Accuracy and Variability of Spatial Localization of Infarct Core Predicted by CT Perfusion

Area of Science:

• Neurology
• Radiology
• Medical Imaging

Background:

• CT perfusion (CTP) is crucial for acute ischemic stroke management, aiding prognostication and treatment decisions.
• Accurate infarct core estimation is vital for patient triage and therapeutic studies.
• Diffusion-weighted MRI (DWI) is the established reference standard for infarct core assessment.

Purpose of the Study:

• To compare the accuracy and variability of infarct core volumes derived from two FDA-cleared CTP software programs against DWI.
• To evaluate the performance of CTP in different infarct size scenarios.

Main Methods:

• Analysis of 61 acute ischemic stroke patients with concurrent CTP and DWI scans within 90 minutes.
• Infarct core estimation using relative cerebral blood flow thresholds in CTP.
• Comparison of CTP-derived infarct core with DWI ground truth, calculating Dice similarity coefficients for topographic agreement.

Main Results:

• CTP-derived infarct core volumes showed significant correlation but substantial variability and bias compared to DWI.
• Limited topographic overlap was observed between CTP and DWI, with median Dice scores of 0.367 (CTP-A) and 0.289 (CTP-B).
• In large infarcts (≥50 mL), CTP provided more reliable spatial agreement, with median Dice scores of 0.61 (CTP-A) and 0.47 (CTP-B).

Conclusions:

• CTP generally demonstrates low accuracy for infarct core estimation in acute ischemic stroke.
• CTP may offer clinically relevant insights for decision-making in cases with large infarcts.
• Findings can inform the design of future clinical trials investigating CTP in stroke.