Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2022 Jun;30(6):682-686.
doi: 10.1038/s41431-021-00994-8. Epub 2021 Nov 22.

Using deep-neural-network-driven facial recognition to identify distinct Kabuki syndrome 1 and 2 gestalt

Flavien Rouxel #  1 , Kevin Yauy #  1 , Guilaine Boursier  2 , Vincent Gatinois  3 , Mouna Barat-Houari  2 , Elodie Sanchez  1 , Didier Lacombe  4 , Stéphanie Arpin  5 , Fabienne Giuliano  6 , Damien Haye  7   8 , Marlène Rio  9 , Annick Toutain  5 , Klaus Dieterich  10 , Elise Brischoux-Boucher  11 , Sophie Julia  12 , Mathilde Nizon  13 , Alexandra Afenjar  14 , Boris Keren  8 , Aurelia Jacquette  8 , Sebastien Moutton  15 , Marie-Line Jacquemont  16 , Claire Duflos  17 , Yline Capri  7 , Jeanne Amiel  9 , Patricia Blanchet  1 , Stanislas Lyonnet  9 , Damien Sanlaville  18 , David Genevieve  19
Affiliations

Using deep-neural-network-driven facial recognition to identify distinct Kabuki syndrome 1 and 2 gestalt

Flavien Rouxel et al. Eur J Hum Genet. 2022 Jun.

Abstract

Kabuki syndrome (KS) is a rare genetic disorder caused by mutations in two major genes, KMT2D and KDM6A, that are responsible for Kabuki syndrome 1 (KS1, OMIM147920) and Kabuki syndrome 2 (KS2, OMIM300867), respectively. We lack a description of clinical signs to distinguish KS1 and KS2. We used facial morphology analysis to detect any facial morphological differences between the two KS types. We used a facial-recognition algorithm to explore any facial morphologic differences between the two types of KS. We compared several image series of KS1 and KS2 individuals, then compared images of those of Caucasian origin only (12 individuals for each gene) because this was the main ethnicity in this series. We also collected 32 images from the literature to amass a large series. We externally validated results obtained by the algorithm with evaluations by trained clinical geneticists using the same set of pictures. Use of the algorithm revealed a statistically significant difference between each group for our series of images, demonstrating a different facial morphotype between KS1 and KS2 individuals (mean area under the receiver operating characteristic curve = 0.85 [p = 0.027] between KS1 and KS2). The algorithm was better at discriminating between the two types of KS with images from our series than those from the literature (p = 0.0007). Clinical geneticists trained to distinguished KS1 and KS2 significantly recognised a unique facial morphotype, which validated algorithm findings (p = 1.6e-11). Our deep-neural-network-driven facial-recognition algorithm can reveal specific composite gestalt images for KS1 and KS2 individuals.

PubMed Disclaimer

Conflict of interest statement

DG was a consultant for the Takeda Society in 2018. Takeda did not have any role in this study.

Figures

Fig. 1
Fig. 1. On the left is the score distribution for KDM6A vs KMT2D, and on the right is the ROC curve obtained by using DeepGestalt analysis.
Binary comparison of facial images of individuals with KDM6A and KMT2D pathologic variants from the collaborative dataset (A) and the collaborative dataset with Caucasian origin only (B).
Fig. 2
Fig. 2. On the left is the score distribution for KDM6A vs KMT2D, and on the right is the ROC curve obtained by using DeepGestalt analysis.
Binary comparison of facial images of individuals with KDM6A and KMT2D pathologic variants from the mixed dataset of the full collaborative and literature datasets (A) and individuals of Caucasian origin only and literature datasets (B).
Fig. 3
Fig. 3. On the left is the composite gestalt based upon 17 KDM6A individual’s pictures form our collaborative dataset, and on the right is the composite gestalt based upon 17 KMT2D individual’s pictures from our collaborative dataset.
Composite gestalt images of individuals with KDM6A and KMT2D variants based on the collaborative dataset.
Fig. 4
Fig. 4. Distribution of scores for each subgroup of clinicians in differentiating between KS1 and KS2 individuals (n = 60).
Normal random distribution was plotted with 60 events: mean 16.5, SD 2.9. 1–2, 3–4, level of expertise.
See this image and copyright information in PMC

References

    1. Ng SB, Bigham AW, Buckingham KJ, Hannibal MC, McMillin MJ, Gildersleeve HI, et al. Exome sequencing identifies MLL2 mutations as a cause of Kabuki syndrome. Nat Genet. 2010;42:790–3. doi: 10.1038/ng.646. - DOI - PMC - PubMed
    1. Lederer D, Grisart B, Digilio MC, Benoit V, Crespin M, Ghariani SC, et al. Deletion of KDM6A, a Histone Demethylase Interacting with MLL2, in Three Patients with Kabuki Syndrome. Am J Hum Genet. 2012;90:119–24. doi: 10.1016/j.ajhg.2011.11.021. - DOI - PMC - PubMed
    1. Miyake N, Mizuno S, Okamoto N, Ohashi H, Shiina M, Ogata K, et al. KDM6A point mutations cause Kabuki syndrome. Hum Mutat. 2013;34:108–10. doi: 10.1002/humu.22229. - DOI - PubMed
    1. Bögershausen N, Gatinois V, Riehmer V, Kayserili H, Becker J, Thoenes M, et al. Mutation Update for Kabuki Syndrome Genes KMT2D and KDM6A and Further Delineation of X-Linked Kabuki Syndrome Subtype 2. Hum Mutat. 2016;37:847–64. doi: 10.1002/humu.23026. - DOI - PubMed
    1. Gilissen C, Hehir-Kwa JY, Thung DT, van de Vorst M, van Bon BWM, Willemsen MH, et al. Genome sequencing identifies major causes of severe intellectual disability. Nature. 2014;511:344–7. doi: 10.1038/nature13394. - DOI - PubMed

Publication types

Supplementary concepts