Your browser doesn't support javascript.
loading
The value of wholeGtumor texture analysis of contrastGenhanced T1 WI in differentiating cystic glioma from brain abscess / 实用放射学杂志
Journal of Practical Radiology ; (12): 857-860,868, 2019.
Article em Zh | WPRIM | ID: wpr-752449
Biblioteca responsável: WPRO
ABSTRACT
Objective To evaluate the feasibility of wholeGtumor texture analysis of contrastGenhanced T1 WI (T1 Ce)in differentiating cystic glioma (CG)from brain abscess (BA).Methods MRI data of 25 cases of ringGenhanced CG and 24 cases of BA proven pathologically were retrospectively studied.All the patients underwent preGsurgery MRI plain and contrastGenhanced scans.FireVoxel software was used to outline the ROI of the wholeGtumor.The signal intensity histogram and related texture parameters of the 3D ROI were obtained,including mean, median,standard deviation,inhomogeneity,skewness,kurtosis and entropy.The data were first tested for normality and the differences in wholeG tumor texture analysis parameters of T1 Ce between CGs and BAs were compared using the independentGsample t test(normal distribution)and MannGWhitney rank sum test (skewed distribution).ROC curve was used to evaluate the efficacy of the parameter in differentiating CG from BA.Results There were statistical significances in the parameters of mean,median,standard deviation,inhomogeneity and skewness between CGs and BAs(P<0.05),and there were no any statistical significances in kurtosis and entropy between CGs and BAs(P>0.05).In all the texture parameters,the AUC of inhomogeneity was the largest(0.988),and when the threshold was 0.314, the sensitivity and the specificity were 92.60% and 9 7.1 0%,respectively.Conclusion Some of the quantitative parameters of the wholeGtumor texture analysis of T1 Ce(mean,median,standard deviation,inhomogeneity and skewness)could provide reliable and objective evidences for imaging differential diagnosis of CG and BA preGsurgery.
Palavras-chave
Texto completo: 1 Base de dados: WPRIM Tipo de estudo: Prognostic_studies Idioma: Zh Ano de publicação: 2019 Tipo de documento: Article
Texto completo: 1 Base de dados: WPRIM Tipo de estudo: Prognostic_studies Idioma: Zh Ano de publicação: 2019 Tipo de documento: Article