Automatic brain tumor segmentation and tumor tissue classification based on multiple MR protocols

Introduction Segmentation of brain tumors in Magnetic Resonance (MR) images and classification of the tumor tissue into vital, necrotic, and perifocal edematous is required in a variety of clinical applications, as tumor diagnosis, grading and follow-up studies via volumetry [1,2], radiation therapy planning [3], surgery planning [4], or automatic region-of-interest segmentation for quantitative analysis, as vascularity-related parameters [5]. Manual delineation of the tumor tissue boundaries is a tedious and errorprone task, and the results are not reproducible. Hence, an automatic procedure for segmenting and classifying brain tumor tissue is needed. A variety of segmentation approaches for brain tumors in MR images can be found in the literature, which are mainly based on statistical [6-8] and variational [9-12] methods. Tumor tissue classification mostly requires information of several MR protocols and contrasts, as T1, T1contrast enhanced (T1CE), T2, FLAIR, MPRAGE, VASO [4,13-16]. The aim of this work was to realize a segmentation tool based on a 3D region growing algorithm for depiction of vital tumor, necrotic area and perifocal edema in T1CE and FLAIR images. Both image types are included in brain tumor protocols that are used in regular clinical routine.