Dental anomaly detection using intraoral photos via deep learning

Children with orofacial clefting (OFC) present with a wide range of dental anomalies. Identifying these anomalies is vital to understand their etiology and to discern the complex phenotypic spectrum of OFC. Such anomalies are currently identified using intra-oral exams by dentists, a costly and time-consuming process. We claim that automating the process of anomaly detection using deep neural networks (DNNs) could increase efficiency and provide reliable anomaly detection while potentially increasing the speed of research discovery. This study characterizes the use of` DNNs to identify dental anomalies by training a DNN model using intraoral photographs from the largest international cohort to date of children with nonsyndromic OFC and controls (OFC1). In this project, the intraoral images were submitted to a Convolutional Neural Network model to perform multi-label multi-class classification of 10 dental anomalies. The network predicts whether an individual exhibits any of the 10 anomalies and can do so significantly faster than a human rater can. For all but three anomalies, F1 scores suggest that our model performs competitively at anomaly detection when compared to a dentist with 8 years of clinical experience. In addition, we use saliency maps to provide a post-hoc interpretation for our model's predictions. This enables dentists to examine and verify our model's predictions.

Tooth Agenesis Patterns in Orofacial Clefting Using Tooth Agenesis Code: A Meta-Analysis

Individuals with orofacial clefting (OFC) have a higher prevalence of tooth agenesis (TA) overall. Neither the precise etiology of TA, nor whether TA occurs in patterns that differ by gender or cleft type is yet known. This meta-analysis aims to identify the spectrum of tooth agenesis patterns in subjects with non-syndromic OFC and controls using the Tooth Agenesis Code (TAC) program. An indexed search of databases (PubMed, EMBASE, and CINAHL) along with cross-referencing and hand searches were completed from May to June 2019 and re-run in February 2022. Additionally, unpublished TAC data from 914 individuals with OFC and 932 controls were included. TAC pattern frequencies per study were analyzed using a random effects meta-analysis model. A thorough review of 45 records retrieved resulted in 4 articles meeting eligibility criteria, comprising 2182 subjects with OFC and 3171 controls. No TA (0.0.0.0) was seen in 51% of OFC cases and 97% of controls. TAC patterns 0.2.0.0, 2.0.0.0, and 2.2.0.0 indicating uni- or bi-lateral missing upper laterals, and 16.0.0.0 indicating missing upper right second premolar, were more common in subjects with OFC. Subjects with OFC have unique TA patterns and defining these patterns will help increase our understanding of the complex etiology underlying TA.

FOXE1 association with both isolated cleft lip with or without cleft palate, and isolated cleft palate

Nonsyndromic orofacial clefts are a common complex birth defect caused by genetic and environmental factors and/or their interactions. A previous genome-wide linkage scan discovered a novel locus for cleft lip with or without cleft palate (CL/P) at 9q22-q33. To identify the etiologic gene, we undertook an iterative and complementary fine mapping strategy using family-based CL/P samples from Colombia, USA and the Philippines. Candidate genes within 9q22-q33 were sequenced, revealing 32 new variants. Concurrently, 397 SNPs spanning the 9q22-q33 2-LOD-unit interval were tested for association. Significant SNP and haplotype association signals (P = 1.45E - 08) narrowed the interval to a 200 kb region containing: FOXE1, C9ORF156 and HEMGN. Association results were replicated in CL/P families of European descent and when all populations were combined the two most associated SNPs, rs3758249 (P = 5.01E - 13) and rs4460498 (P = 6.51E - 12), were located inside a 70 kb high linkage disequilibrium block containing FOXE1. Association signals for Caucasians and Asians clustered 5' and 3' of FOXE1, respectively. Isolated cleft palate (CP) was also associated, indicating that FOXE1 plays a role in two phenotypes thought to be genetically distinct. Foxe1 expression was found in the epithelium undergoing fusion between the medial nasal and maxillary processes. Mutation screens of FOXE1 identified two family-specific missense mutations at highly conserved amino acids. These data indicate that FOXE1 is a major gene for CL/P and provides new insights for improved counseling and genetic interaction studies.

Genome scan, fine-mapping, and candidate gene analysis of non-syndromic cleft lip with or without cleft palate reveals phenotype-specific differences in linkage and association results

OBJECTIVES

Non-syndromic orofacial clefts, i.e. cleft lip (CL) and cleft palate (CP), are among the most common birth defects. The goal of this study was to identify genomic regions and genes for CL with or without CP (CL/P).

METHODS

We performed linkage analyses of a 10 cM genome scan in 820 multiplex CL/P families (6,565 individuals). Significant linkage results were followed by association analyses of 1,476 SNPs in candidate genes and regions, utilizing a weighted false discovery rate (wFDR) approach to control for multiple testing and incorporate the genome scan results.

RESULTS

Significant (multipoint HLOD >or=3.2) or genome-wide-significant (HLOD >or=4.02) linkage results were found for regions 1q32, 2p13, 3q27-28, 9q21, 12p11, 14q21-24 and 16q24. SNPs in IRF6 (1q32) and in or near FOXE1 (9q21) reached formal genome-wide wFDR-adjusted significance. Further, results were phenotype dependent in that the IRF6 region results were most significant for families in which affected individuals have CL alone, and the FOXE1 region results were most significant in families in which some or all of the affected individuals have CL with CP.

CONCLUSIONS

These results highlight the importance of careful phenotypic delineation in large samples of families for genetic analyses of complex, heterogeneous traits such as CL/P.

CRISPLD2: a novel NSCLP candidate gene

Non-syndromic cleft lip with or without cleft palate (NSCLP) results from the complex interaction between genes and environmental factors. Candidate gene analysis and genome scans have been employed to identify the genes contributing to NSCLP. In this study, we evaluated the 16q24.1 chromosomal region, which has been identified by multiple genome scans as an NSCLP region of interest. Two candidate genes were found in the region: interferon regulatory factor 8 (IRF8) and cysteine-rich secretory protein LCCL domain containing 2 (CRISPLD2). Initially, Caucasian and Hispanic NSCLP multiplex families and simplex parent-child trios were genotyped for single nucleotide polymorphisms (SNPs) in both IRF8 and CRISPLD2. CRISPLD2 was subsequently genotyped in a data set comprised of NSCLP families from Colombia, South America. Linkage disequilibrium analysis identified a significant association between CRISPLD2 and NSCLP in both our Caucasian and Hispanic NSCLP cohorts. SNP rs1546124 and haplotypes between rs1546124 and either rs4783099 or rs16974880 were significant in the Caucasian multiplex population (P=0.01, P=0.002 and P=0.001, respectively). An altered transmission of CRISPLD2 SNPs rs8061351 (P=0.02) and rs2326398 (P=0.06) was detected in the Hispanic population. No association was found between CRISPLD2 and our Colombian population or IRF8 and NSCLP. In situ hybridization showed that CRISPLD2 is expressed in the mandible, palate and nasopharynx regions during craniofacial development at E13.5-E17.5, respectively. Altogether, these data suggest that genetic variation in CRISPLD2 has a role in the etiology of NSCLP.