Absolute hypodensity sign by noncontrast computed tomography as a reliable predictor for early hematoma expansion

Brain Hemorrhages(2020)

引用 1|浏览3
暂无评分
摘要
Early hematoma expansion in intracerebral hemorrhage (ICH) patients is strongly associated with poor outcome and potentially preventable if high-risk patients are timely identified. We aimed to investigate whether a novel sign detected by noncontrast computed tomography (NCCT) named absolute hypodensity sign (AHS) was an appropriate candidate for hematoma expansion prediction. Spontaneous ICH patients all underwent baseline NCCT scan within 6 h after ICH onset and the follow-up NCCT scan within 24 h after initial CT scan. AHS was defined as a hypoattenuating area with a distinct margin within the hematoma, the minimal density of which should be ≤30 Hounsfeld units. We used univariate and multivariate logistic regression analyses for determining the association between hematoma expansion and the presence of AHS. We also compared the diagnostic efficacy with blend sign and black hole sign. A total of 348 ICH patients were included, 87 of which were found hematoma expansion. AHS was positive in 76 patients (21.8%) (κ=0.88 for interrater reliability) and 63.2% of the patients with hematoma expansion were observed AHS. The multivariate logistic regression analysis demonstrated AHS was an independent predictor for hematoma expansion (odds ratio: 33.454, 95% confidence interval: 13.628-82.126, P<0.001). The sensitivity, specificity, positive predictive value, negative predictive value and accuracy of AHS for hematoma expansion prediction were 63.2%, 92.0%, 72.4%, 88.2%, 84.8%, respectively, which were higher than blend sign and black hole sign except the specificity of black hole sign. Our findings indicate AHS is a reliable predictor for early hematoma expansion due to its feasibility and convenience for clinical practice.
更多
查看译文
关键词
Intracerebral hemorrhage,Hematoma expansion,Noncontrast computed tomography,Absolute hypodensity sign
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要