MRI-Based Fetal Weight Estimation for Predicting Neonatal Morbidity and Severe Smallness in Small for Gestational Age Pregnancies
2Gray Faculty of Medical and Health Sciences, Tel Aviv University, Israel
3School of Neurobiology, Biochemistry and Biophysics, The George S. Wise Faculty of Life Sciences, Tel Aviv University, Israel
4Obstetrics and Gynecology, Lis Maternity and Women’s Hospital, Tel Aviv Sourasky University Medical Center (Ichilov), Israel
5Medical Imaging Sciences, University of Haifa, Israel
6Biomedical Engineering, Technion – IIT, Israel
7May-Blum-Dahl Technion Human MRI Research Center, Technion – IIT, Israel
8Neonatal Intensive Care Unit, Tel Aviv Sourasky University Medical Center (Ichilov), Israel
9Division of Ultrasound in Obstetrics and Gynecology, Tel Aviv Sourasky University Medical Center (Ichilov), Israel
10Department of Maternal-Fetal Medicine, Tel Aviv Sourasky University Medical Center (Ichilov), Israel
11School of Computer Science and Engineering, The Hebrew University of Jerusalem, Israel
12Edmond and Lily Safra Center for Brain Sciences, The Hebrew University of Jerusalem, Israel
13Sagol Brain Institute, Tel Aviv Sourasky University Medical Center (Ichilov), Israel
14Sagol School of Neuroscience, Tel Aviv University, Israel
Background: To compare MRI-based (mrEFW) with sonographic (sEFW) estimated fetal weight for predicting neonatal morbidity and severe smallness (birth weight <3rd centile) in small for gestational age (SGA) pregnancies.
Methods: This prospective single-center study included singleton pregnancies with sEFW below the 10th centile. Fetuses underwent 3-T MRI using TruFISP sequences. Fetal body volume was segmented with an automated deep-learning tool with manual correction, converted to mrEFW, and then to centiles using a previously published formula and growth charts; sEFW centiles were derived using the Hadlock 4 formula and Hadlock charts. Centiles were analyzed both continuously and after dichotomization at the 3rd centile. Outcomes were a composite adverse neonatal outcome (CANO) and birth weight <3rd centile. Logistic regression models incorporating mrEFW, sEFW, or both were compared using log likelihood ratio tests.
Results: Sixty-four SGA pregnancies were included (mean gestational age at MRI, 33.7 ± 1.8 weeks). CANO occurred in 39 neonates (61%), and 25 (39%) had a birth weight <3rd centile. Using continuous centiles, the area under the curve (AUC) for predicting CANO was 0.82 (95% confidence interval [CI], 0.71–0.92) for mrEFW versus 0.72 (95% CI, 0.59–0.85) for sEFW; for birth weight <3rd centile, 0.89 (95% CI, 0.81–0.96) versus 0.80 (95% CI, 0.69–0.90). mrEFW identified birth weight <3rd centile with a sensitivity of 0.96 when analyzed either continuously (specificity, 0.81) or dichotomously (specificity, 0.62). Adding mrEFW to sEFW significantly improved prediction of CANO (P = 0.01) and of birth weight <3rd centile (P < 0.001), whereas adding sEFW to mrEFW improved only the latter (P = 0.03; CANO, P = 0.11).
Conclusions: In pregnancies with sonographically suspected SGA, mrEFW outperformed sEFW in predicting neonatal morbidity and severe smallness at birth and may serve as a selective adjunct for antenatal risk stratification.