CTA-Derived Radiomics for Predicting Stroke Recurrence in Carotid Atherosclerosis Patients

A retrospective study of 162 patients with unilateral carotid atherosclerosis who underwent head and neck computed tomography angiography (CTA) between May 2016 and March 2024. The dataset, authored by Jianyong Wei and shared under CC-BY-4.0, contains radiomics features extracted from carotid plaque and perivascular adipose tissue to build a machine learning model for predicting ipsilateral stroke recurrence.

Use Cases

• Developing stroke recurrence risk models based on radiomics features of carotid plaque and perivascular adipose tissue.
• Comparing machine learning algorithm performance (e.g., support vector machines) for clinical prediction tasks.
• Investigating feature importance for stroke risk using SHAP analysis on combined radiomics and clinical factors.
• Validating predictive models through repeated cross-validation and independent testing cohorts.

Strengths

• Model performance is validated, with the combined model achieving an AUC of 0.87 in the independent testing set.
• Dataset includes 162 patient records with a median follow-up of 1 year and 63 observed stroke recurrence events.
• Interpretability is addressed using SHAP analysis to explain feature contributions.

Limitations

• Row count for the underlying feature data is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• Data may reflect temporal and selection bias inherent to the single-center retrospective study design.

Provenance

Source

figshare, author Jianyong Wei.

Collection Method

Retrospective study using CTA scans and follow-up MRI, with radiomics features extracted via a deep learning-based segmentation model.

Time Range

Patient CTA scans from May 2016 to March 2024.

Freshness

Last updated 2026-04 16 04:20:26; freshness should be verified.

Geography

null