Describing Crouzon and Pfeiffer syndrome based on principal component analysis

Variations in orbital morphology, globe:orbit volume relation, and ophthalmological outcome in unicoronal synostosis

Nonsyndromic unicoronal synostosis is associated with variability of severity in orbital morphology and ophthalmological manifestations. The relation between the two is not fully understood, nor how surgical treatment with fronto-orbital advancement and remodelling (FOAR) changes the relation. The aim of this study was to elucidate associations between ophthalmological manifestations and variations in orbital morphology and globe:orbit volume ratios preoperatively and at long-term follow-up after surgery. Twelve children referred to Uppsala Craniofacial Center who underwent computed tomography and standardized ophthalmological examinations regarding strabismus, spherical equivalent, astigmatism, anisometropia, and subnormal vision preoperatively and at 3 years of age were included. Orbits and globes were segmented. Principal component analysis elucidated morphological variation, and symmetry between orbital pairs was measured as the Dice similarity coefficient and globe:orbit volume ratios were calculated. The defined orbital shape variations were correlated with strabismus, refractive error, and subnormal vision. Different shape variations were associated with strabismus pre- and postoperatively and ipsi- and contralateral astigmatism. Greater improvement in orbital symmetry after surgery was associated with improvement in astigmatic anisometropia and new onset strabismus at follow-up. A small globe:orbit volume ratio was associated with preoperative strabismus, while the opposite was seen at follow-up. Different mechanisms seem to cause strabismus pre- and postoperatively, and FOAR might not sufficiently correct orbital morphology.

AI-based diagnosis and phenotype - Genotype correlations in syndromic craniosynostoses

Apert (AS), Crouzon (CS), Muenke (MS), Pfeiffer (PS), and Saethre Chotzen (SCS) are among the most frequently diagnosed syndromic craniosynostoses. The aims of this study were (1) to train an innovative model using artificial intelligence (AI)-based methods on two-dimensional facial frontal, lateral, and external ear photographs to assist diagnosis for syndromic craniosynostoses vs controls, and (2) to screen for genotype/phenotype correlations in AS, CS, and PS. We included retrospectively and prospectively, from 1979 to 2023, all frontal and lateral pictures of patients genetically diagnosed with AS, CS, MS, PS and SCS syndromes. After a deep learning-based preprocessing, we extracted geometric and textural features and used XGboost (eXtreme Gradient Boosting) to classify patients. The model was tested on an independent international validation set of genetically confirmed patients and non-syndromic controls. Between 1979 and 2023, we included 2228 frontal and lateral facial photographs corresponding to 541 patients. In all, 70.2% [0.593-0.797] (p < 0.001) of patients in the validation set were correctly diagnosed. Genotypes linked to a splice donor site of FGFR2 in Crouzon-Pfeiffer syndrome (CPS) caused a milder phenotype in CPS. Here we report a new method for the automatic detection of syndromic craniosynostoses using AI.

Understanding the influence of surgical parameters on craniofacial surgery outcomes: a computational study

Sagittal craniosynostosis (SC) is a congenital condition whereby the newborn skull develops abnormally owing to the premature ossification of the sagittal suture. Spring-assisted cranioplasty (SAC) is a minimally invasive surgical technique to treat SC, where metallic distractors are used to reshape the newborn's head. Although safe and effective, SAC outcomes remain uncertain owing to the limited understanding of skull-distractor interaction and the limited information provided by the analysis of single surgical cases. In this work, an SC population-averaged skull model was created and used to simulate spring insertion by means of the finite-element analysis using a previously developed modelling framework. Surgical parameters were varied to assess the effect of osteotomy and spring positioning, as well as distractor combinations, on the final skull dimensions. Simulation trends were compared with retrospective measurements from clinical imaging (X-ray and three-dimensional photogrammetry scans). It was found that the on-table post-implantation head shape change is more sensitive to spring stiffness than to the other surgical parameters. However, the overall end-of-treatment head shape is more sensitive to spring positioning and osteotomy size parameters. The results of this work suggest that SAC surgical planning should be performed in view of long-term results, rather than immediate on-table reshaping outcomes.

