New computer vision technology to safely monitor heart and respiratory rates of premature babies has the potential to replace ECG monitoring in neonatal units.
The University of South Australia (UniSA) School of Nursing and Midwifery and Engineering have trialled a new non-contact way to monitor pre-term infants in intensive care.
The study involved 10 premature babies at the Flinders Medical Centre Neonatal Unit.
The infants were filmed using two high-resolution Nikon cameras at close range. Babies’ physiological data was extracted using advanced signal processing techniques to detect subtle colour changes and movement not visible to the human eye.
“Our computer vision system captures subtle signals in a pre-term baby, such as invisible skin colour variations that can be amplified to measure cardiac activity.
“We used colour change in skin and motion which is quite unique. Other research being done uses infra-red and other methods to get data on heart rate or respiratory rate,” neonatal critical care specialist nurse and UniSA lecturer in the School of Nursing and Midwifery Kim Gibson said.
Pre-term babies were selected as they are prone to episodes of bradycardia and apnoea which are difficult to monitor without the use of an ECG which relies on expensive, adhesive electrodes that can damage infants’ fragile skin and leave them vulnerable to infection.
“Premature babies of 25-26 weeks have wafer skin. Adhesive electrodes take off the outer layer of the skin which allows having an entry point for bacteria and increases the risk of infection,” Ms Gibson said.
An unexpected finding was that the system was able to accurately detect apnoea when the ECG monitor did not.
During filming, one baby stopped breathing for about eight seconds which was not picked up by the ECG and recorded 14-17 breaths per minute. It was well documented that the ECG was not 100% perfect, Ms Gibson said.
“We can also apply algorithms to magnify movement to give nursing staff a clear picture of what is going on with pre-term infants.”
“When you magnify something a lot with an iPhone, such as zooming in on a crowd, the image appears grainy or distorted. We can use an algorithm to enhance the image to improve its accuracy.”
Preliminary results showed that the non-contact system could help monitor the health of pre-term babies, particularly when resources were scarce, and the risk of infection was high, Ms Gibson said.
“It’s about improving care. We know babies experience pain and we are learning more that when experiencing lots of painful stimuli that has an impact on brain development.
“Why not do something to decrease the pain experience and reduce the infection risk for babies, especially those born prematurely which are immunocompromised; and minimise those practices and procedures that can essentially harm them.”
The research was published in journal Pediatric Research.