Automated craniofacial biometry with 3D T2w fetal MRI

Area of Science:

• Medical Imaging
• Fetal Medicine
• Genetics

Background:

• Prenatal evaluation of craniofacial phenotypes is crucial but challenging with 3D ultrasound due to subjectivity.
• Automated methods are needed to improve accuracy and efficiency in fetal craniofacial measurements.

Purpose of the Study:

• To develop and validate an automated pipeline for fetal craniofacial biometrics using 3D slice-to-volume reconstructed (SVR) fetal MRI.
• To identify significant craniofacial differences in fetuses with Down syndrome (T21) using this automated method.

Main Methods:

• An atlas-based automated label propagation pipeline was created using 3D SVR fetal MRI.
• 31 craniofacial biometrics were defined, and an MRI atlas with landmarks was used for registration and auto-labelling.
• The pipeline was tested on 108 healthy controls and 24 fetuses with T21 (29-36 weeks GA).

Main Results:

• The automated pipeline achieved a 0.3% placement error rate across 132 subjects.
• Seven measurements, including anterior base of skull length and maxillary length, showed significant differences in the T21 cohort (p<0.001).
• Automated measurements took ~5 minutes, compared to 25-35 minutes for manual methods; growth charts were generated.

Conclusions:

• This is the first automated atlas-based protocol for fetal craniofacial biometrics using 3D SVR MRI.
• The method accurately identified morphological differences in fetuses with T21, aiding prenatal diagnosis.
• Future work will focus on improving reliability, larger cohorts, and technical advancements for enhanced prenatal care.