The Accuracy of Computer-Assisted Surgical Planning in Predicting Soft Tissue Responses After Le Fort I Osteotomy: Retrospective Analysis

PURPOSE

Mismatch between preoperative planning and surgical outcome in maxillofacial surgery relate to on-table replication of presurgical planning and predictive algorithm inaccuracy: software error was hereby decoupled from planning inaccuracy to assess a commercial software. The hypothesis was that soft tissue prediction error would be minimized if the surgical procedure was replicated precisely as planned and is independent of the extent of bone repositioning.

MATERIALS AND METHODS

Cone-beam computed tomography scans of 16 Le Fort I osteotomy patients were collected at Boston Children's Hospital. Preoperative and postoperative models of bone and soft tissue were constructed and the maxilla repositioning was replicated. Each model was subdivided into 6 regions: mouth, nose, eyes, and cheeks. Soft tissue prediction (performed using Proplan CMF-Materialise) for each patient was compared with the relative postoperative reconstruction and error was determined. P <0.05 was considered significant.

RESULTS

Le Fort I segment repositioning was replicated within 0.70±0.18 mm. The highest prediction error was found in the mouth (1.49±0.77 mm) followed by the cheeks (0.98±0.34 mm), nose (0.86±0.23 mm), and eyes (0.76±0.32). Prediction error on cheeks correlated significantly with mouth ( r =0.63, P < 0.01) and nose ( r =0.67, P < 0.01). Mouth prediction error correlated with total advancement ( r =0.52, P =0.04).

CONCLUSIONS

ProPlan CMF is a useful outcome prediction tool; however, accuracy decreases with the extent of maxillary advancement even when errors in surgical replication are minimized.

Applications of 3D photography in craniofacial surgery

Three-dimensional (3D) photography is becoming more common in craniosynostosis practice and may be used for research, archiving, and as a planning tool. In this article, an overview of the uses of 3D photography will be given, including systems available and illustrations of how they can be used. Important innovations in 3D computer vision will also be discussed, including the potential role of statistical shape modeling and analysis as an outcomes tool with presentation of some results and a review of the literature on the topic. Potential future applications in diagnostics using machine learning will also be presented.

Rare Genetic Syndromes and Oral Anomalies: A Review of the Literature and Case Series with a New Classification Proposal

Rare genetic syndromes, conditions with a global average prevalence of 40 cases/100,000 people, are associated with anatomical, physiological, and neurological anomalies that may affect different body districts, including the oral district. So far, no classification of oral abnormalities in rare genetic syndromes is present in the literature. The aim of this narrative review is to analyze literature on rare genetic syndromes affecting dento-oro-maxillofacial structures (teeth, maxillary bones, oral soft tissues, or mixed) and to propose a classification according to the detected oral abnormalities. In addition, five significant cases of rare genetic syndromes are presented. The Scale for the Assessment of Narrative Review Articles (SANRA) was followed for this review. From 674 papers obtained through PubMed search, 351 were selected. Sixty-two rare genetic syndromes involving oral manifestations were found and classified. The proposed classification aims to help the clinician to easily understand which dento-oro-maxillofacial findings might be expected in the presence of each rare genetic syndrome. This immediate framework may both help in the diagnosis of dento-oro-maxillofacial anomalies related to the underlying pathology as well as facilitate the drafting of treatment plans with the involvement of a multidisciplinary team.

The 3D skull 0-4 years: A validated, generative, statistical shape model

Background

This study aims to capture the 3D shape of the human skull in a healthy paediatric population (0-4 years old) and construct a generative statistical shape model.

Methods

The skull bones of 178 healthy children (55% male, 20.8 ± 12.9 months) were reconstructed from computed tomography (CT) images. 29 anatomical landmarks were placed on the 3D skull reconstructions. Rotation, translation and size were removed, and all skull meshes were placed in dense correspondence using a dimensionless skull mesh template and a non-rigid iterative closest point algorithm. A 3D morphable model (3DMM) was created using principal component analysis, and intrinsically and geometrically validated with anthropometric measurements. Synthetic skull instances were generated exploiting the 3DMM and validated by comparison of the anthropometric measurements with the selected input population.

