Status: Funded - Open
Aviad Rabinowich, MD
BACKGROUND: Fetal growth restriction (FGR) complicates 5-10% of pregnancies and increases the risk of perinatal morbidity and mortality by ~44%. GAP: Diagnosis and risk stratification are conducted with ultrasound; however, unfortunately, perinatal prognostication is limited, resulting in misdiagnosis, excessive monitoring, and unnecessary interventions. HYPOTHESIS: Magnetic resonance imaging (MRI) anthropometric parameters have the potential to be sensitive to the severity of FGR. Our hypothesis is that by employing quantitative multi-parametric data in conjunction with machine learning techniques, predicting perinatal outcomes for FGR can be enhanced, allowing for the anticipation of adverse events. METHODS: This prospective single-center study will involve MRI scans of 380 participants, comprising 300 fetuses with appropriate for gestational age development and 80 fetuses with FGR. Deep learning networks will be employed to quantify disease-specific anthropometric parameters, such as body volume, fat deposition, and placental morphometrics. The correlation between these disease-specific anthropometric features and perinatal outcomes will be investigated using machine learning techniques. RESULTS: Pending. IMPACT: The proposed study aims to provide novel biomarkers to grade FGR severity. Ultimately, integrating fetal MRI with ultrasound and clinical data, using machine learning, will aid in identifying fetuses at risk for adverse perinatal outcomes, target pregnancy and newborn monitoring, and potentially improve patient care and outcome.