liu.seSearch for publications in DiVA
Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • oxford
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Detection of Acromegaly From Facial Images Using Machine Learning: A Comparison With Clinical Experts
Umeå Univ, Sweden.
Linköping University, Department of Science and Technology, Media and Information Technology. Linköping University, Faculty of Science & Engineering.
Umeå Univ, Sweden.
Karolinska Univ Hosp, Sweden; Karolinska Inst, Sweden.
Show others and affiliations
2026 (English)In: Journal of the Endocrine Society, E-ISSN 2472-1972, Vol. 10, no 2, article id bvaf203Article in journal (Refereed) Published
Abstract [en]

Context Substantial diagnostic delay in acromegaly contributes to increased morbidity and mortality. Screening attempts in high-risk groups have yielded few positive cases, underscoring the need for simple and precise prescreening methods.Objective Machine-learning analysis of facial images shows promise for acromegaly detection but requires validation in larger, well-characterized cohorts using robust machine-learning frameworks as performed in this study.Methods Facial images from different angles were collected via smartphone from 155 acromegaly patients (79% biochemically controlled) and 153 matched controls at all Swedish university hospitals. Six machine-learning models were trained to distinguish acromegaly from control images, including 3 deep neural networks pretrained on diverse image datasets (ImageNet models: ResNet50, InceptionV2, and DenseNet121) and 1 network pretrained specifically on facial images (FaRL). Model performance was compared to assessment by 12 experienced endocrinologists.Results The diagnostic accuracy of the FaRL-based model was superior to all ImageNet models and matched the accuracy of human experts (area under the receiver operating characteristic curve 0.89 for both) with similar specificity (0.87 vs 0.93) but higher sensitivity (0.82 vs 0.66). Classification agreement between the best machine-learning model (FaRL) and human experts was 86% for true negatives and 60% for true positives. Machine-learning models and human experts both showed greater sensitivity in identifying acromegaly in male patients.Conclusion A deep learning model pretrained on facial features (FaRL) can detect acromegaly from standard photographs with accuracy comparable to that of expert endocrinologists. This supports the feasibility of face analysis as a screening tool for acromegaly.

Place, publisher, year, edition, pages
ENDOCRINE SOC , 2026. Vol. 10, no 2, article id bvaf203
Keywords [en]
acromegaly; diagnostic delay; screening; face classification; face photographs; machine learning; deep learning
National Category
Computer graphics and computer vision
Identifiers
URN: urn:nbn:se:liu:diva-220926DOI: 10.1210/jendso/bvaf203ISI: 001670333700001PubMedID: 41608201OAI: oai:DiVA.org:liu-220926DiVA, id: diva2:2034584
Note

Funding Agencies|Ume University; Arnerska Fund; JC Kempe Foundation

Available from: 2026-02-02 Created: 2026-02-02 Last updated: 2026-03-29

Open Access in DiVA

fulltext(784 kB)29 downloads
File information
File name FULLTEXT01.pdfFile size 784 kBChecksum SHA-512
11d82cb0805abc3a82f2fe6f99b26980a3bc931e162bcc6a430a4f853a99ce632a6110574dfd3467d872546ae969768db0d8de34b8eb222bf00dc5bede7f5e05
Type fulltextMimetype application/pdf

Other links

Publisher's full textPubMed

Search in DiVA

By author/editor
Ylipää, ErikEkman, BertilRobért, Jonas
By organisation
Media and Information TechnologyFaculty of Science & EngineeringDivision of Diagnostics and Specialist MedicineFaculty of Medicine and Health SciencesDepartment of Internal Medicine in Norrköping
In the same journal
Journal of the Endocrine Society
Computer graphics and computer vision

Search outside of DiVA

GoogleGoogle Scholar
Total: 29 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

doi
pubmed
urn-nbn

Altmetric score

doi
pubmed
urn-nbn
Total: 31 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • oxford
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf