Multiparametric Model for Penumbral Flow Prediction in Acute Stroke

Area of Science:

• Neuroimaging
• Stroke Research
• Medical Diagnostics

Background:

• Dynamic susceptibility contrast magnetic resonance imaging (DSC-MRI) is crucial for identifying salvageable penumbra tissue in acute stroke.
• Prior research has not integrated multiple perfusion maps into a single predictive model for stroke stratification.
• This study aimed to develop and validate a multiparametric perfusion imaging model.

Purpose of the Study:

• To establish a novel multiparametric perfusion imaging model using DSC-MRI for stroke patient stratification.
• To cross-validate the developed model against positron emission tomography (PET) perfusion for accurate detection of penumbral flow.
• To evaluate the predictive performance of the integrated model compared to individual perfusion parameters.

Main Methods:

• Retrospective analysis of 17 subacute stroke patients with DSC-MRI and H2O15 PET scans.
• Construction of perfusion maps (cerebral blood flow, cerebral blood volume, mean transit time, time-to-maximum, time-to-peak).
• Application of a generalized linear model (GLM) to combine perfusion maps and voxel-wise prediction of penumbral flow, validated using receiver-operating characteristic analysis.

Main Results:

• The GLM demonstrated a significantly improved model fit compared to single perfusion maps (P<1×e-5).
• The GLM achieved a high performance with an area under the curve (AUC) of 0.91 for penumbral flow prediction.
• The performance difference between the GLM and the best single parameter (time-to-maximum) was minimal (AUC difference = 0.04).

Conclusions:

• A DSC-MRI-based GLM is a superior model for predicting penumbral flow in stroke patients.
• This multiparametric model provides a straightforward and observer-independent tool for therapy stratification.
• The findings support the clinical utility of advanced imaging analysis for acute stroke management.