Identifying clinically applicable machine learning algorithms for glioma segmentation: recent advances and discoveries