New imaging software can accurately diagnose jaundice in the blink of an eye

Imagine a system that can detect in a second if a newborn baby has jaundice, which afflicts 60 percent of babies and in severe cases leads to brain damage and hearing loss.

Visualize that the system starts the treatment program immediately and notifies the caregiver via text message.

Thanks to Iraqi and Adelaide researchers, all three scenarios are now possible.

Engineers at the University of South Australia and Middle Technical University have designed imaging software that can accurately diagnose jaundice in the blink of an eye, automatically turn on a blue LED to combat it, and send the diagnosis as a text message to a caregiver.

Jaundice is a common disease in newborns, especially premature babies, in which there is an overload of an orange-yellow pigment in the bloodstream called bilirubin. It usually disappears quickly when the baby’s liver is mature enough to remove it from the body.

However, in severe cases of jaundice due to sickle cell anemia, blood diseases, and deficiency of certain enzymes, light therapy using fluorescent blue light to break down bilirubin in the baby’s skin is usually used to treat the condition.

Professor Javaan Chahl, UniSA’s remote sensing engineer, says jaundice is especially common in developing countries, which often lack the equipment or trained medical staff to treat it effectively.

With image processing techniques extracted from the data stored by the camera, we can inexpensively and accurately screen newborn jaundice in a non-invasive manner before taking a blood test.

When bilirubin levels reach a certain threshold, the microcontroller triggers a blue LED light therapy and sends the data to a cell phone.

This can be done in one second, literally, which can have an impact in severe cases where brain damage and hearing loss can follow if treatment is not given quickly. “

Professor Javaan Chah, UniSA Remote Sensing Engineer

The researchers tested the system at an intensive care unit in Mosul, Iraq, with 20 newborns diagnosed with jaundice. In the second data set, 16 images of newborns were taken, five of which were healthy and the rest yellow. The system was also successfully tested on four other dolls with white and brown skin color, with or without jaundice pigmentation.

“Previous studies using a non-invasive way to detect jaundice with sensors have failed. The methods tested have been unreliable, expensive, ineffective, and in some cases caused infections and allergies where the sensors required skin contact,” says Professor Chahl.

“Our system overcomes these barriers by detecting jaundice immediately in a new digital color display that enables high diagnostic accuracy at a relatively low cost. It could be widely used in hospitals worldwide and in medical centers where laboratory facilities and trained medical personnel are not available.”

The study has been published in a technical journal Plans.


University of South Australia

Journal reference:

Hashim, W., et al. (2021) Detection of neonatal jaundice by computer vision system. Plans.


Leave a Comment