Knowledge of molecular status for a variety of markers in gliomas has moved to the forefront in clinical decision-making. This requires direct tissue sampling either from an invasive brain biopsy or open surgical resection. In this project, we are developing and validating a non-invasive method to determine a panel of the most critical molecular markers (IDH, 1p/19q and MGMT methylation) with near tissue-level accuracies using routine T2-weighted (T2w) MR images and deep learning algorithms to significantly reduce and in many cases eliminate the need for stereotactic biopsy in glioma.