THE ROLE OF ARTIFICIAL INTELLIGENCE IN THE AUTOMATIC DETECTION OF FETAL MALFORMATIONS BY ULTRASOUND

Abstract

Artificial intelligence (AI) has emerged as a valuable tool to enhance prenatal ultrasonography, particularly in the detection of fetal malformations. Recent studies show that deep learning models achieve higher sensitivity, specificity, and accuracy compared with conventional approaches, with notable results in screening for congenital heart defects such as ventricular septal defect, tetralogy of Fallot, and aortic coarctation, as well as in identifying central nervous system anomalies. Integrated platforms, such as the HeartAssist system, have already reported accuracy rates above 98% in cardiac structure analysis. Despite its potential, challenges remain, including heterogeneous datasets, the need for multicenter validation, and ethical and privacy concerns. In conclusion, AI represents a promising resource to standardize and expand prenatal diagnosis, providing valuable support to specialists and contributing to improved perinatal outcomes.