Machine Learning in Maternity: Early Diagnosis of Placenta Accreta Spectrum

Placenta accreta spectrum (PAS) is a life-threatening obstetric condition marked by abnormal placental attachment and invasion into the uterine wall. Early and accurate prenatal diagnosis is crucial to improving maternal and fetal outcomes as it reduces morbidity, mortality, and complications such as severe hemorrhage and cesarean hysterectomy. Current diagnostic approaches utilize magnetic resonance and ultrasound imaging and rely heavily on clinician interpretation. This study aims to evaluate recent advancements in machine learning (ML) techniques for the early diagnosis of PAS using ultrasound and magnetic resonance imaging (MRI). A comprehensive review of the literature evaluated 14 studies highlighting various ML techniques, including linear, ensemble, deep learning, and hybrid models, highlighting the study outcomes. ML techniques demonstrated improved diagnostic performance compared to traditional methods. Ultrasound-based models achieved accuracy rates ranging from 84.6% to 92.3%, particularly with ensemble methods and deep dictionary learning. MRI-based approaches showed even higher performance, particularly with texture analysis using k-nearest neighbors (k-NN), achieving up to 98.1% accuracy. Despite these promising results, notable challenges included model generalizability across diverse populations and variability in imaging quality due to differences in medical equipment and patient demographics. Overall, ML offers significant potential as a tool for the early diagnosis of PAS by improving diagnostic accuracy, consistency, and reducing human error. However, further research is needed to address limitations related to generalizability and standardization before widespread clinical implementation.