Automatic segmentation of brain tumors in magnetic resonance images

Segmentation of tumors in magnetic resonance images (MRI) is an important task but is quite time consuming when performed manually by experts. Automating this process is challenging due to the high diversity in appearance of tumor tissue in different patients, and in many cases, similarity between tumor and normal tissues. This paper presents an automatic method for segmentation of brain tumors in MRI. We use images of patients with glioblastoma multiform tumors. After pre-processing and removal of the regions that do not have useful information (e.g., eyes and scalp), we create a projection image for determining the primary location of the tumor. This image provides an overall view of the tumor. Then, we grow the primary region to segment the entire tumor. This method is automatic and independent of the operator. It segments low contrast tumors without requiring their exacta tissue boundaries. The segmentation results obtained by the proposed approach are compared with those of an expert radiologist showing excellent correlations among them (R²=0.97).