AI spots heart and lung conditions in eye images of premature infants
An artificial intelligence (AI) algorithm can help identify serious cardiovascular and pulmonary complications in premature infants by analyzing retinal images captured during routine screens for retinopathy of prematurity (ROP), according to new data published in JAMA Ophthalmology.[1]
ROP is a potentially blinding eye disorder found in premature or low birthweight neonates and requires screening in the first weeks of life. But researchers found the routine eye images can also show early signs of both bronchopulmonary dysplasia (BPD) and pulmonary hypertension (PH). Both of these conditions are difficult to diagnose early and requires specialized imaging of the back of the eye, or the use of blood tests, imaging and sometimes more invasive tests. The researchers said earlier detection could make a meaningful difference in outcomes with earlier interventions and better treatment planning.
"AI allows us to detect subtle patterns in retinal images that are not visible to the human eye," explained lead author Praveer Singh, PhD, assistant professor of ophthalmology at the University of Colorado Anschutz in a statement. "This opens the possibility of using a simple photograph to gain insights into a premature infant's overall health."
The study included retinal images from 493 infants before clinical diagnoses of BPD or PH were made. These were collected at seven NICUs participating in the Retinopathy of Prematurity (i-ROP) study, a long-running, multi-institutional research project supported by the National Institutes of Health. Hospitals included University of Colorado, Massachusetts General Hospital, Oregon Health & Science University, Northeastern University in Boston, and the University of Illinois Chicago.
Using a deep learning algorithm, researchers tried to detect disease-related patterns using the images alone, patient data alone and then when the two were options combined. The combined data was the most accurate, with a 82% accuracy for BPD and 91% for PH.
Researchers then removed images that showed any clear, clinical signs of ROP and the results remained consistent. The authors said this suggests the AI identified information beyond what can be seen in traditional eye disease markers.
"One of the challenges with realizing the potential of many oculomics algorithms is that imaging the back of the eye is not yet part of the normal care pathway for many populations of patients," explained co-author Peter Campbell, MD, MPH, professor of ophthalmology at Oregon Health and Science University, in the same statement. "For more and more NICUs imaging is part of the care pathway for ROP, which means the barriers to implement technologies like this are significantly lower."
Persistent PH is usually caused when a newborn's circulatory system does not adapt to breathing air outside of the uterus. The pulmonary arteries typically fully open after a neonate takes its first breath, but if they do not, it restricts the amount of oxygen they receive. The conditional also can be caused by congenital heart and lung abnormalities.
BPD is caused by incomplete lung growth or lung damage in premature babies. Damage can be cause by infection or mechanical ventilation.
Dave Fornell has covered healthcare for more than 17 years, with a focus in cardiology and radiology. Fornell is a 5-time winner of a Jesse H. Neal Award, the most prestigious editorial honors in the field of specialized journalism. The wins included best technical content, best use of social media and best COVID-19 coverage. Fornell was also a three-time Neal finalist for best range of work by a single author. He produces more than 100 editorial videos each year, most of them interviews with key opinion leaders in medicine. He also writes technical articles, covers key trends, conducts video hospital site visits, and is very involved with social media. E-mail: [email protected]