Computer-extracted MR imaging features are associated with survival in glioblastoma patients.

摘要: Automatic survival prognosis in glioblastoma (GBM) could result improved treatment planning for the patient. The purpose of this research is to investigate association GBM patients with tumor features pre-operative magnetic resonance (MR) images assessed using a fully automatic computer algorithm. MR imaging data 68 from two US institutions were used study. obtained Cancer Imaging Archive. A vision algorithm was applied segment and extract eight MRI studies. included side, proportion enhancing tumor, necrosis, T1/FLAIR ratio, major axis length, minor volume, thickness margin. We constructed multivariate Cox proportional hazards regression model likelihood ratio test establish whether are prognostic survival. also evaluated individual value each feature through analysis univariate models feature. found that automatically extracted predictive (p = 0.031). Multivariate showed survival: (p = 0.013), length (p = 0.026). Univariate indicated same as significant (p = 0.021, p = 0.017 respectively). conclude that computer-extracted can be patients.