Non-invasive Technique for Detecting Neonatal Jaundice

Abstract

Neonatal jaundice is a common ailment observed in infants. There is a lot of diagnostic approaches being made to detect it currently. When the infant’s bilirubin level goes beyond 5 mg/L (85 mmol/L) it is indicative of jaundice. Bilirubin level increases often in infants just after birth till about a week. Almost 60 and 80% of full-term and premature babies, respectively, have jaundice. This paper aims for the early detection of jaundice in newly born infants. A total of 15 jaundice affected and 22 normal infants were considered for the purpose of this study, including babies of varied skin tones. A smartphone camera was used to capture images of the infant’s skin. MATLAB was used to process the images that were taken, using algorithms for skin and facial detection. Color map transformations like RGB and YCbCr were used. Multiple quantitative features such as energy, entropy, mean, standard deviation and skewness were studied. PCA principle is the main technique that is used to reduce data redundancy. PCA uses energy and other features to come to a well-defined conclusion. It was noticed that the standard deviation and energy are much higher in jaundice-affected infants than normal ones. This research concludes that images can be used to identify the occurrence of neonatal jaundice.