Radiomics-based functional outcome prediction in acute single subcortical infarction

BackgroundClinical outcomes vary considerably among patients with acute single subcortical infarction (SSI). We aimed to construct a model incorporating radiomic features and clinical factors to predict functional outcomes in patients with acute SSI. MethodsWe enrolled patients who experienced acute SSI within 14 days of stroke onset and randomly divided them into training (n=118) and test (n=30) cohorts. Unfavorable functional outcome was defined as a modified Rankin Scale score >1 at 3 months. We extracted and selected radiomics features from baseline diffusion-weighted imaging and perfusion-weighted imaging to develop a radiomics model. Multivariate logistic regression was performed to construct a clinical model using clinical factors and imaging features. Finally, a combined model was built using both clinical and radiomics features. Receiver operating characteristic curves were used to evaluate the discriminatory ability of these models. ResultsThe radiomics model, encompassing 13 radiomics features, exhibited good predictive performance for unfavorable functional outcomes with area under the curve (AUC) values of 0.774 and 0.824 in the training and test cohorts, respectively. The combined model, which included clinical factors (early neurological deterioration, hypertension, baseline National Institutes of Health Stroke Scale score, infarct volume, and summary cerebral small vessel disease score) and radiomics features, improved performance in the training (AUC = 0.915) and test (AUC = 0.846) cohorts. ConclusionsThe clinical-radiomics model provided improved accuracy for the prognostic prediction of SSI, which may help clinicians in the decision-making process and improve long-term outcomes in patients with SSI.