Context-sensitive Classication Forests for Segmentation of Brain Tumor Tissues

摘要: We describe our submission to the Brain Tumor Segmenta- tion Challenge (BraTS) at MICCAI 2012, which is based on method for tissue-specic segmentation of high-grade brain tumors (3). The main idea cast as a classication task, and use discriminative power context information. realize this by equipping forest (CF) with spatially non-local features represent data, providing CF initial probability estimates single tissue classes additional input (along-side MRI channels). probabilities are patient-specic, com- puted test time learned model intensity. Through combination features, approach able capture information each data point. Our fully automatic, run times in range 1-2 minutes per patient. evaluate cross- validation real synthetic, high- low-grade tumor BraTS sets.