Magnetic resonance imaging compared to ultrasonography in fetal weight and volume estimation in diabetic and normal pregnancy

Abstract

To estimate fetal volume and weight in diabetic and normal pregnancy using high-resolution magnetic resonance imaging. T1-weighted magnetic resonance imaging was combined with semiautomatic segmentation technique. The accuracy of fetal volume estimations thus obtained was compared with conventional ultrasound-based weight estimations in ten pregnant women with insulin-dependent diabetes mellitus and ten women with normal pregnancy. Examinations were made within 48 hours before delivery. Ultrasound-based estimations of fetal weight showed a correlation rate of r=0.77 with the actual birth weights in the whole material, while volume determinations based on magnetic resonance imaging showed a significantly better correlation rate of r=0.95. Diabetic women did not differ from the normal pregnancy group with regard to birth weight or the accuracy of weight estimations. High-resolution magnetic resonance imaging combined with semiautomatic segmentation software was found to be accurate in determining fetal volume and, consequently, better than conventional ultrasound-based techniques in estimating fetal weight. The use of magnetic resonance imaging in fetal weight estimation may be recommended for clinical situations where an accurate weight estimate is considered essential.