Whole tumor ADC-derived texture features in grading of brain glioma
10.13929/j.1003-3289.201811139
- Author:
Dongdong MEI
1
Author Information
1. Department of Radiology, Shenzhen People's Hospital, Jinan University
- Publication Type:Journal Article
- Keywords:
Apparent diffusion coefficient;
Glioma;
Magnetic resonance imaging;
Radiomics
- From:
Chinese Journal of Medical Imaging Technology
2019;35(7):976-980
- CountryChina
- Language:Chinese
-
Abstract:
Objective: To investigate the value of whole tumor texture features derived from ADC mapping in distinguishing high grade glioma (HGG) from low grade glioma (LGG) of brain. Methods: Totally 66 patients with pathologic proven brain glioma were enrolled, including 41 HGGs and 25 LGGs. Then 107 texture features were derived from whole tumor ADC mapping. The texture features and clinical characteristics were compared, and the variates with statistical significance at univariate analysis were entered into Logistic analysis to find out the independent risk factors for HGG. ROC curves were constructed to determine the diagnostic performance of HGG. Results: The univariate analysis revealed that the gender and age of patients as well as 3 texture features were different between HGGs and LGGs. Logistic analysis showed that age (P=0.002, OR=1.090) and ZoneEntropy (P=0.003, OR=2.984) were independent risk factors for HGG. Combining age and ZoneEntropy, the AUC of identifying HGG was 0.844, with a sensitivity of 75.6% and a specificity of 88.0%. Conclusion: The whole tumor ADC-derived texture features are useful for grading of brain glioma grade. Combining texture features with clinical characteristics can obtain high diagnostic performance.