Results

The 3DMM of the paediatric skull 0-4 years was successfully constructed. The model was reasonably compact - 90% of the model shape variance was captured within the first 10 principal components. The generalisation error, quantifying the ability of the 3DMM to represent shape instances not encountered during training, was 0.47 mm when all model components were used. The specificity value was <0.7 mm demonstrating that novel skull instances generated by the model are realistic. The 3DMM mean shape was representative of the selected population (differences <2%). Overall, good agreement was observed in the anthropometric measures extracted from the selected population, and compared to normative literature data (max difference in the intertemporal distance) and to the synthetic generated cases.

Conclusion

This study presents a reliable statistical shape model of the paediatric skull 0-4 years that adheres to known skull morphometric measures, can accurately represent unseen skull samples not used during model construction and can generate novel realistic skull instances, thus presenting a solution to limited availability of normative data in this field.

Towards a Complete 3D Morphable Model of the Human Head

Three-dimensional morphable models (3DMMs) are powerful statistical tools for representing the 3D shapes and textures of an object class. Here we present the most complete 3DMM of the human head to date that includes face, cranium, ears, eyes, teeth and tongue. To achieve this, we propose two methods for combining existing 3DMMs of different overlapping head parts: (i). use a regressor to complete missing parts of one model using the other, and (ii). use the Gaussian Process framework to blend covariance matrices from multiple models. Thus, we build a new combined face-and-head shape model that blends the variability and facial detail of an existing face model (the LSFM) with the full head modelling capability of an existing head model (the LYHM). Then we construct and fuse a highly-detailed ear model to extend the variation of the ear shape. Eye and eye region models are incorporated into the head model, along with basic models of the teeth, tongue and inner mouth cavity. The new model achieves state-of-the-art performance. We use our model to reconstruct full head representations from single, unconstrained images allowing us to parameterize craniofacial shape and texture, along with the ear shape, eye gaze and eye color.

Influence of Monobloc/Le Fort III Surgery on the Developing Posterior Maxillary Dentition and Its Resultant Effect on Orthognathic Surgery

BACKGROUND

Timing of frontofacial surgery for the syndromic craniosynostosis as it relates to various surgical risks has not been adequately studied. The purpose of this study was to investigate posterior dental complications of midface advancement in patients with syndromic craniosynostosis undergoing surgery at different ages and the effects on subsequent orthognathic surgery.

METHODS

A retrospective chart review of patients with syndromic craniosynostosis treated with midface advancement (monobloc or Le Fort III) from 1999 to 2018 was carried out. Patient demographics, records, and imaging studies were reviewed. A subanalysis of those patients who were also treated with orthognathic surgery from 2014 to 2018 with imaging studies available for analysis was also performed.

RESULTS

Thirty-seven patients met the inclusion criteria. Sixty-four percent of the patients had radiographic evidence of maxillary molar dental abnormality. Older age at the time of surgery was significantly associated with a lower odds of sustaining dental injury (OR, 0.55; p = 0.034). The odds of damaging second or third maxillary molars was significantly higher with a younger age at the time of surgery (p = 0.021 and p = 0.034). The odds of sustaining dental injury increased moving posteriorly, showing the risk of abnormal pattern of M3 greater than M2 greater than M1. Advanced age at the time of surgery was significantly associated with decreased odds of dental injury (OR, 0.55; p = 0.034).

CONCLUSIONS

Damage to the developing permanent maxillary molars may affect orthodontic management, mastication, and potentially maxillary development. Delaying frontofacial surgery until development of the permanent maxillary dentition should be considered if other indications do not mandate earlier intervention.

Modelling growth curves of the normal infant's mandible: 3D measurements using computed tomography

Objectives

Data on normal mandibular development in the infant is lacking though essential to understand normal growth patterns and to discriminate abnormal growth. The aim of this study was to provide normal linear measurements of the mandible using computed tomography performed in infants from 0 to 2 years of age.

