|| Checking for direct PDF access through Ovid
Background: Collateral blood flow plays a major role to sustain oxygenation to the ischemic tissue at risk of irreversible infarction. It is an important marker during diagnosis and treatment decision in acute stroke. Presence of reliable collateral flow to the region at risk has been linked to better outcome and may be used in the future to extend the treatment window for endovascular thrombectomy. Automatic and objective scoring of collaterals would greatly facilitate treatment decisions. However, such capability remains beyond current imaging methods. In this study, we introduce an imaging marker computed from routine magnetic resonance (MR) perfusion studies that directly reflects the trajectories and the strength of the blood flow within the major arteries of the brain.Methods: MR perfusion studies were analyzed retrospectively from patients treated for acute ischemic stroke. All patients were diagnosed with a large artery occlusion (ICA or proximal M1 MCA) as observed on single-phase CT angiography (CTA) acquired immediately prior to endovascular thrombectomy. Spatial-temporal gradient was automatically extracted for every voxel to quantify the direction and the magnitude of the flow. A tractography algorithm was then used to recover the trajectories of the flow and a color mapping was applied to visualize both direction and strength of the flow. Regions of interest were delineated bilaterally and each image was reviewed by a neurologist for the ASITN score immediately prior to endovascular intervention.Results: The data of 80 patients satisfied the inclusion criteria (33 female, average age 71) and was included in the study. Collateral grade prior to intervention included 2 ASITN grade 4, 26 grade 3, 23 grade 2, 6 grade 1 and 0 grade 0. oTICI2C reperfusion scores after thrombectomy included 2 TICI 3 (100%), 22 TICI 2C (90-99%), 25 TICI o2B (67-89%), 9 TICI m2B (50-66%), 19 TICI 2A (<50%) and 3 TICI 0/1. The spatio-temporal flow tracks were predictive of the ASITN when used as input to a machine learning model that mapped the tractography patterns of the collaterome to a continuous score.Conclusion: SFT, or Spatio-temporal Flow Tractography, is introduced to provide an automatic assessment of collateral flow from routine perfusion MRI.