Neonatal Jaundice Diagnosis Using a Smartphone Camera Based on Eye, Skin, and Fused Features with Transfer Learning

Abstract

Neonatal jaundice is a common condition worldwide. Failure of timely diagnosis and treatment can lead to death or brain injury. Current diagnostic approaches include a painful and time-consuming invasive blood test and non-invasive tests using costly transcutaneous bilirubinometers. Since periodic monitoring is crucial, multiple efforts have been made to develop non-invasive diagnostic tools using a smartphone camera. However, existing works rely either on skin or eye images using statistical or traditional machine learning methods. In this paper, we adopt a deep transfer learning approach based on eye, skin, and fused images. We also trained well-known traditional machine learning models, including multi-layer perceptron (MLP), support vector machine (SVM), decision tree (DT), and random forest (RF), and compared their performance with that of the transfer learning model. We collected our dataset using a smartphone camera. Moreover, unlike most of the existing contributions, we report accuracy, precision, recall, f-score, and area under the curve (AUC) for all the experiments and analyzed their significance statistically. Our results indicate that the transfer learning model performed the best with skin images, while traditional models achieved the best performance with eyes and fused features. Further, we found that the transfer learning model with skin features performed comparably to the MLP model with eye features.

Management challenges in the treatment of severe hyperbilirubinemia in low- and middle-income countries: Encouraging advancements, remaining gaps, and future opportunities

“…Another study in China demonstrated decreased neonatal readmission rate and decreased maternal anxiety scores for mother-infant dyads that used a smartphone app to monitor for neonatal jaundice at home under web-based guidance of pediatricians compared to the control group that received routine care ( 108 ). Similar smartphone-based detection tools are being studied in other regions of the world, including Singapore ( 109 ), Ghana ( 110 ), and Saudi Arabia ( 111 ), and Nigeria ( 112 ), just to name a few. Although these tools are best suited as screening devices, as with the TcB, they are encouraging uses of state-of-the-art technology to improve the recognition and diagnosis of NJ.…”

A Non-invasive Methods for Neonatal Jaundice Detection and Monitoring to Assess Bilirubin Level: A Review

“…Althnian et al [25] propose a thorough exchange learning technique based on eye, skin, and merged images. In addition, they created conventional machine learning models, such as multi-layer perceptron (MLP), decision tree (DT), support vector machine (SVM), and random forest (RF), and compared their performance to that of the exchange learning signal.…”

Validation of a smartphone‐based screening tool (Biliscan) for neonatal jaundice in a multi‐ethnic neonatal population

“…Smartphone-based applications [17][18][19][20][21][22][23][24][25][26][27] that estimate the severity of jaundice raise the possibility of remote monitoring of neonatal jaundice (see Table 1). The 'Biliscan (Shenzhen Beishen Healthcare Technology Co, Shen Zhen, People's Republic of China)' is one such application.…”