Abstract:Objectives: Craniofacial phenotype-genotype correlations prenatally remain subjective, and detailed evaluation with 3D ultrasound is challenging and time-consuming. Recent methods for automating MRI fetal brain biometry have shown technical feasibility. Using 3D motion-corrected, slice-to-volume reconstructed (SVR) fetal MRI, we developed an automated landmark propagation pipeline specifically for a large set of cranial and facial measurements. Methods: A literature review and expert consensus identified 31 craniofacial biometrics for fetal MRI. A 3D spatiotemporal SVR fetal MRI atlas was used to define anatomical landmark points and as a template for subject registration, auto-labelling and finally landmark-based biometric calculation. Visual inspections of subject-level landmark labels, as well as multivariate and univariate analyses, were performed in 108 healthy control fetuses and 24 fetuses with Down syndrome, T21, (gestational age, GA, 29-36 weeks), to assess differences between groups and identify meaningful landmarks in T21. Reliability and reproducibility were assessed in 10 random datasets by four observers. Results: Automated labels were produced for all 132 subjects, with a 0.03% major error rate. Significant differences in biometrics were found between T21 and control groups (MANOVA, p<0.001). There were significant differences with large effect sizes for seven measurements (ANOVA, p<0.001), including the anterior base of skull length and maxillary length. Manual measurements took 25-35 minutes per case, while automated extraction took approximately 5 minutes. Bland-Altman plots showed a relative mean bias and limits of agreement within the ranges seen for manual observers except for the mandibular width, which had an agreement threshold for manual observers of good (ICC = 0.904, 95% CI = 0.677-0.975) versus moderate for automated plus manual observers (ICC = 0.635, 95%, CI = 0.176-0.888). Extended GA growth charts (19-39 weeks) based on 280 control fetuses were produced for all craniofacial biometrics and made publicly available for future research. Conclusion: We present the first automated atlas-based protocol using 3D motion-corrected MRI for fetal craniofacial biometrics and, using this method, accurately characterise differences in craniofacial morphology in Downs syndrome fetuses, with findings consistent with the known phenotype. Future work should include studies to improve measurement reliability, larger clinical cohort studies, technical development of craniofacial optimised MRI sequences and 3D deep learning for anatomical landmarking. Future development may reduce clinical reporting time, improve accuracy, increase understanding of prenatal development and enhance personalised prenatal care with accurate genotype-phenotype correlations.

Automatic facial axes standardization of 3D fetal ultrasound images

Reconstruction of the fetus face from three-dimensional ultrasound using a newborn face statistical shape model

EP01.09: Deep learning‐based standardisation of the anatomical fetal facial axes in 3D prenatal ultrasounds

Early Pregnancy Fetal Facial Ultrasound Standard Plane-Assisted Recognition Algorithm

AutoFB: Automating Fetal Biometry Estimation from Standard Ultrasound Planes

Detection of fetal facial anatomy in standard ultrasonographic sections based on real‐time target detection network

Automatic Recognition of Fetal Facial Ultrasound Standard Planes Based on Improved YOLOv4

FetalBrainAwareNet: Bridging GANs with anatomical insight for fetal ultrasound brain plane synthesis

Ultrasound Plane Pose Regression: Assessing Generalized Pose Coordinates in the Fetal Brain

Human-level Performance On Automatic Head Biometrics In Fetal Ultrasound Using Fully Convolutional Neural Networks

Automating the Human Action of First-Trimester Biometry Measurement from Real-World Freehand Ultrasound

Automatic Fetal Face Detection from Ultrasound Volumes Via Learning 3D and 2D Information

Automatic quality assessment for 2D fetal sonographic standard plane based on multi-task learning

An Ultrasound Standard Plane Detection Model of Fetal Head Based on Multi-Task Learning and Hybrid Knowledge Graph

Deep learning fetal ultrasound video model match human observers in biometric measurements

Fetal biometric based abnormality detection during prenatal development using deep learning techniques

Fetal face shape analysis from prenatal 3D ultrasound images

SonoNet: Real-Time Detection and Localisation of Fetal Standard Scan Planes in Freehand Ultrasound

Automatic Fetal Ultrasound Standard Plane Detection Using Knowledge Transferred Recurrent Neural Networks

Fully-automated alignment of 3D fetal brain ultrasound to a canonical reference space using multi-task learning

Automated Craniofacial Biometry with 3D T2w Fetal MRI