Automated acquisition of thyroid cytology image features using self-supervised learning and application to image classification
Abstract
Background: Thyroid fine-needle aspiration cytology is widely used to evaluate thyroid nodules before surgery. Recent advances in deep learning have increased the potential for artificial intelligence-assisted cytological diagnosis. However, developing accurate models usually requires large numbers of images labelled by experts, making dataset preparation time-consuming and costly. We investigate whether self-supervised learning can improve thyroid cytology image classification while reducing the need for labelled data.
Methods: We retrospectively analysed cytology images from 393 thyroid nodules and extracted 142,158 image patches. Three self-supervised learning methods were used to pre-train vision transformer models on unlabelled images, followed by classification training with different amounts of labelled data. Performance in classifying eight lesion types was assessed using five-fold cross-validation, precision–recall area under the curve, and macro-averaged F1 scores. Model performance was compared with ImageNet-pretrained models and four cytotechnologists.
Results: Here we show that self-supervised learning maintains high classification performance when labelled training data are limited and consistently outperforms ImageNet pre-training. With 1% of the labelled data, the best self-supervised model achieves a precision–recall area under the curve of 0.899, compared with 0.707 for ImageNet pre-training. The diagnostic pipeline achieves an average class-specific accuracy of 0.94, exceeding that of the four participating cytotechnologists. Self-supervised image representations also form lesion-specific clusters without diagnostic labels.
Conclusions: Self-supervised learning enables accurate classification of thyroid cytology images while substantially reducing the need for expert annotation. This approach may facilitate the development of more efficient artificial intelligence-assisted cytology systems, although validation using data from multiple institutions is needed.
Related
- CERVEX: A Foundation-Model Framework With Built-In Explainable AI for Automated Cervical Cytology Classification From Pap Smear Images
- Weakly-Supervised Gated Attention Multiple Instance Learning for Multi-Center Cervical Cytology Slide Triage
- Developing an automated cytology-based DNA ploidy test for detecting oral squamous cell carcinoma in Taiwan: A feasibility study.
- A novel morphological assessment approach for abdominal aortic aneurysms: automated segmentation and measurement using a semi-supervised deep learning algorithm
- A user-friendly, no-code, application for HIPAA-compliant automated analysis of tabular data at scale
BibTeX
@article{abe2026automated,
title = {{Automated acquisition of thyroid cytology image features using self-supervised learning and application to image classification}},
author = {Masatoshi Abe and Ayana Suzuki and Kumiko Kamada and Toshiya Kanno and Mayu Shimasaki and Toshinori Mori and Hajime Nagahara and Akira Miyauchi and Takashi Akamizu and Mitsuyoshi Hirokawa and Hirohiko Niioka},
journal = {Communications Medicine},
year = {2026},
month = oct,
doi = {10.1038/s43856-026-01874-2},
url = {https://www.nature.com/articles/s43856-026-01874-2},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Automated acquisition of thyroid cytology image features using self-supervised learning and application to image classification
AU - Masatoshi Abe
AU - Ayana Suzuki
AU - Kumiko Kamada
AU - Toshiya Kanno
AU - Mayu Shimasaki
AU - Toshinori Mori
AU - Hajime Nagahara
AU - Akira Miyauchi
AU - Takashi Akamizu
AU - Mitsuyoshi Hirokawa
AU - Hirohiko Niioka
PY - 2026
DA - 2026/10/06
JO - Communications Medicine
DO - 10.1038/s43856-026-01874-2
UR - https://www.nature.com/articles/s43856-026-01874-2
AB - Background: Thyroid fine-needle aspiration cytology is widely used to evaluate thyroid nodules before surgery. Recent advances in deep learning have increased the potential for artificial intelligence-assisted cytological diagnosis. However, developing accurate models usually requires large numbers of images labelled by experts, making dataset preparation time-consuming and costly. We investigate whether self-supervised learning can improve thyroid cytology image classification while reducing the need for labelled data. Methods: We retrospectively analysed cytology images from 393 thyroid nodules and extracted 142,158 image patches. Three self-supervised learning methods were used to pre-train vision transformer models on unlabelled images, followed by classification training with different amounts of labelled data. Performance in classifying eight lesion types was assessed using five-fold cross-validation, precision–recall area under the curve, and macro-averaged F1 scores. Model performance was compared with ImageNet-pretrained models and four cytotechnologists. Results: Here we show that self-supervised learning maintains high classification performance when labelled training data are limited and consistently outperforms ImageNet pre-training. With 1% of the labelled data, the best self-supervised model achieves a precision–recall area under the curve of 0.899, compared with 0.707 for ImageNet pre-training. The diagnostic pipeline achieves an average class-specific accuracy of 0.94, exceeding that of the four participating cytotechnologists. Self-supervised image representations also form lesion-specific clusters without diagnostic labels. Conclusions: Self-supervised learning enables accurate classification of thyroid cytology images while substantially reducing the need for expert annotation. This approach may facilitate the development of more efficient artificial intelligence-assisted cytology systems, although validation using data from multiple institutions is needed.
SN - 2730-664X
ER -