Ultrasound Radiomics Model for Predicting Cervical Stromal Invasion in Endometrial Cancer

294 patient records with pathologically confirmed endometrial cancer were used to develop a non-invasive prediction model. The study extracted 837 radiomics features from 2D ultrasound images, reducing them to 25 for a radiomics score. The resulting model achieved an AUC of 0.906 in the test cohort, outperforming clinical predictors.

Use Cases

• Training machine learning models for cancer diagnosis based on ultrasound radiomics features.
• Comparing the predictive performance of radiomics models against clinical variable models.
• Developing tools for preoperative surgical planning in endometrial cancer cases.

Strengths

• Model performance is quantitatively validated with an AUC of 0.906 in a test cohort of 89 patients.
• The dataset includes 294 patient cases with pathologically confirmed diagnoses.
• Feature engineering is described in detail, starting from 837 extracted radiomics features.

Limitations

• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• The dataset is small in file size (280.4 KB), suggesting limited raw data scope.

Provenance

Source

Xiaoli Peng via figshare

Collection Method

Retrospective analysis of patient ultrasound images and clinical records.

Freshness

Last updated 2026-05-07 04:45:34; freshness should be verified.