Abstract WP30: Delay and Dispersion Correction Leads to More Accurate Core Estimation for Endovascular Thrombectomy Patients

Abstract

Objective: Endovascular thrombectomy (EVT) significantly improves clinical outcomes in acute ischemic stroke with large vessel occlusion. Clinical benefits are inversely proportional to size of the pre-treatment ischemic core. Therefore, accurate measurement of the size of core is critical in selecting patients for EVT. Different post-processing perfusion algorithms for automated core calculation on perfusion CT (CTP) are based on variations of deconvolution of the tissue concentration time curve with the arterial input function (single value decomposition, or SVD). In this study, we compared ischemic core estimated by two different CTP automated algorithms to the final infarct volume as demonstrated by follow up diffusion weighted imaging (DWI). Methods: We performed a retrospective analysis of patients who underwent EVT. Inclusion criteria were CT perfusion scan prior to EVT, successful EVT with mTICI 2b-3 reperfusion, and DWI scan 24-48 hours post-EVT. CTP data were processed by two different post-processing algorithms: ‘delay-insensitive’ single value decomposition (DISVD) and delay and dispersion corrected single value decomposition (ddSVD) using the respective commercially available automated CTP software. CTP core volumes from both methods were compared with DWI final infarct volumes using an independent software (MRIcron) for concordance. The agreement between a given algorithm and MRI was estimated using Lin’s concordance coefficient and further investigated using reduced major axis regression. Results: One hundred and three patients who underwent EVT and achieved successful mTICI 2b-3 reperfusion were included. Both algorithms had excellent agreement with MRI (Lin’s concordance coefficients: DISVD 0.8 (95% CI: 0.73; 0.87), ddSVD 0.92 (95% CI: 0.89; 0.95). Compared to ddSVD (reduced major axis slope = 0.95), DISVD exhibited larger extent of proportional bias (slope = 1.12). Conclusion: Both algorithms showed excellent agreement with FIV calculated on MRI but DISVD post-processing overestimated the larger ischemic cores, which may lead to unnecessary exclusion of patients from EVT due to a 'large core'.