CT perfusion analysis by nonlinear regression for predicting hemorrhagic transformation in ischemic stroke

Area of Science:

• Neuroimaging and Stroke Research
• Cerebrovascular Disease Diagnostics
• Medical Physics and Image Analysis

Background:

• Intravenous thrombolysis improves outcomes in acute ischemic stroke but increases the risk of hemorrhagic transformation (HT).
• Blood-brain barrier damage, indicated by vascular permeability, is a potential predictor of HT.
• Current methods like Patlak analysis may not optimally quantify this permeability for HT prediction.

Purpose of the Study:

• To evaluate if a novel fast nonlinear regression (NLR) method improves the prediction of hemorrhagic transformation (HT) in acute ischemic stroke patients.
• To compare the discriminative power of vascular permeability measurements using NLR versus standard Patlak analysis for predicting HT.

Main Methods:

• A prospective multicenter cohort study included 20 patients with HT and 40 without HT.
• Vascular permeability (K(trans)) and permeability-surface area product (PS) were measured using standard Patlak analysis and a novel NLR method.
• Relative values (rK(trans), rPS) were calculated and analyzed using Mann-Whitney U tests and ROC analyses to assess discriminative power.

Main Results:

• Relative K(trans) (rK(trans)) values measured with the NLR method were significantly higher in patients who developed HT.
• The relative permeability-surface area product (rPS) measured with NLR showed the highest discriminative power (AUC = 0.75) for predicting HT.
• Standard Patlak analysis for rK(trans) did not yield significantly different values between groups and had lower AUCs compared to NLR-derived rPS.

Conclusions:

• CT perfusion analysis, particularly using the NLR method, can aid in predicting HT after acute ischemic stroke.
• The NLR-derived rPS in the infarct core demonstrated superior discriminative power for HT prediction compared to K(trans) methods and conventional perfusion parameters.
• Standard Patlak analysis was less effective in identifying patients at risk for HT based on vascular permeability.