You need to agree to share your contact information to access this dataset

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

YAML Metadata Warning: empty or missing yaml metadata in repo card (https://huggingface.co/docs/hub/datasets-cards)

🧾 If you use this dataset, please cite the following paper: "ThyroidXL: Advancing Thyroid Nodule Diagnosis with an Expert-Labeled, Pathology-Validated Dataset"

ThyroidXL Dataset

The ThyroidXL dataset. The annotations are provided for classification, detection and segmentation tasks.

Dataset stats

Total number of patients: 4093 (patients)
Total number of images: 11545 (images)

Train set:

Number of patients: 3275 (2477 benigns, 877 malignants)
Number of images: 9541
Benign-Malignant Ratio: 64.6% - 35.4%

Test set:

Number of patients: 739 (386 benigns, 453 malignants)
Number of images: 2094
Benign-Malignant Ratio: 52.2% - 47.8%

Category mapping

Gender:

Male: 1
Female: 2

Thyroid diagnostic results:

Benign: 0
Malignant: 1

Male-Female ratios in the trainset and testset:

Age distributions:

TIRADS distributions by Patients

TIRADS distributions by Images

Samples

Downloads last month: 221