Defining Ischemic Core in Acute Ischemic Stroke Using CT Perfusion: A Multiparametric Bayesian-Based Model.

Area of Science:

• Neuroimaging
• Radiology
• Medical Statistics

Background:

• Bayesian probabilistic methods show potential for reducing noise in perfusion analysis.
• Computed tomography perfusion (CTP) is used for analyzing perfusion in acute ischemic stroke.

Purpose of the Study:

• To determine optimal Bayesian-estimated thresholds for ischemic core estimation in acute ischemic stroke patients.
• To utilize multiparametric voxel-level models with CTP for improved accuracy.

Main Methods:

• Included patients with anterior circulation acute ischemic stroke and successful recanalization.
• Constructed multiparametric voxel-based models using Bayesian-processed CTP maps and follow-up MRIs.
• Compared Bayesian model CTP-estimated ischemic core volumes against routine methods and final infarct volume.

Main Results:

• Identified TTP, relative CBF, differential arterial tissue delay, and differential mean transit time as predictors of infarction.
• Achieved nearly 80% accuracy in identifying infarcted voxels with the Bayesian model.
• Demonstrated improved volumetric agreement for the Bayesian model (-25 to 27 mL) compared to routine methods (-61 to 52 mL).

Conclusions:

• Established thresholds for the Bayesian model to accurately estimate the ischemic core.
• The multiparametric Bayesian model enhances consistency in CTP-based ischemic core estimation.
• This method offers improved performance over current clinical routine methodologies.