Multimodal CT Radiomics and Clinical Data for Predicting Stroke Treatment Outcomes

A dataset of 101 patients diagnosed with acute ischemic stroke who underwent successful endovascular treatment. It includes 946 radiomics features extracted from non-contrast CT, contrast-enhanced CT, and CT perfusion maps, alongside demographic, clinical, laboratory, and angiographic variables. The data was used to develop a machine learning model predicting futile recanalization, achieving an AUC of 0.918.

Use Cases

• Predicting futile recanalization after endovascular treatment based on CT radiomics and clinical features.
• Training ensemble machine learning models using multimodal medical data.
• Analyzing the relationship between CT perfusion map features and stroke treatment outcomes.
• Benchmarking feature selection methods like SelectKBest on clinical-radiomics datasets.

Strengths

• Includes data from 101 patients, with 66 instances of the target outcome (futile recanalization).
• Extracts 946 radiomics features from multiple CT imaging modalities per patient.
• Model performance reported with an AUC of 0.918, indicating strong predictive capability.

Limitations

• Row count for the underlying feature table is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• The dataset's 2.4 MB size suggests it may contain summary or model results rather than the full raw feature matrix.

Provenance

Source

Zhenxiong Wang via figshare.

Collection Method

Data collected from a study enrolling 101 acute ischemic stroke patients who underwent endovascular treatment. Radiomics features were extracted using PyRadiomics.

Freshness

Last updated 2026-05-12 05:42:44; freshness should be verified.