Segmentation of Liver Tumor based on Mobilenetv3 U-shape Net

Segmentation of tumor regions from medical images is crucial for clinical diagnosis. U-Net is a practical encoder-decoder framework proposed for better image segmentation. In segmentation of medical image, U-Net performs more excellent than other network. At present, medical image segmentation usually continuously deepens the network to improve the feature extraction capability for the accuracy of image segmentation, rarely considering the practical performance of the network. This paper aims to simplify the network feature extraction layer, proposed to segment the liver tumor area by mobilenetv3. Then, the attention mechanism is added to the skip connection layer of the encoder-decoder to increase the correlation ofit. Finally, Experimental results shows that the precision, recall, dice, and AUC were 0.8854, 0.9704, 0.9187, and 0.9864, respectively, outperforming other state-of-the-art networks. Therefore, the proposed method has the potential for the automatic segmentation of liver tumors in clinical setting.