Subacute Infarct Volume With Edema Correction in Computed Tomography Is Equivalent to Final Infarct Volume After Ischemic Stroke

OBJECTIVES: Final infarct volume is regularly used as an end point of tissue outcome in stroke trials; however, the reported volumes are most commonly derived from early follow-up imaging. Those volumes are significantly impaired by ischemic edema, which causes an overestimation of the true final lesion volume. As net water uptake within ischemic brain can be quantified densitometrically in computed tomography (CT) as recently described, we hypothesized that the final lesion volume can be better estimated by correcting the lesion volume in early follow-up for the corresponding proportion of edema. MATERIALS AND METHODS: After retrospective consecutive screening of the local registry, 20 patients with acute middle cerebral artery large vessel occlusion met the inclusion criteria with early and late follow-up CT; the latter acquired at least 4 weeks after admission. In early follow-up imaging 24 hours after onset, the proportion of edema contributing to the infarct lesion was calculated by quantifying the total volume of ischemic net water uptake. Edema volume was then subtracted from the total lesion volume to obtain edema-corrected lesion volumes. Finally, these corrected lesion volumes were compared with the final lesion volume on late follow-up serving as ground truth. RESULTS: The median lesion volume in the early follow-up was 115.1 mL (range, 21.9-539.9 mL) and significantly exceeded the median final lesion volume in the late follow-up CT, which was 86.6 mL (range, 11.2-399.0 mL; p < 0.001). The calculated mean proportion of edema within the early lesion volume was 25.8% (±5.9%; range, 11.1%-35.9%. The median edema-corrected lesion volume measured after 24 hours was 87.1 mL (range, 18.2-376.3 mL). The estimation of final lesion volume in the early follow-up CT was therefore improved by a mean of 31.4% (±2.1%) when correcting for the proportion of edema and did not differ significantly from the true final infarct volume (p = 0.2). CONCLUSIONS: Edema-corrected volumes of early follow-up infarct lesion in CT were in close agreement with the actual final infarct volumes. Computed tomography-based edema correction of subacute infarct lesions improves the estimation of final tissue outcome. This could especially improve the comparability of imaging end points and facilitate patient recruitment in clinical trials.