187 - Enhanced Fetal MRI Safety Assessment Through Incorporation of Realistic Fetal Motion

Publication Number: 3180.187

Filiz Yetisir, Harvard Medical School, Boston, MA, United States; Neel Dey, Massachusetts General Hospital, Cambridge, MA, United States; Ilias I. Giannakopoulos, New York University Grossman School of Medicine, New York, NY, United States; José E.. Cruz Serrallés, Center for Biomedical Imaging, Department of Radiology, New York University Grossman School of Medicine, New York, NY, USA, New York, NY, United States; Georgy Guryev, Massachusetts Institute of Technology, Cambridge, MA, United States; Esra Abaci Turk, Boston Children's Hospital, Harvard Medical School, Boston, MA, United States; Adi Titelman Ashkenazy, Boston Children's Hospital, Watertown, MA, United States; Riccardo Lattanzi, New York University Grossman School of Medicine, New York, NY, United States; Jacob White, Massachusetts Institute of Technlogy, Cambridge, MA, United States; Polina Golland, Massachusetts Institute of Technology, Cambridge, MA, United States; Elfar Adalsteinsson, Massachusetts Institute of Technology, Cambridge, MA, United States; P. Ellen grant, Boston Children's Hospital, Boston, MA, United States

Background: Tissue heating due to radiofrequency (RF) energy must be limited during Magnetic Resonance Imaging (MRI). Fetal MRI in the standard mode of imaging has a long history of safe use. However, before implementing new technologies which can improve image quality and imaging speed, their RF safety need to be assessed. Both average (core body) and peak local tissue heating must be limited during MRI. Local tissue heating is challenging to measure; therefore, it is estimated using anatomical numerical body models and electromagnetic simulations. Simulation studies so far only used static fetus models, failing to accurately assess RF safety of fetal MRI. In this study, we assess the variability of fetal tissue heating estimations due to realistic fetal motion using 100s of fetal positions for one pregnant subject.

Objective: To assess the variability of fetal MRI tissue heating estimations due to realistic fetal motion.

Design/Methods: Data acquisition: Structural data of the maternal body (3 breath-hold T1-weighted DIXON volumes, each 9.8s) and ~15 mins time series 3D echo planar imaging (EPI) data of the uterine volume (fetal motion sampling interval: 2.1s) were collected in one pregnant subject (gestational age: 24 weeks). Body model generation: DIXON volumes were segmented into 21 tissues. Each EPI volume capturing a fetal position was segmented into amniotic fluid, placenta, umbilical cord and the fetus's body, brain, and eyes using segmentation networks trained on 176 EPI volumes. Using ANTs, each uterine volume was registered into the maternal volume, generating one pregnant body model per fetal position. SAR simulations: Electromagnetic fields in the resulting pregnant body models were rapidly (~2.5 mins/body model) simulated using MARIE, for 3 Tesla MRI. Fetal and maternal tissue heating quantified by average and peak local specific absorption rate (SAR) were computed in each body model for 2 W/kg maternal whole-body SAR (regulatory limit).

Results: Fig.2 shows the anatomical maternal body model, and the dynamic fetus model (270 fetal positions). Among the 270 positions, fetal peak SAR and aveSAR varied by up to 17% and 23%. The location of the fetal peak SAR was at different parts of the fetus including the back and the left extremities (Fig.3).

Conclusion(s): For the one subject and 270 fetal positions evaluated, the value and location of fetal tissue heating varied considerably indicating that fetal motion should be taken into account when evaluating the safety of new technologies. Future work will focus on improving the speed and accuracy of the segmentations and including more pregnant body models.

Figure 1
Pipeline to generate several body models with different fetal positions that sample fetal motion from one pregnant subject.

Figure 2
Left: maternal body model loaded into the RF coil model, right: dynamic fetus model. Fetal motion was observed and sampled during ~15 mins of EPI acquisition, however the effective fetal motion duration shown is ~10 mins after the elimination of fetal positions with severe motion artifacts or poor segmentation quality. Time stamp shows the time for 270 fetal positions with a sampling interval of 2.1 s, not the actual time of each fetal position.

Figure 3
Normalized local SAR maps through peak maternal SAR (left) and peak fetal SAR for example fetal positions. Black plus signs show the peak maternal or fetal SAR location.