Publication

Abstract : Neonatal jaundice is a common disease that develops as a result of high bilirubin, which, in case of unidentified diagnosis, may lead to severe complications. Traditional methodologies total serumbilirubin (TSB) test are invasive blooddraws, which may be stressful to the neonates. The system proposes a deep learning system to measure the bilirubin concentration in the images of head, face, and chest using digital cameras with the help of convolutional neural networks(CNNs)to measure the bilirubin level in the images. The system gained experience with the Neo-Jaundice data set labeled by pediatricians. Using jaundice pattern analysis, the system offers a effective, non-invasive screening, eliminating the need of repeated blood draws. It increases early diagnosis, improves convenience in the neonatal care settings and it offers a cheaper, scalable alternative to existing diagnostic technology Owing to the integration of deeplearning, a solution that is capable of screening jaundice can be automated, alleviate suffering of infants and allow prompt medical care, enhancing overall neonatal health outcomes.