Abstract:
Newborn jaundice (hyperbilirubinemia), which is seen in 65% of healthy newborns, is usually a harmless condition and passes without any treatment. However, 5% of the jaundiced babies develop kernictures which may cause irreversible brain damage. Therefore, detection and follow-up of jaundiced babies are very important. The most common method for the evaluation and follow-up of neonatal jaundice is measuring the level of total serum bilirubin (TSB). Transcutaneous bilirubin (TcB) measurement, which is an alternative for jaundice detection and monitoring, is a safe, easy, painless, cost-effective and fast method as well as being noninvasive. The application of TcB measurement uses visible field reflection spectroscopy to determine the level of jaundice in newborns. The aim of this thesis is to design algorithms which can determine the level of jaundice by nonlinear approaches using diffuse reflection spectra obtained from a specially developed device. In order to achieve this, TcB and TSB measurements were taken from 314 infants and prediction algorithms were developed. The TcB values predicted by the designed algorithms were compared with the TSB values of the paired measurement. In addition, the performance of the designed algorithms was compared with the widely used JM-103 transcutaneous bilirubinometer. All designed algorithms can predict jaundice level with correlation values between 0.932 and 0.943. Theresultsofthecomparisonswerefoundtobepromisingfortheuseoftranscutaneous bilirubinometer devices.|Keywords : Newborn Jaundice, Transcutaneous Bilirubinometry, Nonlinear Regression Methods, Hyperbilirubinemia.