ThyroidXL: Expert-Labeled and Pathology-Validated Thyroid Nodule Data

ThyroidXL is an expert-labeled and pathology-validated dataset for thyroid nodule diagnosis published in 2025 by Viet Hung Duong and a multi-institutional research team. The data provides ultrasound-based diagnostic information where findings are confirmed against gold-standard pathology results. It was developed to improve the accuracy of computer-aided diagnosis in endocrinology.

Use Cases

• Training binary classification models to distinguish benign from malignant nodules using pathology-validated labels
• Benchmarking computer-aided diagnosis (CAD) systems against expert-labeled ultrasound findings
• Statistical analysis of the correlation between ultrasound features and biopsy-confirmed pathology outcomes

Strengths

• Pathology-validated ground truth labels
• Expert-labeled by a multi-author medical research team
• Peer-reviewed publication context (2025)

Limitations

• Lack of explicit row counts or column headers in the metadata
• Narrow domain focus restricted to thyroid pathology

Provenance

Source

Viet Hung Duong et al., 'ThyroidXL: Advancing Thyroid Nodule Diagnosis with an Expert-Labeled, Pathology-Validated Dataset' (2025)

Collection Method

Expert annotation and clinical pathology validation

Freshness

Last updated February 2026; reflects 2025 research standards.