Rethinking Pooling Operation for Liver and Liver-Tumor Segmentations

Deep convolutional neural networks (DCNNs) have been widely used in medical image segmentation due to their excellent feature learning ability. In these DCNNs, the pooling operation is usually used for image down-sampling, which can gradually reduce the image resolution and thus expands the receptive field of convolution kernel. Although the pooling operation has the above advantages, it inevitably causes information loss during the down-sampling of the pooling process. This paper proposes an effective weighted pooling operation to address the problem of information loss. First, we set up a pooling window with learnable parameters, and then update these parameters during the training process. Secondly, we use weighted pooling to improve the full-scale skip connection and enhance the multi-scale feature fusion. We evaluated weighted pooling on two public benchmark datasets, the LiTS2017 and the CHAOS. The experimental results show that the proposed weighted pooling operation effectively improve network performance and improve the accuracy of liver and liver-tumor segmentation.

Effect of Lipopolysaccharides on Liver Tumor Metastasis of twist1a/krasV12 Double Transgenic Zebrafish

The poor prognosis of patients diagnosed with hepatocellular carcinoma (HCC) is directly associated with the multi-step process of tumor metastasis. TWIST1, a basic helix-loop-helix (bHLH) transcription factor, is the most important epithelial-mesenchymal transition (EMT) gene involved in embryonic development, tumor progression, and metastasis. However, the role that TWIST1 gene plays in the process of liver tumor metastasis in vivo is still not well understood. Zebrafish can serve as a powerful model for cancer research. Thus, in this study, we crossed twist1a+ and kras+ transgenic zebrafish, which, respectively, express hepatocyte-specific mCherry and enhanced green fluorescent protein (EGFP); they also drive overexpression of their respective transcription factors. This was found to exacerbate the development of metastatic HCC. Fluorescence of mCherry and EGFP-labeled hepatocytes revealed that approximately 37.5% to 45.5% of the twist1a+/kras+ double transgenic zebrafish exhibited spontaneous tumor metastasis from the liver to the abdomen and tail areas, respectively. We also investigated the inflammatory effects of lipopolysaccharides (LPS) on the hepatocyte-specific co-expression of twist1a+ and kras+ in double transgenic zebrafish. Following LPS exposure, co-expression of twist1a+ and kras+ was found to increase tumor metastasis by 57.8%, likely due to crosstalk with the EMT pathway. Our results confirm that twist1a and kras are important mediators in the development of metastatic HCC. Taken together, our in-vivo model demonstrated that co-expression of twist1a+/kras+ in conjunction with exposure to LPS enhanced metastatic HCC offers a useful platform for the study of tumor initiation and metastasis in liver cancer.