Cameras in place of ECG sensors for premature babies

April 18, 2017 // By Julien Happich
A team of researchers from EPFL and CSEM have devised a contactless and wireless camera system able to continuously monitor premature babies’ vital signs such as heart rate and breathing patterns, based on image processing.

According to Jean-Claude Fauchère, attending neonatologist at University Hospital Zurich (USZ)'s division of neonatology, today's skin sensors placed on the babies’ chest are so sensitive that they generate false alarms nearly 90% of the time, mainly caused by the babies moving around.

“This is a source of discomfort for the babies, because we have to check on them every time. It’s also a significant stress factor for nurses and a poor use of their time – it distracts them from managing real emergencies and can affect quality of care", explains Fauchère.

About to be tested at the USZ, the camera system requires no physical contact. The babies’ pulse is detected through their skin colour, which changes ever so slightly every time their heart beats. And their breathing is monitored through movements in their thorax and shoulders. At night, infrared cameras take over.

The optical system was designed by the CSEM researchers, who chose cameras that are sensitive enough to detect minute changes in skin color. They teamed up with the EPFL researchers to design algorithms for processing the data in real time. CSEM focused on respiration, while EPFL worked on the heart rate.

“We ran an initial study on a group of adults, where we looked at a defined patch of skin on their forehead," said Sibylle Fallet, a PhD student at EPFL. “With our algorithms we can track this area when the person moves, isolate the skin pixels and use minor changes in the colour of these pixels to determine the pulse. The tests showed that the cameras produced practically the same results as conventional sensors.”

"We plan to take measurements on as many preemies as possible to see whether, under real-life conditions, the results we get from our algorithms match the data collected by the on-skin sensors”, said Virginie Moser, the CSEM researcher in charge of the set-up at USZ. If successful, the camera system could one day