PEDIA

prioritization of exome data by image analysis

Ivanovski, Ivan

Research output: Contribution to journalArticle

2 Citations (Scopus)

Abstract

Purpose: Phenotype information is crucial for the interpretation of genomic variants. So far it has only been accessible for bioinformatics workflows after encoding into clinical terms by expert dysmorphologists. Methods: Here, we introduce an approach driven by artificial intelligence that uses portrait photographs for the interpretation of clinical exome data. We measured the value added by computer-assisted image analysis to the diagnostic yield on a cohort consisting of 679 individuals with 105 different monogenic disorders. For each case in the cohort we compiled frontal photos, clinical features, and the disease-causing variants, and simulated multiple exomes of different ethnic backgrounds. Results: The additional use of similarity scores from computer-assisted analysis of frontal photos improved the top 1 accuracy rate by more than 20–89% and the top 10 accuracy rate by more than 5–99% for the disease-causing gene. Conclusion: Image analysis by deep-learning algorithms can be used to quantify the phenotypic similarity (PP4 criterion of the American College of Medical Genetics and Genomics guidelines) and to advance the performance of bioinformatics pipelines for exome analysis.

Original languageEnglish
JournalGenetics in Medicine
DOIs
Publication statusPublished - Jan 1 2019
Externally publishedYes

Fingerprint

Exome
Computational Biology
Computer-Assisted Image Processing
Workflow
Artificial Intelligence
Genomics
Learning
Guidelines
Phenotype
Genes

Keywords

  • computer vision
  • deep learning
  • dysmorphology
  • exome diagnostics
  • variant prioritization

ASJC Scopus subject areas

  • Genetics(clinical)

Cite this

PEDIA : prioritization of exome data by image analysis. / Ivanovski, Ivan.

In: Genetics in Medicine, 01.01.2019.

Research output: Contribution to journalArticle

@article{d8603105a29943d5b0ac76b11d2e8cb5,
title = "PEDIA: prioritization of exome data by image analysis",
abstract = "Purpose: Phenotype information is crucial for the interpretation of genomic variants. So far it has only been accessible for bioinformatics workflows after encoding into clinical terms by expert dysmorphologists. Methods: Here, we introduce an approach driven by artificial intelligence that uses portrait photographs for the interpretation of clinical exome data. We measured the value added by computer-assisted image analysis to the diagnostic yield on a cohort consisting of 679 individuals with 105 different monogenic disorders. For each case in the cohort we compiled frontal photos, clinical features, and the disease-causing variants, and simulated multiple exomes of different ethnic backgrounds. Results: The additional use of similarity scores from computer-assisted analysis of frontal photos improved the top 1 accuracy rate by more than 20–89{\%} and the top 10 accuracy rate by more than 5–99{\%} for the disease-causing gene. Conclusion: Image analysis by deep-learning algorithms can be used to quantify the phenotypic similarity (PP4 criterion of the American College of Medical Genetics and Genomics guidelines) and to advance the performance of bioinformatics pipelines for exome analysis.",
keywords = "computer vision, deep learning, dysmorphology, exome diagnostics, variant prioritization",
author = "{Ivanovski, Ivan} and Hsieh, {Tzung Chien} and Mensah, {Martin A.} and Pantel, {Jean T.} and Dione Aguilar and Omri Bar and Allan Bayat and Luis Becerra-Solano and Bentzen, {Heidi B.} and Saskia Biskup and Oleg Borisov and Oivind Braaten and Claudia Ciaccio and Marie Coutelier and Kirsten Cremer and Magdalena Danyel and Svenja Daschkey and Eden, {Hilda David} and Koenraad Devriendt and Sandra Wilson and Sofia Douzgou and Dejan Đukić and Nadja Ehmke and Christine Fauth and Bj{\"o}rn Fischer-Zirnsak and Nicole Fleischer and Heinz Gabriel and Luitgard Graul-Neumann and Gripp, {Karen W.} and Yaron Gurovich and Asya Gusina and Nechama Haddad and Nurulhuda Hajjir and Yair Hanani and Jakob Hertzberg and Konstanze Hoertnagel and Janelle Howell and Ivan Ivanovski and Angela Kaindl and Tom Kamphans and Susanne Kamphausen and Catherine Karimov and Hadil Kathom and Anna Keryan and Alexej Knaus and Sebastian K{\"o}hler and Uwe Kornak and Alexander Lavrov and Maximilian Leitheiser and Lyon, {Gholson J.} and Elisabeth Mangold",
year = "2019",
month = "1",
day = "1",
doi = "10.1038/s41436-019-0566-2",
language = "English",
journal = "Genetics in Medicine",
issn = "1098-3600",
publisher = "Nature Publishing Group",

}

TY - JOUR

T1 - PEDIA

T2 - prioritization of exome data by image analysis

AU - Ivanovski, Ivan

AU - Hsieh, Tzung Chien

AU - Mensah, Martin A.

