To explore the value of Delta radiomics based on cerebral CT perfusion (CTP) in predicting hemorrhagic transformation after intravenous thrombolysis in acute cerebral infarction (HT-ACI).

A retrospective study was conducted on 419 patients with acute cerebral infarction who underwent CTP and received intravenous thrombolysis at the Neurology Department of Xiangyang No. 1 People's Hospital affiliated to Hubei University of Medicine from October 2016 to October 2024. Based on post-thrombolysis cranial CT or MRI findings, patients were categorized into an HT-ACI group (hemorrhagic transformation after acute cerebral infarction, n=114) and a non-HT-ACI group (n=305). Data were chronologically partitioned into a training set (HT-ACI: 80 cases, non-HT-ACI: 214 cases) and a test set (HT-ACI: 34 cases, non-HT-ACI: 91 cases) at a 7∶3 ratio. Within the training set, the infarct region of interest was delineated using 3D Slicer to extract the Delta radiomics score (Delta Radscore). Concurrently, CTP hemodynamic parameters [cerebral blood volume (CBV), cerebral blood flow (CBF), time to peak (TTP)] and clinical data [including age, sex, National Institutes of Health stroke scale (NIHSS) score, and medical history] were collected. Multivariate Logistic regression analysis was employed to develop predictive models of HT-ACI: a clinical model (based on clinical factors: age, NIHSS, etc.), an imaging model (based on imaging features: CBF, Delta Radscore, etc.), and a combined model (integrating both clinical and imaging data). Model performance was compared using DeLong's test for the area under the curve (AUC), clinical net benefit was assessed via decision curve analysis (DCA), and prediction outcomes were validated using the XGBoost algorithm.

In the training set, there were statistically significant differences in NIHSS score, anticoagulant use, age, infarct area, apparent diffusion coefficient difference, CBF, and Delta Radscore between the two groups (P<0.05). The combined model (incorporating clinical, radiological, and Delta radiomic features) demonstrated superior predictive performance (AUC=0.878) compared to the clinical model (AUC=0.725) and the imaging model (AUC=0.818). XGBoost further validated that infarct area, age, CBF, and Delta Radscore were associated with HT-ACI. The decision curve analysis indicated that the combined model had a higher clinical net benefit. The results verified in the test set were consistent with those in the training set, and a nomogram was constructed to simplify the prediction process.

The combined model based on CTP-derived Delta radiomics can early reflect the hemodynamic status of ischemic brain tissue and holds significant clinical value for predicting HT-ACI, aiding in guiding treatment decisions, reducing the risk of hemorrhage, and improving patient prognosis.