Detection and localization of posterior circulation large vessel occlusions using perfusion imaging

Area of Science:

• Neurology
• Radiology
• Medical Imaging

Background:

• Cerebral perfusion imaging is valuable for anterior circulation stroke, but its utility in posterior circulation large vessel occlusion (PC-LVO) requires further definition.
• Accurate localization of PC-LVO is crucial for timely and effective treatment.

Purpose of the Study:

• To evaluate the diagnostic accuracy and inter-reader reliability of CT and MR perfusion imaging.
• To assess the ability of perfusion imaging to classify occlusions and localize vessels within the posterior circulation.

Main Methods:

• A retrospective multicenter diagnostic accuracy study included 184 patients with stroke, enriched for posterior circulation stroke.
• Four blinded readers classified cases into five categories: no occlusion, AC-LVO, basilar, PCA, or vert/PICA occlusion.
• Classification accuracy was assessed for 3-class (LVO(-), AC-LVO, PC-LVO) and 5-class tasks, with posterior circulation localization accuracy also analyzed.

Main Results:

• Perfusion imaging achieved 88.6% accuracy in classifying LVO(-), AC-LVO, and PC-LVO, with 100% sensitivity for LVO(-) and AC-LVO.
• In the posterior circulation subgroup, localization accuracy was 74.8%, with specific accuracies for basilar (76.6%), PCA (83.3%), and vert/PICA (56.3%) occlusions.
• Inter-reader reliability was high for the 3-class task (κ=0.889) and moderate for posterior localization (κ=0.604).

Conclusions:

• Perfusion imaging demonstrates moderate accuracy for detecting and localizing PC-LVO.
• Its performance in PC-LVO detection is less robust compared to AC-LVO detection.
• CT and MR perfusion imaging offer valuable, albeit imperfect, insights into posterior circulation stroke etiology.