Material and methods

3D voxel software was used to calculate mandibular body length, mandibular ramus length, bicondylar width, bigonial width and the gonial angle. Intra- and inter-rater reliability was assessed for these measurements. They were found to be sufficient for all distances; intra-class correlation coefficients were all above 0.9. Regression analysis for growth modelling was performed.

Results

In this multi-centre retrospective study, 109 CT scans were found eligible that were performed for various reasons (e.g. trauma, craniosynostosis, craniofacial abscesses). Craniosynostosis patients had larger mandibular measurements compared to non-craniosynostosis patients and were therefore excluded. Fifty-one CT scans were analysed.

Conclusions

Analysis showed that the mandible increases more in size vertically (the mandibular ramus) than horizontally (the mandibular body). Most of the mandibular growth occurs in the first 6 months.

Clinical relevance

These growth models provide insight into normal mandibular development in the first 2 years of life. This reference data facilitates discrimination between normal and abnormal mandibular growth.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00784-021-03937-1.

Statistical shape modelling for the analysis of head shape variations

The aim of this study is, firstly, to create a population-based 3D head shape model for the 0 to 2-year-old subjects to describe head shape variability within a normal population and, secondly, to test a combined normal and sagittal craniosynostosis (SAG) population model, able to provide surgical outcome assessment. 3D head shapes of patients affected by non-cranial related pathologies and of SAG patients (pre- and post-op) were extracted either from head CTs or 3D stereophotography scans, and processed. Statistical shape modelling (SSM) was used to describe shape variability using two models - a normal population model (MODEL1) and a combined normal and SAG population model (MODEL2). Head shape variability was described via principal components analysis (PCA) which calculates shape modes describing specific shape features. MODEL1 (n = 65) mode 1 showed statistical correlation (p < 0.001) with width (125.8 ± 13.6 mm), length (151.3 ± 17.4 mm) and height (112.5 ± 11.1 mm) whilst mode 2 showed correlation with cranial index (83.5 mm ± 6.3 mm, p < 0.001). The remaining 9 modes showed more subtle head shape variability. MODEL2 (n = 159) revealed that post-operative head shape still did not achieve full shape normalization with either spring cranioplasty or total calvarial remodelling. This study proves that SSM has the potential to describe detailed anatomical variations in a paediatric population.

On Learning 3D Face Morphable Model from In-the-Wild Images

As a classic statistical model of 3D facial shape and albedo, 3D Morphable Model (3DMM) is widely used in facial analysis, e.g., model fitting, image synthesis. Conventional 3DMM is learned from a set of 3D face scans with associated well-controlled 2D face images, and represented by two sets of PCA basis functions. Due to the type and amount of training data, as well as, the linear bases, the representation power of 3DMM can be limited. To address these problems, this paper proposes an innovative framework to learn a nonlinear 3DMM model from a large set of in-the-wild face images, without collecting 3D face scans. Specifically, given a face image as input, a network encoder estimates the projection, lighting, shape and albedo parameters. Two decoders serve as the nonlinear 3DMM to map from the shape and albedo parameters to the 3D shape and albedo, respectively. With the projection parameter, lighting, 3D shape, and albedo, a novel analytically-differentiable rendering layer is designed to reconstruct the original input face. The entire network is end-to-end trainable with only weak supervision. We demonstrate the superior representation power of our nonlinear 3DMM over its linear counterpart, and its contribution to face alignment, 3D reconstruction, and face editing. Source code and additional results can be found at our project page: http://cvlab.cse.msu.edu/project-nonlinear-3dmm.html.

Average Models and 3-dimensional Growth Patterns of the Healthy Infant Cranium

Treatment of cranial deformity is often performed during infancy in cases such as craniosynostosis and deformational plagiocephaly. To acquire morphologic standards for the treatment goals of these conditions, we created cranial average models and elucidated the growth patterns of the cranium of healthy infants in 3-dimension (3D) using homologous modeling.

