The Diagnostic Value of MRI-Based Texture Analysis in Discrimination of Tumors Located in Posterior Fossa: A Preliminary Study

Objectives To investigate the diagnostic value of MRI-based texture analysis in discriminating common posterior fossa tumors: medulloblastoma, brain metastasis and hemangioblastoma. Methods A total number of 185 patients were included in this study, 63 with medulloblastoma, 56 with brain metastasis and 66 with hemangioblastoma. Texture features derived from Histogram based matrix and Grey-level co-occurrence matrix were extracted from contrast-enhanced T1-weighted (T1C) image and fluid-attenuation inversion recovery (FLAIR) image. Mann-Whitney U test was conducted to identify whether texture features were significantly different. Binary logistic regression analysis was performed to assess if they could be taken as independent predictors, which were combined to build integrated models. Receiver operating characteristic analysis was conducted to evaluate their practical value in discrimination. Results Several texture features from T1C image and FLAIR image were found to be significantly different in discrimination. The integrated models represented higher discriminatory power than any single texture feature, with the AUC of 0.808 (medulloblastoma versus hemangioblastoma), 0.871 (medulloblastoma versus brain metastasis) and 0.880 (brain metastasis versus hemangioblastoma). Conclusion Texture analysis had the potential to be served as a feasible method in the differentiation among common posterior fossa tumors. Moreover, the integration of single texture feature displayed more promising value in discrimination.