work combines a set of statistical features to quantify the quality of the HR signal by automatically detecting the artifacts in neonatal ECG. The HRV signal is obtained by detecting R-Peaks using the adapted Pan-Tompkins algorithm. Four features are extracted from HR signal to discriminate normal and corrupted signal. The performance of these features in discrimination is then assessed using statistical tests. It has been shown that there is a significant difference of proposed features between artifact and normal signals (p<;0.001). The discrimination power is increased by combing the current features using Support Vector Machine. The median AUC was 0.9941 (IQR: 0.98-1.00).