Feasibility of Fetal Fat Volume Assessment using 3D Water-Fat MRI
Stephanie Giza1, Craig Olmstead2, Kevin Sinclair1, Charles A McKenzie1,3, and Barbra de Vrijer3,4

1Medical Biophysics, Western University, London, ON, Canada, 2Schulich School of Medicine and Dentistry, Western University, London, ON, Canada, 3Division of Maternal, Fetal and Newborn Health, Children's Health Research Institute, London, ON, Canada, 4Obstetrics and Gynecology, Western University, London, ON, Canada

Synopsis

Conventional ultrasound techniques perform adequately in assessment of fetal size in obese mothers but fail to identify the fetus that is chronically stressed inside the womb. A reliable MRI measurement of fetal fat volume and distribution would therefore be a powerful tool in the assessment of fetal condition. Using a 3D LAVA-Flex sequence during maternal breath hold, fat signal fraction images were generated to measure the fat volume of the fetus, while correcting for partial volume effects. 3D water-fat MRI was found to provide a reliable measurement of fetal fat volumes that could be used to assess size and growth.

Target Audience

Researchers interested in the use of Magnetic Resonance Imaging to detect fetal abnormalities.

Purpose

The proportion of pregnant women with body mass index (BMI) above normal (BMI>25 kg/m2) is increasing1, resulting in an increased risk of negative fetal outcomes including macrosomia and fetal growth restriction2. Detecting changes in fetal fat distribution and overall volume before birth may help physicians evaluate the risk of macrosomia, growth restriction, stillbirth, and future metabolic health and determine the best time for early intervention, such as induction of preterm delivery. Magnetic Resonance Imaging (MRI) complements routine ultrasound methodologies, by providing a volumetric measurement of fetal adipose tissue distribution. Previous studies have made measurements of fetal fat using 2D fat-only MRI3; however, 3D water-fat imaging allows for accurate fat volume measurement with correction of partial volume effects. This study aims to develop a reliable 3D MRI method to assess fetal size and fat distribution in utero.

Methods

Women with singleton pregnancies in their second or third trimester were recruited from low risk and specialized high BMI obstetrical clinics and underwent a fetal MRI in a wide-bore (70cm diameter) 1.5T MRI (GE MR 450w). 20 women consented to the study, however 1 participant was unable to commence scanning due to claustrophobia. The 19 imaged participants had a BMI between 19-53 kg/m2 (7 normal BMI, 10 high BMI) and gestational age between 29 and 34 weeks. The participants also underwent routine obstetric ultrasounds through their clinics, and the biometry measurements from the ultrasounds immediately preceding and following the MRI examination were used for measurement comparisons.

Fat-only and water-only images oriented axial to the fetus were acquired using a 3D LAVA-Flex sequence during a maternal breath hold (TR 6.2 ms, flip angle 5°, FOV 48 cm, acquisition matrix 160×160, slice thickness 4 mm, 32-48 slices, 2× parallel MRI acceleration, acquisition time 19-24 s). Fat signal fraction (FSF=fat/(water+fat)) images were generated and reformatted to be axial to the fetal abdomen. From these images, the subcutaneous fat was manually segmented by two independent readers (SG, CO) from where the arm meets the abdomen to where the leg meets the abdomen. (See Figure1 for an example segmentation.) Segmented subcutaneous fat volumes were multiplied by the mean fat signal fraction of this region to correct for partial volume errors in the segmented fat voxels. Intraclass correlation coefficients (ICC) were used to assess interrater reliability of the subcutaneous fat volumes segmented by the two readers. Estimated fetal weight percentiles were calculated using the Shepard formula4 from ultrasound and MRI measurements and a growth percentile calculator5, then ICC was used to compare agreement of the MRI and US measurements.

Results and Discussion

Comparison of subcutaneous fat volumes measured by multiple readers gave an ICC of 0.950 (P<0.001), indicating excellent interrater agreement. The ICC between ultrasound- and MRI-determined fetal growth percentile was 0.978 (P<0.001), showing that the MRI assessment of fetal size agrees with standard US measurements. This is an encouraging result because the calculation of fetal growth percentile is sensitive to small changes in the input measurements. A moderate correlation was found between the fat fraction corrected subcutaneous fat volumes and the growth percentile at MRI (Figure 2). This is an unsurprising result as fat volume is only one component of fetal size. For example, Figure 3 shows two fetuses with similar gestational age and estimated growth percentile that have very different subcutaneous fat volumes. This demonstrates the value of including a fat signal fraction based partial volume correction as there is a striking visual difference in the fat signal fraction of the two segmented volumes. This difference accounts for part of the large difference in the calculated subcutaneous fat volumes.

All images were of sufficient quality to be segmented, so fetal motion during 3D acquisition was not a significant problem. The LAVA-Flex sequence is limited for calculation of fat fraction since it can not account for biases such as R2* correction and accurate fat spectrum modelling. These biases could be eliminated by using quantitative fat fraction measurements that account for these sources of bias (e.g. [6]). Techniques that correct fat fraction biases have the potential to improve our measurements of fetal adipose tissue.

Conclusion

3D water-fat MRI with correction of partial volume effects reliably assesses fetal fat volumes.

Acknowledgements

Grant support from Children’s Health Research Institute and Western University.

References

1. Kim SY, Dietz PM, England L, et al. Trends in Pre-pregnancy Obesity in Nine States, 1993-2003. Obesity. 2007; 15:986-993.

2. Ornoy A. Prenatal origin of obesity and their complications: Gestational diabetes, maternal overweight and the paradoxical effects of fetal growth restriction and macrosomia. Reprod Toxical. 2011;32:205-212.

3. Anblagan D, Deshpande R, Jones NW, et al. Measurement of fetal fat in utero in normal and diabetic pregnancies using magnetic resonance imaging. Ultrasound Obst Gyn. 2013; 42: 335-340.

4. Shepard MJ, Richards VA, Berkowitz RL, et al. An evaluation of two equations for predicting fetal weight by ultrasound. Am J Obstet Gynecol. 1982; 42: 47-54.

5. Fetus Measurement Growth Calculators (2015). Retrieved from http://www.baby2see.com/medical/charts.html.

6. Hines CD, Frydrychowicz A, Hamilton G, et al. T(1) independent, T(2) (*) corrected chemical shift based fat-water separation with multi-peak fat spectral modeling is an accurate and precise measure of hepatic steatosis. J Magn Reson Imaging 2011;33:873–881.

Figures

Figure 1: Manual segmentation of fat of fetus with A) small subcutaneous volume (130mL) and B) large subcutaneous volume (210mL). These images were taken at 33 and 30 weeks of gestational age respectively.

Figure 2: Subcutaneous volume vs. growth percentile at MRI. The fat signal fraction corrected subcutaneous volume is moderately correlated with the growth percentile calculated from MRI measurements. R2 = 0.36. The line in the graph shows the linear regression.

Figure 3: Example of two fetuses with similar growth percentile but different fat volumes. Fetus A was gestational age 32 weeks, growth percentile 57%, and fat volume 20mL. Fetus B was gestational age 34 weeks, growth percentile 61%, and fat volume 135mL.



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
3891