Super-Resolution Ultrasound Radiomics for Predicting Nondiagnostic Thyroid Nodules

437 patients with thyroid nodules were included in a study evaluating a GAN-based super-resolution radiomics pipeline for predicting nondiagnostic fine-needle aspiration outcomes. The dataset, shared by Shaozheng He, contains processed ultrasound images and radiomic features extracted using PyRadiomics. It was last updated in May 2026 and is shared under a CC-BY-4.0 license.

Use Cases

• Training machine learning classifiers to predict nondiagnostic cytology outcomes based on ultrasound radiomic features.
• Comparing the performance of super-resolution versus normal-resolution ultrasound image analysis for medical diagnosis.
• Conducting qualitative error analysis on false-positive and false-negative predictions in thyroid nodule assessment.

Strengths

• Includes data from 437 patients, split into development (n=338) and temporally independent validation (n=99) cohorts.
• Describes a multistep feature selection pipeline using Mann-Whitney U test, Spearman correlation, RFE, and LASSO regression.
• Reports model performance metrics, such as an AUC of 0.7435 for the SR-RF model on the independent validation set.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Data is from a single-center retrospective study, which may limit generalizability.

Provenance

Source

figshare, author Shaozheng He.

Collection Method

Retrospective study using 2D B-mode ultrasound images processed with a pretrained GAN-based super-resolution model.

Freshness

Last updated 2026-05-01 04:18:40; freshness should be verified.