Automatic acromegaly detection using deep learning on hand images: a multicenter observational study

Ohmachi, Y; Nishio, M; Abe, I; Kobayashi, K; Iida, T; Shirakawa, M; Nagata, Y; Takeuchi, K; Taguchi, A; Kinoshita, Y; Fukuhara, N; Nishioka, H; Tahara, S; Fujio, S; Ogura, T; Kurosaki, M; Hada, Y; Susa, S; Otsuka, Y; Otsuka, F; Ishida, I; Kameda, H; Oyama, K; Yamada, S; Kobatake, M; Oi-Yo, Y; Fujii, G; Tomofuji, S; Sasaki, Y; Bando, H; Yamamoto, M; Iguchi, G; Motomura, Y; Tsujimoto, Y; Yamamoto, N; Suzuki, M; Urai, S; Takahashi, M; Murakami, T; Ogawa, W; Fukuoka, H

doi:10.1210/clinem/dgag027

Pituitary & hypothalamus

Automatic acromegaly detection using deep learning on hand images: a multicenter observational study

J Clin Endocrinol MetabPublished 27 Feb 2026Diagnostic accuracy study

Read the original article ↗PubMed

Cite

Authors' conclusion

This study highlights the utility of dorsal hand and fist sign as diagnostic clues for acromegaly, which the AI model captured more accurately than endocrinologists.

Study context

Population: Adults aged ≥18 years with acromegaly (317) and controls (399), 15 Japanese pituitary centers
Participants: 716 adults11,480 images

Comparison: ResNet-50-based deep learning model on hand images vs endocrinologists
Outcome: Detection of acromegaly from hand images (test dataset)

Methodological detailsnationwide multicentre, observational, training/validation (12 centers) and test (3 centers) split, 5-fold cross-validation

Nationwide multicentre
Observational
Training/validation (12 centers) and test (3 centers) split
5-fold cross-validation

Performance

As reported in the abstract

Test or model	AUC	Sensitivity	Specificity	PPV	NPV	F1-score	F1-score range
Model	0.96	0.89	0.91	0.88	0.93	0.89	— not stated in the abstract
Specialists	— not stated in the abstract	— not stated in the abstract	— not stated in the abstract	— not stated in the abstract	— not stated in the abstract	— not stated in the abstract	0.43-0.63

AUC

Model0.96

Specialists—

Sensitivity

Model0.89

Specialists—

Specificity

Model0.91

Specialists—

PPV

Model0.88

Specialists—

NPV

Model0.93

Specialists—

F1-score

Model0.89

Specialists—

F1-score range

Model—

Specialists0.43-0.63

— not stated in the abstract

Reference

Authors: Ohmachi Y, Nishio M, Abe I, Kobayashi K, Iida T, Shirakawa M, et al.
All 41 authors
Ohmachi Y, Nishio M, Abe I, Kobayashi K, Iida T, Shirakawa M, Nagata Y, Takeuchi K, Taguchi A, Kinoshita Y, Fukuhara N, Nishioka H, Tahara S, Fujio S, Ogura T, Kurosaki M, Hada Y, Susa S, Otsuka Y, Otsuka F, Ishida I, Kameda H, Oyama K, Yamada S, Kobatake M, Oi-Yo Y, Fujii G, Tomofuji S, Sasaki Y, Bando H, Yamamoto M, Iguchi G, Motomura Y, Tsujimoto Y, Yamamoto N, Suzuki M, Urai S, Takahashi M, Murakami T, Ogawa W, Fukuoka H
Journal: The Journal of Clinical Endocrinology & Metabolism
Published: 27 Feb 2026
DOI: 10.1210/clinem/dgag027
PubMed: 41757900

Topic: Pituitary & hypothalamus

Diagnostic definitions

Model: ResNet-50-based deep learning model trained with PyTorch, data augmentation and 5-fold cross-validation

Key notes

Model outperformed specialists

Original publication

Automatic acromegaly detection using deep learning on hand images: a multicenter observational study. J Clin Endocrinol Metab. Published 27 February 2026. doi:10.1210/clinem/dgag027

Read the original article ↗

Composed by AI from the published abstract and checked against it. How Briefs are made →