AU - Pantel, Jean T.

AU - Aguilar, Dione

AU - Bar, Omri

AU - Bayat, Allan

AU - Becerra-Solano, Luis

AU - Bentzen, Heidi B.

AU - Biskup, Saskia

AU - Borisov, Oleg

AU - Braaten, Oivind

AU - Ciaccio, Claudia

AU - Coutelier, Marie

AU - Cremer, Kirsten

AU - Danyel, Magdalena

AU - Daschkey, Svenja

AU - Eden, Hilda David

AU - Devriendt, Koenraad

AU - Wilson, Sandra

AU - Douzgou, Sofia

AU - Đukić, Dejan

AU - Ehmke, Nadja

AU - Fauth, Christine

AU - Fischer-Zirnsak, Björn

AU - Fleischer, Nicole

AU - Gabriel, Heinz

AU - Graul-Neumann, Luitgard

AU - Gripp, Karen W.

AU - Gurovich, Yaron

AU - Gusina, Asya

AU - Haddad, Nechama

AU - Hajjir, Nurulhuda

AU - Hanani, Yair

AU - Hertzberg, Jakob

AU - Hoertnagel, Konstanze

AU - Howell, Janelle

AU - Ivanovski, Ivan

AU - Kaindl, Angela

AU - Kamphans, Tom

AU - Kamphausen, Susanne

AU - Karimov, Catherine

AU - Kathom, Hadil

AU - Keryan, Anna

AU - Knaus, Alexej

AU - Köhler, Sebastian

AU - Kornak, Uwe

AU - Lavrov, Alexander

AU - Leitheiser, Maximilian

AU - Lyon, Gholson J.

AU - Mangold, Elisabeth

PY - 2019/1/1

Y1 - 2019/1/1

N2 - Purpose: Phenotype information is crucial for the interpretation of genomic variants. So far it has only been accessible for bioinformatics workflows after encoding into clinical terms by expert dysmorphologists. Methods: Here, we introduce an approach driven by artificial intelligence that uses portrait photographs for the interpretation of clinical exome data. We measured the value added by computer-assisted image analysis to the diagnostic yield on a cohort consisting of 679 individuals with 105 different monogenic disorders. For each case in the cohort we compiled frontal photos, clinical features, and the disease-causing variants, and simulated multiple exomes of different ethnic backgrounds. Results: The additional use of similarity scores from computer-assisted analysis of frontal photos improved the top 1 accuracy rate by more than 20–89% and the top 10 accuracy rate by more than 5–99% for the disease-causing gene. Conclusion: Image analysis by deep-learning algorithms can be used to quantify the phenotypic similarity (PP4 criterion of the American College of Medical Genetics and Genomics guidelines) and to advance the performance of bioinformatics pipelines for exome analysis.

AB - Purpose: Phenotype information is crucial for the interpretation of genomic variants. So far it has only been accessible for bioinformatics workflows after encoding into clinical terms by expert dysmorphologists. Methods: Here, we introduce an approach driven by artificial intelligence that uses portrait photographs for the interpretation of clinical exome data. We measured the value added by computer-assisted image analysis to the diagnostic yield on a cohort consisting of 679 individuals with 105 different monogenic disorders. For each case in the cohort we compiled frontal photos, clinical features, and the disease-causing variants, and simulated multiple exomes of different ethnic backgrounds. Results: The additional use of similarity scores from computer-assisted analysis of frontal photos improved the top 1 accuracy rate by more than 20–89% and the top 10 accuracy rate by more than 5–99% for the disease-causing gene. Conclusion: Image analysis by deep-learning algorithms can be used to quantify the phenotypic similarity (PP4 criterion of the American College of Medical Genetics and Genomics guidelines) and to advance the performance of bioinformatics pipelines for exome analysis.

KW - computer vision

KW - deep learning

KW - dysmorphology

KW - exome diagnostics

KW - variant prioritization

UR - http://www.scopus.com/inward/record.url?scp=85066983034&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85066983034&partnerID=8YFLogxK

U2 - 10.1038/s41436-019-0566-2

DO - 10.1038/s41436-019-0566-2

M3 - Article

JO - Genetics in Medicine

JF - Genetics in Medicine

SN - 1098-3600

ER -