A novel RBF-based predictive tool for facial distraction surgery in growing children with syndromic craniosynostosis

Purpose

Predicting changes in face shape from corrective surgery is challenging in growing children with syndromic craniosynostosis. A prediction tool mimicking composite bone and skin movement during facial distraction would be useful for surgical audit and planning. To model surgery, we used a radial basis function (RBF) that is smooth and continuous throughout space whilst corresponding to measured distraction at landmarks. Our aim is to showcase the pipeline for a novel landmark-based, RBF-driven simulation for facial distraction surgery in children.

Methods

An individual’s dataset comprised of manually placed skin and bone landmarks on operated and unoperated regions. Surgical warps were produced for ‘older’ monobloc, ‘older’ bipartition and ‘younger’ bipartition groups by applying a weighted least-squares RBF fitted to the average landmarks and change vectors. A ‘normalisation’ warp, from fitting an RBF to craniometric landmark differences from the average, was applied to each dataset before the surgical warp. The normalisation was finally reversed to obtain the individual prediction. Predictions were compared to actual post-operative outcomes.

Results

The averaged change vectors for all groups showed skin and bone movements characteristic of the operations. Normalisation for shape–size removed individual asymmetry, size and proportion differences but retained typical pre-operative shape features. The surgical warps removed the average syndromic features. Reversing the normalisation reintroduced the individual’s variation into the prediction. The mid-facial regions were well predicted for all groups. Forehead and brow regions were less well predicted.

Conclusions

Our novel, landmark-based, weighted RBF can predict the outcome for facial distraction in younger and older children with a variety of head and face shapes. It can replicate the surgical reality of composite bone and skin movement jointly in one model. The potential applications include audit of existing patient outcomes, and predicting outcome for new patients to aid surgical planning.

Analysis of Cranial Morphology of Healthy Infants Using Homologous Modeling

OBJECTIVE

Data on cranial morphology of healthy individuals can be used as the guide in the treatment of cranial deformity. There are many reports analyzing the cranial morphology of healthy children in the past. But most of them focus on 2-dimensional values, and there are only a few reports, which analyzed the cranial morphology of Japanese healthy infants. We report a novel method that enables the comprehensive analysis of cranial morphology of Japanese healthy infants in 3D.

METHODS

Craniofacial CT data of 20 healthy infants (9 males, 11 females) ranging in age from 1 to 11 months were collected. Based on the CT data, we created 20 homologous models of cranium using software specifically designed to support homologous modeling. We averaged vertex coordinates of the homologous models to create average model. We further performed principal component analysis, and created virtual models based on each principal component. The contribution rate was calculated, and the features described by each principal component were interpreted.

RESULTS

We created the average cranial model of Japanese healthy infants. Seven principal components (cumulative contribution rate: 89.218%) were interpreted as to which part of the cranial shape each component was related to. The elements were extracted that may characterize the cranial morphology of some of the clinical conditions such as dolico/brachycephaly and deformational plagiocephaly. Some of these elements have not been mentioned in the past literature.

CONCLUSION

Homologous modeling was considered to be valid and strong tool for comprehensive analysis of cranial morphology.

Using principal component analysis to describe the midfacial deformities in patients with craniofacial microsomia

PURPOSE

Craniofacial microsomia (CFM) is the result of a disturbance in embryologic development and is characterised by an asymmetric, mostly unilateral facial underdevelopment. The aim of this study is to understand the midfacial involvement in CFM using principal component analysis (PCA).

MATERIALS AND METHODS

Pre-operative data from 19 CFM and 23 control patients were collected. A set of 71 landmarks was placed on three-dimensional (3D) reconstructions of all skulls to compare both populations. PCA visualised variation within both groups and calculated the vector of change. Linear measurements were taken to compare ratios between the populations and between the affected and unaffected sides in CFM patients.

RESULTS

PCA defined a vector that described shape changes between both populations. Videos showed the variation within the control and CFM group and the transformation from a mean CFM skull into a normal phenotype. Linear measurements showed a significant difference between the affected and unaffected sides in CFM patients.

