Temporal Similarity Mapping

Introduction: Interpretation of the extent of perfusion deficits in stroke MRI is highly dependent on the method used for analyzing the perfusion-weighted signal intensity time-series after gadolinium injection. In this study, we introduce a new model-free standardized method of temporal similarity perfusion (TSP) mapping for perfusion deficit detection and test its ability and reliability in acute stroke. Materials and Methods: Forty patients with an ischemic stroke or transient ischemic attacks were included. Two blinded readers compared real-time generated interactive maps (described below) and automatically generated TSP maps to traditional TTP/MTT maps for presence of perfusion deficits. Lesion volumes were compared for volumetric inter-rater reliability, spatial concordance between perfusion deficits and healthy tissue and contrast-to-noise ratio. Results: Perfusion deficits were correctly detected in all patients with acute ischemia. Inter-rater reliability was higher for TSP when compared to TTP/MTT maps and there was a high similarity between the lesion volumes depicted on TSP and TTP/MTT (r(18)=0.73). The Pearson's correlation between lesions calculated on TSP and traditional maps was high (r(18)=0.73, p<0.0003), however the effective CNR was greater for TSP compared to TTP (352.3 vs 283.5, t(19)=2.6, p<0.03.) and MTT (228.3, t(19)=2.8, p<0.03). Discussion: TSP maps provide a reliable and robust model-free method for accurate perfusion deficit identification and improve lesion delineation compared to traditional methods. This simple method is also computationally faster and more easily automated than model-based methods. This method can potentially improve the speed and accuracy in perfusion lesion identification for acute stroke treatment decision-making and clinical trial inclusion.