Automated Craniofacial Biometry with 3D T2w Fetal MRI

Abstract

Objectives: Evaluating craniofacial phenotype-genotype correlations prenatally is increasingly important; however, it is subjective and challenging with 3D ultrasound. We developed an automated landmark propagation pipeline using 3D motion-corrected, slice-to-volume reconstructed (SVR) fetal MRI for craniofacial measurements.

Methods: A literature review and expert consensus identified 31 craniofacial biometrics for fetal MRI. An MRI atlas with defined anatomical landmarks served as a template for subject registration, auto-labelling, and biometric calculation. We assessed 108 healthy controls and 24 fetuses with Down syndrome (T21) in the third trimester (29-36 weeks gestational age, GA) to identify meaningful biometrics in T21. Reliability and reproducibility were evaluated in 10 random datasets by four observers.

Results: Automated labels were produced for all 132 subjects with a 0.03% placement error rate. Seven measurements, including anterior base of skull length and maxillary length, showed significant differences with large effect sizes between T21 and control groups (ANOVA, p<0.001). Manual measurements took 25-35 minutes per case, while automated extraction took approximately 5 minutes. Bland-Altman plots showed agreement within manual observer ranges except for mandibular width, which had higher variability. Extended GA growth charts (19-39 weeks), based on 280 control fetuses, were produced for future research.

Conclusion: This is the first automated atlas-based protocol using 3D SVR MRI for fetal craniofacial biometrics, accurately revealing morphological craniofacial differences in a T21 cohort. Future work should focus on improving measurement reliability, larger clinical cohorts, and technical advancements, to enhance prenatal care and phenotypic characterisation.

Figure 1.

a. GA distribution of control subject datasets in the study per MRI protocol; b. proportional distribution of MRI protocols in the whole control group (MRI protocol (field strength/TE) = 1.5T/ 180ms; 1.5T/ 80ms; 3T/ 180ms; or, 3T/ 250ms); and examples of a 3D SVR fetal head reconstructions at different acquisition parameters.

Figure 2.

Visual representation of 3D landmarks placed within a 3D population-averaged MRI atlas (31 weeks GA)

Figure 3.

Proposed pipeline for atlas-based 3D craniofacial biometry for fetal MRI (orange boxes)

Figure 4.

Upper row: Face SVR quality assessment results stratified by MRI field strength (1.5T/3T) and gestational age. Lower Row: i. automated landmarks with labels, ii. Example: Excellent quality SVR with good landmark placement, iii. Example: Moderate quality SVR (poor contrast resolution lower face), with poor lip label placement due to limited boundary definition adjacent to maternal tissue. (Label key: Ve=vertex; Oc=occiput; Si=sinciput; NaIn=inner nasion; NaO=outer nasion; Fc=foramen caecum; NBt=nasal bone tip; v=vomer; H=hormion; PcP=posterior clinoid process; Ppw=posterior pharyngeal wall; Pav=palate vault; Ba=basion; ANS=anterior nasal spine; PNS; posterior nasal spine; Tp=posterior tongue; Me=bony mentum; Lip=lip; Chin=chin)

Figure 5.

Bland Altman plots of absolute and relative difference for observer 0, grey diamond (using manual landmark-based indirect measurement method), observers 2 and 3, grey square and triangle respectively (using direct manual measurement method), and automated biometry, red circle, all compared to expert observer 1 (direct manual method) for a selection of biometrics (ABSL, IFA, HPL, and NASO). Grey dash=mean difference for automated biometry and red dash=upper and lower limits of agreement for automated method.

Figure 6.

Growth charts for seven craniofacial biometrics* with statistically significant differences from control group and a large effect size, and biparietal diameter, BPD, with no significant differences between groups. (*ASBL, anterior skull base length; VPL, velopharyngeal length; HPL, hard palate length; MXL, maxillary length, IFA, inferior facial angle; NASO, nasopharyngeal area; and, occipitofontal diameter.)

Figure 7.

Selected growth charts for craniofacial biometry from 280 normal controls (blue circles) during 2nd and 3rd trimesters. The quadratic or linear regression equation for the 50th centile bestfit line is included on the chart (y = the mean measurement of the variable under investigation, and GA is the selected gestational age).