Discrimination of thrombus types in ischemic stroke using Dual-Energy CT

This study explores the clinical utility of dual-energy computed tomography (DECT) in discriminating thrombus types in ischemic stroke.

Patients with acute ischemic stroke who underwent brain DECT non-contrast scanning and brain CT perfusion (CTP) before thrombectomy were included, and the thrombus composition was analyzed by postoperative pathology. DECT data was conducted to reconstruct polychromatic images and effective atomic number images. Computed tomography (CT) values, effective atomic numbers, and spectral curve slopes of the thrombus were measured and calculated. Thrombus attenuation increase was obtained from CTP data. Parameters were compared between red blood cell (RBC)-dominant thrombi and fibrin/platelet (F/P)-dominant thrombi. Thresholds, sensitivity, specificity, and area under the curve (AUC) were analyzed to distinguish these thrombi. The associations between DECT parameters and proportion of RBCs were analyzed by Spearman's correlation.

Pathological analysis of 42 enrolled patients revealed 24 cases of RBC-dominant thrombi and 18 cases of F/P-dominant thrombi. Effective atomic numbers, spectral curve slopes, and polychromatic images CT values were significantly higher in the RBC-dominant thrombi group compared with the F/P-dominant thrombi group. Although the average thrombus attenuation increase was greater in the F/P-dominant thrombi group, this difference was not statistically significant. Among the DECT parameters, polychromatic images CT values had the greatest AUC at 0.924 (0.848-0.999) for discriminating RBC-dominant and F/P-dominant thrombi, with a threshold of 59 HU, sensitivity of 79.2 %, and specificity of 94.4 %. The combined diagnostic AUC reached 0.938 (0.863-1.012), with 87.5 % sensitivity and 94.4 % specificity. DECT polychromatic images CT values, effective atomic numbers, and spectral curve slopes were significantly correlated with proportion of RBCs (r = 0.673, 0.574, and 0.571, all p < 0.01).

DECT non-contrast scan parameters are associated with thrombus composition, which could be effective in distinguishing between RBC-dominant and F/P-dominant thrombi.