Differentiation of glioblastoma from solitary brain metastases using radiomic machine-learning classifiers. (1st June 2019)

• Record Type: Journal Article
• Authors: Qian, Zenghui; Li, Yiming; Wang, Yongzhi; Li, Lianwang; Li, Runting; Wang, Kai; Li, Shaowu; Tang, Ke; Zhang, Chuanbao; Fan, Xing; Chen, Baoshi; Li, Wenbin
• Is Part Of: Cancer letters. Volume 451(2019)
• Page Start: 128
• Page End: 135
• Publication Date: 2019-06-01
• Subjects: Brain metastases -- Glioblastoma -- Radiomics -- Machine learning; GBM glioblastoma -- MET metastases -- AUC area under the curve -- RSD relative standard deviation -- ROC receiver operating characteristic -- SVM support vector machines -- LASSO least absolute shrinkage and selection operator -- CE contrast-enhancement -- TCGA the Cancer Genome Atlas -- T1 T1-weighted -- T2 T2-weighted -- ROI region of interest -- LoG Laplacian of Gaussian -- ICCs intraclass correlation coefficients -- T-test-score TSCR -- RELF Relief -- IFGN information gain -- GNRO gain ratio -- EUDT Euclidean distance -- FAOV F-anova -- WLCX Wilcoxon rank sum -- LG logistic regression -- MUIF mutual information -- SVM support vector machine -- LASSO least absolute shrinkage and selection operator -- RF random forest -- Ada Adaboost Classifier -- KNN k-nearest neighbor -- MLP multi-layer perceptron -- DT decision tree -- NB naïve Bayes -- ACC accuracy -- PPV positive prediction value -- NPV negative predictive value; Cancer -- Periodicals; Neoplasms -- Periodicals; Cancer -- Périodiques; Electronic journals; 616.994
• DOI: 10.1016/j.canlet.2019.02.054
• Languages: English
• ISSNs: 0304-3835

Abstract

Abstract: This study aimed to identify the optimal radiomic machine-learning classifier for differentiating glioblastoma (GBM) from solitary brain metastases (MET) preoperatively. Four hundred and twelve patients with solitary brain tumors (242 GBM and 170 solitary brain MET) were divided into training (n = 227) and test (n = 185) cohorts. Radiomic features extraction was performed with PyRadiomics software. In the training cohort, twelve feature selection methods and seven classification methods were evaluated to construct favorable radiomic machine-learning classifiers. The performance of the classifiers was evaluated using the mean area under the curve (AUC) and relative standard deviation in percentile (RSD). In the training cohort, thirteen classifiers had favorable predictive performances (AUC≥0.95 and RSD ≤6). In the test cohort, receiver operating characteristic (ROC) curve analysis revealed that support vector machines (SVM) + least absolute shrinkage and selection operator (LASSO) (AUC, 0.90) classifiers had the highest prediction efficacy. Furthermore, the clinical performance of the best classifier was superior to neuroradiologists in accuracy, sensitivity, and specificity. In conclusion, employing radiomic machine-learning technology could help neuroradiologist in differentiating GBM from solitary brain MET preoperatively. Highlights: Radiomic analysis could help to differentiate brain metastases from glioblastoma. LASSO combined with SVM achieved the best performance in an independent test cohort. The best radiomic classifier performed much better than neuroradiologists.