Development and Evaluation of a Smartphone Camera Method for Transcutaneous Bilirubin Measurements
Student thesis: Master Thesis and HD Thesis
- Sarah Bach Munkholm
5. Term (Master thesis), Medicine, Master (Master Programme)
Introduction
Worldwide, neonatal hyperbilirubinemia remains a common cause of readmission to the healthcare system, as it may cause kernicterus at high levels. This condition is treatable, why an easy and accessible screening tool, which enables self-measurement of bilirubin, is valuable and advantageous. However, such device is not yet available. Hence, we aim to examine if the embedded camera on a smartphone can be a reliable tool in order to screen and monitor for neonatal hyperbilirubinemia.
Methods
A total of 49 Caucasian neonates admitted to Aalborg University Hospital were randomly selected for the study between 15th of July and the 3rd of December 2015. Three methods were evaluated using an iPhone 6 for the assessment. In the first method, we simply used the camera on the smartphone to apply moderate pressure on the glabella of each neonate while recording a video or taking a picture. Secondly, we used a Dermatoscope (DermaLite Pro II) attached to the iPhone to obtain each video. Thirdly, we applied a Wratten No. 11 Colour Filter between the DermaLite Pro II and the camera on the iPhone 6. A transcutaneous bilirubinometer (TcB) or a total serum bilirubin measurement was used as reference. The red, green and blue colour intensities of each video were analysed and correlated to the respective bilirubin reference values.
Results
No significant correlation between the iPhone 6 analysis and reference bilirubin levels using the first and second method was observed (Pearson´s correlation coefficient; red channel 0.05 and -0.38; green channel -0.03 and -0.05; blue channel -0.11 and 0.01). However, method three revealed a significant negative Pearson´s correlation coefficient of 0.61 (p = 0.015) between the blue channel intensities and the TcB with a coefficient of variance of 22.5 %.
Conclusion
The presented methods demonstrate differential results. Method three did, however, indicate a proof of concept and therefore we believe that a combination of standardised light, even pressure and a filter excluding the reflectance of haemoglobin along with further improvements and studies, can be useful in the screening process of neonatal hyperbilirubinemia.
Worldwide, neonatal hyperbilirubinemia remains a common cause of readmission to the healthcare system, as it may cause kernicterus at high levels. This condition is treatable, why an easy and accessible screening tool, which enables self-measurement of bilirubin, is valuable and advantageous. However, such device is not yet available. Hence, we aim to examine if the embedded camera on a smartphone can be a reliable tool in order to screen and monitor for neonatal hyperbilirubinemia.
Methods
A total of 49 Caucasian neonates admitted to Aalborg University Hospital were randomly selected for the study between 15th of July and the 3rd of December 2015. Three methods were evaluated using an iPhone 6 for the assessment. In the first method, we simply used the camera on the smartphone to apply moderate pressure on the glabella of each neonate while recording a video or taking a picture. Secondly, we used a Dermatoscope (DermaLite Pro II) attached to the iPhone to obtain each video. Thirdly, we applied a Wratten No. 11 Colour Filter between the DermaLite Pro II and the camera on the iPhone 6. A transcutaneous bilirubinometer (TcB) or a total serum bilirubin measurement was used as reference. The red, green and blue colour intensities of each video were analysed and correlated to the respective bilirubin reference values.
Results
No significant correlation between the iPhone 6 analysis and reference bilirubin levels using the first and second method was observed (Pearson´s correlation coefficient; red channel 0.05 and -0.38; green channel -0.03 and -0.05; blue channel -0.11 and 0.01). However, method three revealed a significant negative Pearson´s correlation coefficient of 0.61 (p = 0.015) between the blue channel intensities and the TcB with a coefficient of variance of 22.5 %.
Conclusion
The presented methods demonstrate differential results. Method three did, however, indicate a proof of concept and therefore we believe that a combination of standardised light, even pressure and a filter excluding the reflectance of haemoglobin along with further improvements and studies, can be useful in the screening process of neonatal hyperbilirubinemia.
| Language | English |
|---|---|
| Publication date | 21 Dec 2015 |
| Number of pages | 23 |