Article Figures & Data
Figures
- FIG 1.
Axial FLAIR (A and D), ADC maps (B and E), and manual tumor segmentation using semiautomatic tools in 3D Slicer (C and F). Upper row: a 15-year-old adolescent boy with a supratentorial BRAF-mutated low-grade glioma. Lower row: a 3-year-old boy with a cerebellar BRAF-fused low-grade glioma.
- FIG 2.
Flow chart of the study.
- FIG 3.
The repetitive classification approach.
- FIG 4.
Boxplots of test AUCs. Note that P values are calculated on the basis of Student t tests.
Tables
Institutional Cohort Total Toronto Stanford No. of patients 229 23 252 Age (mean) (yr) 8.18 6.32 8.01 Male (No.) (%) 124 (54.1%) 11 (47.8%) 135 (53.6%) Histologic diagnosis (No.) PA 112 16 128 GG 24 5 29 LGA 36 0 36 PMA 8 2 10 PXA 4 0 4 DNET 10 0 10 DA 17 0 17 ODG 3 0 3 GNT 5 0 5 ACG 5 0 5 NC 1 0 1 DIG 1 0 1 PLNTY 1 0 1 Mixed tumor components 2 0 2 Molecular subgroup (No.) (%) KIAA1549-BRAF fusion 114 (49.7%) 18 (78.3%) 132 (52.2%) BRAF V600E mutation 36 (15.7%) 5 (21.7%) 41 (16.6%) Non-BRAF 79 (34.4%) 0 79 (31.2%) Supratentorial (No.) (%) 125 (54.5%) 5 (21.7%) 130 (51.8%) Infratentorial (No.) (%) 104 (45.4%) 18 (78.3%) 122 (48.2%) Note:—ACG indicates angiocentric glioma; DA, diffuse astrocytoma; DIG, desmoplastic infantile ganglioglioma; DNET, dysembryoplastic neuroepithelial tumor; GNT, glioneural tumor; LGA, low-grade astrocytoma; NC, neurocytoma; PMA, pilomyxoid astrocytoma; PXA, pleomorphic xanthoastrocytoma; PLNTY, polymorphous low-grade neuroepithelial tumor of the young; ODG, oligodendroglioma.
Type Source Feature Category Feature Clinical NA NA Tumor location Radiomics 3D wavelet transform Gray-level difference matrix Large dependence high gray-level emphasis Radiomics 3D wavelet transform First order Mean Radiomics 3D wavelet transform First order Kurtosis Radiomics Original Gray-level difference matrix Dependence variance Radiomics Original Gray-level difference matrix Large dependence high gray-level emphasis Radiomics Logarithm Gray-level difference matrix Large dependence high gray-level emphasis Radiomics Local binary pattern 3D Gray-level difference matrix Large dependence emphasis Radiomics Exponential First order 90th Percentile Radiomics 3D wavelet transform First order Kurtosis Note:—NA indicates not applicable.
KIAA1549-BRAF Fusion versus the Rest BRAF V600E Mutation versus the Rest Non-BRAF versus the Rest Average Radiomics features only AUC = 0.80 95% CI, 0.79–0.81 AUC = 0.74 95% CI, 0.73–0.75 AUC = 0.67 95% CI, 0.66–0.68 AUC = 0.74 95% CI, 0.73–0.75 Clinical features only AUC = 0.75 95% CI, 0.74–0.7 AUC = 0.62 95% CI, 0.61–0.63 AUC = 0.63 95% CI, 0.62–0.6 AUC = 0.67 95% CI, 0.66–0.68 Combined radiomics and clinical features AUC = 0.81 95% CI, 0.81–0.82 AUC = 0.75 95% CI, 0.73–0.75 AUC = 0.74 95% CI, 0.672–0.74 AUC = 0.76 95% CI, 0.75–0.7)
