Improved visualisation of early cerebral infarctions after endovascular stroke therapy using dual-energy computed tomography oedema maps.

Grams, Astrid Ellen; Djurdjevic, Tanja; Rehwald, Rafael; Schiestl, Thomas; Dazinger, Florian; Steiger, Ruth; Knoflach, Michael; Gizewski, Elke Ruth; Glodny, Bernhard

Objective: The aim was to investigate whether dual-energy computed tomography (DECT) reconstructions optimised for oedema visualisation (oedema map; EM) facilitate an improved detection of early infarctions after endovascular stroke therapy (EST).Methods: Forty-six patients (21 women; 25 men; mean age: 63 years; range 24-89 years) were included. The brain window (BW), virtual non-contrast (VNC) and modified VNC series based on a three-material decomposition technique optimised for oedema visualisation (EM) were evaluated. Follow-up imaging was used as the standard for comparison. Contralateral side to infarction differences in density (CIDs) were determined. Infarction detectability was assessed by two blinded readers, as well as image noise and contrast using Likert scales. ROC analyses were performed and the respective Youden indices calculated for cut-off analysis.Results: The highest CIDs were found in the EM series (73.3 ± 49.3 HU), compared with the BW (-1.72 ± 13.29 HU) and the VNC (8.30 ± 4.74 HU) series. The EM was found to have the highest infarction detection rates (area under the curve: 0.97 vs. 0.54 and 0.90, p 0.05 each).Conclusions: The EM series allows higher contrast and better early infarction detection than the VNC or BW series after EST.Key Points: • Dual-energy CT EM allows better early infarction detection than standard brain window. • Dual-energy CT EM series allow better early infarction detection than VNC series. • Dual-energy CT EM are modified VNC based on water content of tissue.

DUAL energy CT (Tomography); CEREBRAL infarction; RECEIVER operating characteristic curves; DIAGNOSIS of edema; COMPUTED tomography; DIAGNOSIS

European Radiology, 2018, Vol 28, Issue 11, p4534

0938-7994

Academic Journal

10.1007/s00330-018-5449-4