Accuracy of magnetic resonance imaging for measuring fetal sheep lungs and other organs




Lung volume measurement by fetal magnetic resonance imaging (fetMRI) has been used to predict survival of fetuses with isolated congenital diaphragmatic hernia (CDH). So far, the accuracy and precision of fetMRI for volumetry of either the normal or hypoplastic developing lung has not been formally studied.


A total of nine sheep carrying 14 fetuses underwent fetMRI under general anesthesia at a mean of 118 days' gestational age (term = 145 days). A total of 61 organs were measured in nine normal fetal sheep and five that underwent surgical creation of diaphragmatic hernia (DH), so as to induce pulmonary hypoplasia. Lungs were measured on T2-WI (weighted images) in three different planes, while liver and kidneys were measured in the axial (T1-WI) and sagittal (T2-WI) planes, respectively. Necropsy was done within 24 h after fetMRI to determine the volume postmortem by the water displacement method. Values were linearly correlated and a Bland and Altman analysis was done for volume measurement comparison, calculating means ± SD, bias (mean of the difference of volume measurements), precision (SD of the difference) and absolute and proportionate limits of agreement for both methods. The accuracy of fetMRI volume measurement was determined for different organ groups by calculating the median relative error and precision index, both being measures of error in proportion to the magnitude of the volume measured, as a clinically relevant proxy of potential errors.


The fetMRI volume measurements were on average larger than postmortem volumes, except for the kidneys. Kidney volume determination had a relative error of 29%, while measurements of larger organs had larger relative errors (42% for liver). Normal lungs were less accurately measured in the coronal or sagittal than in the axial plane (relative error 53%, 73% and 38%, respectively; P < 0.05 for sagittal vs. axial). Axially-measured lung volumes were more accurate for lungs of normal sheep compared to DH lungs (relative error 38% vs. 73%, respectively; P < 0.05).


FetMRI measured systematically higher volumes for organs such as fetal liver or lung. This may be related to fluid loss or lack of perfusion at the time of necropsy. Measurement of lung volume by fetMRI was most accurate in the axial plane. Measurements of lung and liver volumes by fetMRI in normal sheep were both in agreement with volumes measured at necropsy. Loss of accuracy for DH-lungs in comparison with the accuracy when measuring other similarly small organs, such as kidneys, suggests that fetMRI measurements can be less accurate for hypoplastic lungs related to CDH. With improving hardware, it might become easier to render the fetal lung and determine its volume reliably. Copyright © 2005 ISUOG. Published by John Wiley & Sons, Ltd.