A diagnostic tree for differentiation of adult pilocytic astrocytomas from high-grade gliomas. Issue 143 (October 2021)

• Record Type: Journal Article
• Authors: Park, Yae Won; Kim, Dain; Eom, Jihwan; Ahn, Sung Soo; Moon, Ju Hyung; Kim, Eui Hyun; Kang, Seok-Gu; Chang, Jong Hee; Kim, Se Hoon; Lee, Seung-Koo
• Is Part Of: European journal of radiology. Issue 143(2021)
• Publication Date: 2021-10
• Subjects: AUC area under the curve -- DTA decision tree analysis -- CART classification and regression tree -- CI confidence interval -- FLAIR fluid-attenuation inversion recovery -- HGG high-grade glioma -- IDH isocitrate dehydrogenase -- MRI magnetic resonance imaging -- PA pilocytic astrocytoma -- T1C postcontrast T1-weighted -- T1 T1-weighted -- T2 T2-weighted -- WHO World Health Organization; Decision tree -- Glioma -- Magnetic resonance imaging -- Pilocytic astrocytoma; Medical radiology -- Periodicals; Radiology -- Periodicals; Radiologie médicale -- Périodiques; Medical radiology; Periodicals; 616.075705
• DOI: 10.1016/j.ejrad.2021.109946
• Languages: English
• ISSNs: 0720-048X

Abstract

Abstract: Background: To develop a diagnostic tree analysis (DTA) model based on demographical information and conventional MRI for differential diagnosis of adult pilocytic astrocytomas (PAs) and high-grade gliomas (HGGs; World Health Organization grade III-IV). Methods: A total of 357 adult patients with pathologically confirmed PA (n = 65) and HGGs (n = 292) who underwent conventional MRI were included. The patients were randomly divided into training (n = 250) and validation (n = 107) datasets to assess the diagnostic performance of the DTA model. The DTA model was created using a classification and regression tree algorithm on the basis of demographical and MRI findings. Results: In the DTA model, tumor location (on cerebellum, brainstem, hypothalamus, optic nerve, or ventricle), cystic mass with mural nodule appearance, presence of infiltrative growth, and major axis (cutoff value, 2.9 cm) were significant predictors for differential diagnosis of adult PAs and HGGs. The AUC, accuracy, sensitivity, and specificity were 0.94 (95% confidence interval 0.86–1.00), 96.2%, 89.5%, and 97.7%, respectively, in the test set. The accuracy of the DTA model was significantly higher than the no-information rate in the test (96.2 % vs 85.0%, P < 0.001) set. Conclusion: The DTA model based on MRI findings may be useful for differential diagnosis of adult PA and HGGs.