Liver tumor segmentation plays a critical role in medical imaging analysis, especially in cancer diagnosis and treatment planning. Accurate segmentation of liver tumors is essential. There has been growing interest in developing automated methods for liver tumor segmentation, practically using deep learning algorithms. In this paper, we use the convolutional neural networks (CNNs), on large datasets of liver images, labelled with the tumor regions. A U-net architecture is builtwith cross connections. The encoder and decoder portion of the Unet is built using a 34 layerResNet. The model was trained on Liver Tumor Segmentation challenge (LITS) dataset of liver CT scans and evaluated on a separate test dataset. The results demonstrated that the proposed method produced great segmentation accuracy and an average dice score of 0.95