Automated imaging analysis as a screening tool for acute recanalization treatment : from primary to comprehensive stroke center

“Time is brain” said Jeffrey L. Saver in 2006. This highlights the role of time from symptom onset in acute ischaemic stroke (AIS). The endovascular treatment (EVT) of large vessel occlusion (LVO) has been shown to be effective within 24 hours of symptom onset in AIS patients. This requires evaluation of the ischaemic core (ischaemic brain tissue destined for infarction) and penumbra (salvageable ischaemic brain tissue) with either computed tomography perfusion (CTP) or comprehensive magnetic resonance diffusion imaging (MRI) especially when more than 6 hours has elapsed from symptom onset. Little is known about individual variance of ischaemic core growth in individual AIS patients, although collateral flow has a major role in this growth. As CTP or MRI is seldomly available around the clock at primary stroke centres, comparability between different CTP automated imaging analysis software for core and penumbra assessment is still largely unknown, and knowledge about probable infarct growth rate during possible transfer for EVT is limited, readily available screening tools should be further developed for acute recanalization treatment candidates. In this dissertation we aimed to study 1) whether automated algorithm and Non-Contrast Computed Tomography (NCCT) based imaging biomarker, NCCTcore (Brainomix Ltd), is able to identify acute recanalization treatment candidates with a large follow-up infarct volume (≥ 50 mL) similarly to CTP RAPID software (CTPcore) (Study I), 2) the concept of fast infarct progression and the accuracy of approximated follow-up infarct volume (eFIV) using a linear growth model based on baseline CTP ischaemic core volume and time from symptom onset alone compared with actual follow up infarct after EVT (Study II) and, 3) the difference in volumes of ischaemic core volume and perfusion lesion and the agreement of target mismatch (ischaemic core < 70 mL, penumbra >15 mL) in acute recanalization treatment candidates screened for EVT between commercially available CTP automated imaging analysis software (AutoMIstar, OLEA and syngo.via) and RAPID software (Study III). The overall aim of all studies was to evaluate the possibilities of imaging biomarkers to help in clinical decision-making when considering transfer of acute stroke code patient for a EVT from primary to a comprehensive stroke center. This dissertation was based on the Helsinki Stroke Registry (HSQR) and includes all patients screened for acute recanalization treatment at Helsinki University Hospital (HUS) from September 2015 to September 2021. Study I included patients who had baseline volumes of NCCTcore, and CTPcore and follow-up NCCT imaging at 24 h (+/- 6 h) available. Study II included patients with known symptom onset, perfusion imaging (RAPID), LVO, and EVT procedure with hypoperfusion intensity ratio (HIR, RAPID) as a marker of indirect collateral flow. Study III included all patients with perfusion imaging (RAPID). Follow-up infarct volumes (mL) were measured by a neuroradiologist from the NCCT in Studies I and II. Study I showed that NCCTcore is a promising tool to identify patients not eligible for EVT because of large cores at baseline. In a cohort of 288 patients, NCCTcore identified 23 patients and CTPcore 26 patients with ischaemic core ≥70 mL. From the 23 patients with a large core in NCCT, only one had a good clinical outcome after EVT. In Study II, in a cohort of 48 patients, the linear estimation of follow-up infarct volume proved to be an oversimplification. In most patients (83%), ischaemic core growth was not linear. High HIR (≥0.4) predicted well both poor clinical outcome and large FIV after EVT. In Study III, in a cohort of 1606 patients with baseline RAPID perfusion maps, the three commonly used automated imaging software showed a small core difference, a larger difference in volume of perfusion lesion and variance in agreement rate of target mismatch compared with RAPID software. To conclude, NCCTcore appears to be beneficial at primary stroke center level with limited access to more advanced imaging. It can also help telemedicine consultations of EVT candidates as a decision-supporting tool. Ischaemic core progression, individual pace of infarct growth, and collateral status are important to evaluate in acute recanalization candidates when considering transfer to the nearest comprehensive stroke center for EVT. As there was variance in ischaemic core volume, volume of perfusion lesion and target mismatch between individual CTP automated imaging analysis software, this highlights the importance of human interpretation of the imaging data and its potential pitfalls when acute recanalization treatment is considered.