Machine Learning in Metopic Craniosynostosis: Does Phenotypic Severity Predict Long-Term Esthetic Outcome?

Abstract

Background: There have been few longitudinal studies assessing the effect of preoperative phenotypic severity on long-term esthetic outcomes in metopic craniosynostosis. This study evaluates the relationship between metopic severity and long-term esthetic outcomes using interfrontal angle (IFA) and CranioRate, a novel metopic synostosis severity measure. Methods: Patients with metopic craniosynostosis who underwent bifrontal orbital advancement and remodeling between 2012 and 2017 were reviewed. Preoperative computed tomography head scans were analyzed for IFA and CranioRate, a machine learning algorithm which generates quantitative severity ratings including metopic severity score (MSS) and cranial morphology deviation (CMD). Long-term esthetic outcomes were assessed by craniofacial surgeons using blinded 3-rater esthetic grading of clinical photos. Raters assessed Whitaker score and the presence of temporal hollowing, lateral orbital retrusion, frontal bone irregularities and/or “any visible irregularities.” Results: Preoperative scans were performed at a mean age of 7.7±3.4 months, with average MSS of 6/10, CMD of 200/300, and IFA of 116.8±13.8 degrees. Patients underwent bifrontal orbital advancement and remodeling at mean 9.9±3.1 months. The average time from operation to esthetic assessment was 5.4±1.0 years. Pearson correlation revealed a significant negative correlation between MSS and age at computed tomography (r=−0.451, P=0.004) and IFA (r=−0.371, P=0.034) and between IFA and age at surgery (r=−0.383, P=0.018). In multinomial logistic regression, preoperative MSS was the only independent predictor of visible irregularities (odds ratio=2.18, B=0.780, P=0.024) and preoperative IFA alone significantly predicted Whitaker score, with more acute IFA predicting worse Whitaker score (odds ratio=0.928, B=−0.074, P=0.928). Conclusions: More severe preoperative phenotypes of metopic craniosynostosis were associated with worse esthetic dysmorphology. Objective measures of preoperative metopic severity predicted long-term esthetic outcomes.

Characterization of Regional Morphological Changes in Metopic Craniosynostosis Following Endoscopic Strip Craniectomy With Postoperative Helmeting: Predictors for Success

“…The recessed temporal areas are often a residual sequela even after prolonged helmet molding. It has also been shown that more severe objective preoperative metopic phenotypes are associated with worse esthetic dysmorphology, 14 which correlates with our findings.…”

Developing a Craniofacial Soft Tissue Anthropomorphic Database with Magnetic Resonance Imaging and Unbiased Diffeomorphic Registration

“…Previous studies reveal high degrees of subjectivity bias and low interrater reliability for popularly used methods (eg, Whitaker classification system). 5–7 Such variability and disagreement in appraising preoperative dysmorphology and appraising postoperative results renders determining surgical efficacy and comparing outcomes of different techniques difficult. Although two-dimensional normative measurements such as Farkas norms provide normative data for linear measurements, challenges persist in quantifying three-dimensional features, such as temporal retrusion, frontal bossing, and occipital bulleting.…”

“…Current popular methods of assessing postoperative results are subjective or demonstrate poor interrater reliability, even among experts. 4–6 Perhaps the most popular and widely adopted system for appraising postoperative craniofacial surgery outcomes is the Whitaker classification system, which assigns a score ranging from 1 to 4 based on the magnitude of subsequent surgery required to achieve aesthetic “normalcy.” 7 However, the Whitaker classification has demonstrated poor interrater reliability, likely secondary to its subjective nature and its pairing of an aesthetic appraisal to surgical management. 5,7 These subjective methods render appraising the quality of postoperative results or comparing the efficacy of surgical techniques difficult.…”

Validating the Supraorbital Notch-Nasion-Supraorbital Notch Angle as a Marker for Metopic Craniosynostosis Severity

“…11,12,[17][18][19][20][21][22] These measures and three-dimensional (3D) cranial morphometric analysis are combined in the CranioRate machine learning model that is now used by surgeons to measure metopic severity. [23][24][25][26] Our group recently proposed the supraorbital notch-nasionsupraorbital notch (SNS) notch angle as a meaningful measurement for identifying MCS and quantifying its severity. The purpose of this study is to validate the SNS angle using surgeon questionnaire responses, as well as the CranioRate tool.…”