Abstract
Objective(s)
Recent advances in fetal three-dimensional (3D) ultrasound (US) enable automated intracranial volume (ICV) measurements. If US yields results comparable to magnetic resonance imaging (MRI), it could serve as a fast, cost-effective, and scalable alternative. This study directly compares and visualizes ICV measurements from fetal 3D US and 3D MRI.
Study Design
We applied automated ICV segmentation to US and MRI data from healthy participants. In a subset of fetuses who underwent both US and MRI within 24 h, we evaluated ICV agreement using the intraclass correlation coefficient (ICC) and the Dice similarity coefficient (DSC). As a proof of principle, we also generated a multimodal ICV trajectory incorporating fetal US, fetal MRI, and neonatal MRI.
Results
Among the 71 participants, 47 had high-quality US scans at 20 and 30 weeks of gestation, 61 had high-quality fetal MRI, and 47 had high-quality neonatal MRI. Thirteen fetuses around 30 weeks had US and MRI scans acquired within 24 h, showing good agreement with an ICC of 0.797 and DSCUS−MRI scores of 0.900–0.925. Mean ICV values were 80 ± 12 ml around 22 weeks (US), 248 ± 28 ml around 30 weeks (US), 294 ± 36 ml around 32 weeks (fetal MRI) and 508 ± 51 ml around 43 weeks (neonatal MRI).
Conclusion(s)
Fetal US and MRI yield comparable ICV measurements around 30 weeks of gestation, validating automatic 3D US volumetric methods. The use of 3D US may support development of normative curves and improve comparisons with at-risk fetuses, including in regions where fetal MRI is less accessible.
Recent advances in fetal three-dimensional (3D) ultrasound (US) enable automated intracranial volume (ICV) measurements. If US yields results comparable to magnetic resonance imaging (MRI), it could serve as a fast, cost-effective, and scalable alternative. This study directly compares and visualizes ICV measurements from fetal 3D US and 3D MRI.
Study Design
We applied automated ICV segmentation to US and MRI data from healthy participants. In a subset of fetuses who underwent both US and MRI within 24 h, we evaluated ICV agreement using the intraclass correlation coefficient (ICC) and the Dice similarity coefficient (DSC). As a proof of principle, we also generated a multimodal ICV trajectory incorporating fetal US, fetal MRI, and neonatal MRI.
Results
Among the 71 participants, 47 had high-quality US scans at 20 and 30 weeks of gestation, 61 had high-quality fetal MRI, and 47 had high-quality neonatal MRI. Thirteen fetuses around 30 weeks had US and MRI scans acquired within 24 h, showing good agreement with an ICC of 0.797 and DSCUS−MRI scores of 0.900–0.925. Mean ICV values were 80 ± 12 ml around 22 weeks (US), 248 ± 28 ml around 30 weeks (US), 294 ± 36 ml around 32 weeks (fetal MRI) and 508 ± 51 ml around 43 weeks (neonatal MRI).
Conclusion(s)
Fetal US and MRI yield comparable ICV measurements around 30 weeks of gestation, validating automatic 3D US volumetric methods. The use of 3D US may support development of normative curves and improve comparisons with at-risk fetuses, including in regions where fetal MRI is less accessible.
| Original language | English |
|---|---|
| Article number | 121633 |
| Number of pages | 8 |
| Journal | NeuroImage |
| Volume | 324 |
| DOIs | |
| Publication status | Published - 15 Dec 2025 |
Bibliographical note
Publisher Copyright:Copyright © 2025. Published by Elsevier Inc.
Keywords
- Intracranial volume
- Magnetic resonance imaging
- Third trimester
- Ultrasound