CONCLUSION

PCA has not been applied on asymmetrical data before, but it has proved to be a useful method to describe CFM. The virtual normalisation of a mean CFM skull enables visualisation of the bony shape changes, which is promising to delineate and to plan surgical correction and could be used as an outcome measure.

Extensive phenotyping of the orofacial and dental complex in Crouzon syndrome

OBJECTIVES

Fibroblast growth factor receptor 2 (FGFR2) ^C342Y/+ mutation is a known cause of Crouzon syndrome that is characterised by craniosynostosis and midfacial hypoplasia. Our aim was to conduct extensive phenotyping of the maxillary, mandibular and dental morphology associated with this mutation.

MATERIALS AND METHODS

Morphometric data were obtained from 40 mice, representing two genotypes (Crouzon and wild-type) and two sexes (males and females) (n=10 in each group). Dental analysis further categorised the first molars into the two jaws (maxillary and mandibular) (n=20 in each group). Maxillary, mandibular and dental morphology was compared by analysing 23 linear landmark-based dimensions in three-dimensional micro-computed tomography reconstructions.

RESULTS

Compared with wild-type, Crouzon (FGFR2^C342Y/+) maxillae were significantly shorter in maximum height, anterior and posterior lengths and middle width, but larger in posterior width (p<0.05 for height; p<0.001 for other comparisons). In the Crouzon mandible, the ascending and descending heights, effective and mandibular lengths, and intercoronoid and intercondylar widths were significantly shorter, whereas intergonial width was larger (p<0.01 for intercondylar width; p<0.001 for other comparisons). Crouzon teeth were significantly smaller mesiodistally, but larger in crown height (p<0.001 for each comparison). All Crouzon mice presented with bifid mandibular condyles and a quarter presented with expansive bone lesions in the mandibular incisor alveolus.

CONCLUSIONS

Our findings of hypoplasia in all three planes in Crouzon maxillae and mandibles, together with the presence of bifid mandibular condyles and expansive bone lesions, may be relevant to maxillofacial surgery and orthodontics. Beyond skeletal effects, the FGFR2^C342Y/+ mutation is now implicated in affecting tooth development. This study's skeletal phenomics data also provides baseline data against which the effect of various treatments can now be assessed.

Statistical shape modelling to aid surgical planning: associations between surgical parameters and head shapes following spring-assisted cranioplasty

PURPOSE

Spring-assisted cranioplasty is performed to correct the long and narrow head shape of children with sagittal synostosis. Such corrective surgery involves osteotomies and the placement of spring-like distractors, which gradually expand to widen the skull until removal about 4 months later. Due to its dynamic nature, associations between surgical parameters and post-operative 3D head shape features are difficult to comprehend. The current study aimed at applying population-based statistical shape modelling to gain insight into how the choice of surgical parameters such as craniotomy size and spring positioning affects post-surgical head shape.

METHODS

Twenty consecutive patients with sagittal synostosis who underwent spring-assisted cranioplasty at Great Ormond Street Hospital for Children (London, UK) were prospectively recruited. Using a nonparametric statistical modelling technique based on mathematical currents, a 3D head shape template was computed from surface head scans of sagittal patients after spring removal. Partial least squares (PLS) regression was employed to quantify and visualise trends of localised head shape changes associated with the surgical parameters recorded during spring insertion: anterior-posterior and lateral craniotomy dimensions, anterior spring position and distance between anterior and posterior springs.

RESULTS

Bivariate correlations between surgical parameters and corresponding PLS shape vectors demonstrated that anterior-posterior (Pearson's [Formula: see text]) and lateral craniotomy dimensions (Spearman's [Formula: see text]), as well as the position of the anterior spring ([Formula: see text]) and the distance between both springs ([Formula: see text]) on average had significant effects on head shapes at the time of spring removal. Such effects were visualised on 3D models.

CONCLUSIONS

Population-based analysis of 3D post-operative medical images via computational statistical modelling tools allowed for detection of novel associations between surgical parameters and head shape features achieved following spring-assisted cranioplasty. The techniques described here could be extended to other cranio-maxillofacial procedures in order to assess post-operative outcomes and ultimately facilitate surgical decision making.

