http://www.iaeme.com/IJEET/index.asp 248 editor@iaeme.com
International Journal of Electrical Engineering and Technology (IJEET)
Volume 11, Issue 3, May 2020, pp. 248-264, Article ID: IJEET_11_03_029
Available online at http://www.iaeme.com/IJEET/issues.asp?JType=IJEET&VType=11&IType=3
ISSN Print: 0976-6545 and ISSN Online: 0976-6553
Journal Impact Factor (2020): 10.1935 (Calculated by GISI) www.jifactor.com
© IAEME Publication
NEONATAL JAUNDICE DETECTION SYSTEM
USING CNN ALGORITHM AND IMAGE
PROCESSING
Ashish Chakraborty
School of Electronics and Communication Engineering,
Vellore Institute of Technology University, Vellore, India
Sushil Goud
School of Electronics and communication Engineering
Vellore Institute of Technology University, Vellore, India
Vandita Shetty
School of Electronics and Communication Engineering,
Vellore Institute of Technology University, Vellore, India
Budhaditya Bhattacharyya
School of Electronics and Communication Engineering,
Vellore Institute of Technology University, Vellore, India
ABSTRACT
Neonatal hyperbilirubinemia or jaundice is a common health condition in newborn
infants because of changes in erythrocyte metabolism in the first week of life itself. It is
a multifactorial disorder with many symptoms. With today’s technological
advancements, we have both invasive and non - invasive systems to facilitate early
neonatal jaundice detection and subsequent treatment at the early stages itself. In this
paper, we shall discuss our proposed non-invasive neonatal jaundice detection system
using CNN algorithm. The various detection systems stated in this paper provides the
accuracy of the method and feasibility when it comes to the implementation. All
methodologies and detection techniques discussed here provide real-world insight and
helps is early detection of neonatal jaundice. These include the use of Support Vector
Machines i.e. the SVM with image processing technique that helps in reading different
bilirubin levels in the baby at the time of disease. The first regressions which was used
was generally linear, but SVR algorithm was non-linear. When determining the
relationships between linear relationships, generally, Support Vector Regressions were
used. The aim of the regression was finding a linear regression function in a high
dimensional feature space. Then, input data was mapped to the space with using the
potential non-linear function. Color card is another detection method used wherein,
based on the skin and eye’s coloration and comparing with the color cards developed,