Deep segmentation of the liver and the hepatic tumors from abdomen tomography images

A pipelined framework is proposed for accurate, automated, simultaneous segmentation of the liver as well hepatic tumors from computed tomography (CT) images. The introduced composed three levels. First, two different transfers deep convolutional neural networks (CNN) are applied to get high-level compact features CT Second, a pixel-wise classifier used obtain output-classified maps each CNN model. Finally, fusion network (FNN) integrate maps. Experimentations performed on MICCAI’2017 database tumor (LITS) challenge, result in dice similarity coefficient (DSC) 93.5% and 74.40% lesion, using 5-fold cross-validation scheme. Comparative results with state-of-the-art techniques same data show competing performance segmentation.