Brain tumor segmentation using cascaded deep convolutional neural network

Gliomas are the most common and threatening brain tumors with little to no survival rate. Accurate detection of such tumors is crucial for survival of the subject. Naturally, tumors have irregular shape and can be spatially located anywhere in the brain, which makes it a challenging task to segment them accurately enough for clinical purposes. In this paper, an automated segmentation algorithm for brain tumor using deep convolutional neural networks (DCNN) is proposed. Deep networks tend to have a lot of parameters thus over-fitting is almost always an issue especially when data are sparse. Max-out and drop-out layers are used to reduce the chances of over-fitting since data are scant. Patch based training method is used for the model where two types of patches sized 37×37 and 19×19 with same center pixel are selected. The proposed algorithm includes preprocessing in which images are normalized and bias field corrected, and post processing where small false positives are removed using morphological operators. BRATS 2013 dataset is used for evaluation of the proposed method, where it outperforms state-of-the-art methods with similar settings in key performance indicators.