Characterizing the skull base in craniofacial microsomia using principal component analysis

The aim of this study was to compare the anatomical differences in the skull base between the affected and non-affected side in patients with craniofacial microsomia (CFM), and to compare the affected and non-affected sides with measurements from a normal population. Three-dimensional computed tomography scans of 13 patients with unilateral CFM and 19 normal patients (age range 7-12 years) were marked manually with reliable homologous landmarks. Principal component analysis (PCA), as part of a point distribution model (PDM), was used to analyse the variability within the normal and preoperative CFM patient groups. Through analysis of the differences in the principal components calculated for the two groups, a model was created to describe the differences between CFM patients and normal age-matched controls. The PDMs were also used to describe the shape changes in the skull base between the cohorts and validated this model. Using thin-plate splines as a means of interpolation, videos were created to visualize the transformation from CFM skull to normal skull, and to display the variability in shape changes within the groups themselves. In CFM cases, the skull base showed significant asymmetry. Anatomical areas around the glenoid fossa and mastoid process showed the most asymmetry and restriction of growth, suggesting a pathology involving the first and second pharyngeal arches.

Quantitative analysis of fetal facial morphology using 3D ultrasound and statistical shape modeling: a feasibility study

BACKGROUND

The antenatal detection of facial dysmorphism using 3-dimensional ultrasound may raise the suspicion of an underlying genetic condition but infrequently leads to a definitive antenatal diagnosis. Despite advances in array and noninvasive prenatal testing, not all genetic conditions can be ascertained from such testing.

OBJECTIVES

The aim of this study was to investigate the feasibility of quantitative assessment of fetal face features using prenatal 3-dimensional ultrasound volumes and statistical shape modeling. STUDY DESIGN: Thirteen normal and 7 abnormal stored 3-dimensional ultrasound fetal face volumes were analyzed, at a median gestation of 29⁺⁴ weeks (25⁺⁰ to 36⁺¹). The 20 3-dimensional surface meshes generated were aligned and served as input for a statistical shape model, which computed the mean 3-dimensional face shape and 3-dimensional shape variations using principal component analysis.

RESULTS

Ten shape modes explained more than 90% of the total shape variability in the population. While the first mode accounted for overall size differences, the second highlighted shape feature changes from an overall proportionate toward a more asymmetric face shape with a wide prominent forehead and an undersized, posteriorly positioned chin. Analysis of the Mahalanobis distance in principal component analysis shape space suggested differences between normal and abnormal fetuses (median and interquartile range distance values, 7.31 ± 5.54 for the normal group vs 13.27 ± 9.82 for the abnormal group) (P = .056).

CONCLUSION

This feasibility study demonstrates that objective characterization and quantification of fetal facial morphology is possible from 3-dimensional ultrasound. This technique has the potential to assist in utero diagnosis, particularly of rare conditions in which facial dysmorphology is a feature.

Large Scale 3D Morphable Models

We present large scale facial model (LSFM)-a 3D Morphable Model (3DMM) automatically constructed from 9663 distinct facial identities. To the best of our knowledge LSFM is the largest-scale Morphable Model ever constructed, containing statistical information from a huge variety of the human population. To build such a large model we introduce a novel fully automated and robust Morphable Model construction pipeline, informed by an evaluation of state-of-the-art dense correspondence techniques. The dataset that LSFM is trained on includes rich demographic information about each subject, allowing for the construction of not only a global 3DMM model but also models tailored for specific age, gender or ethnicity groups. We utilize the proposed model to perform age classification from 3D shape alone and to reconstruct noisy out-of-sample data in the low-dimensional model space. Furthermore, we perform a systematic analysis of the constructed 3DMM models that showcases their quality and descriptive power. The presented extensive qualitative and quantitative evaluations reveal that the proposed 3DMM achieves state-of-the-art results, outperforming existing models by a large margin. Finally, for the benefit of the research community, we make publicly available the source code of the proposed automatic 3DMM construction pipeline, as well as the constructed global 3DMM and a variety of bespoke models tailored by age, gender and ethnicity.