Artificial Intelligence can help to uncover heart defects in babies

Researchers from the Swiss Federal Institute of Technology in Zurich (ETH) and the KUNO Klinik St. Hedwig, a hospital based in the Bavarian city of Regensburg, have developed an algorithm on the basis of Artificial Intelligence (AI) that is capable of detecting pulmonary hypertension, a heart defect that is difficult to diagnose, in newborn babies, further details of which can be found in a report published by ETH. This serious disorder causes the arteries to the lungs to remain narrowed or to close up again in the first few days or weeks after birth.

This restricts the flow of blood to the lungs and reduces oxygen saturation in the bloodstream. A diagnosis requires a highly specific ultrasound examination. If a patient is suffering from this heart defect, doctors need to act quickly. However, the experience and expertise required for this is often unavailable outside of the large perinatal centers.

The team led by data scientist and ETH professor Julia Vogt has, in conjunction with neonatology specialists from the Regensburg-based hospital, developed a computer model that can offer reliable support for the diagnosis of this disease in newborn babies. In their study, the researchers initially trained their algorithm using several hundred video clips of cardiac ultrasound examinations performed on 192 newborns. The dataset also included the respective diagnoses.

The study then sought to verify how well the algorithm performed in interpreting the images by adding a second dataset of ultrasound images from 78 newborn infants. The result: In 80 to 90 percent of cases, the model suggested the correct diagnosis and determined in 65 to 85 percent of cases the correct severity level of the disease. The AI therefore only offers support, Vogt explains, adding that: “The crucial issue for us is that the final decision should always be made by a human, by a doctor”. ce/mm