We introduce PREDATOR, a model for pairwise pointcloud registration with deep attention to the overlap region. Different from previous work, our model is specifically designed to handle (also) point-cloud pairs with low overlap. Its key novelty is an overlap-attention block for early information exchange between the latent encodings of the two point clouds. In this way the subsequent decoding of the latent representations into per-point features is conditioned on the respective other point cloud, and thus can predict which points are not only salient, but also lie in the overlap region between the two point clouds. The ability to focus on points that are relevant for matching greatly improves performance: PREDATOR raises the rate of successful registrations by more than 20% in the low-overlap scenario, and also sets a new state of the art for the 3DMatch benchmark with 89% registration recall. Source code and pretrained models will be available at https://github. com/ShengyuH/OverlapPredator.
Aberrant chromobox (CBX) family protein expression has been reported in a variety of human malignancies. However, the role of CBX6 in hepatocellular carcinoma (HCC) progression and patient prognosis remains unknown. In this study, we found that CBX6 was frequently up-regulated in HCC clinical samples and HCC cell lines and that CBX6 expression was significantly correlated with larger tumor sizes (≥ 5 cm, p = 0.011) and multiple tumors (n ≥ 2, p = 0.018). Survival analyses indicated that patients with higher CBX6 expression levels had significantly shorter recurrence-free survival (RFS) and overall survival (OS) than patients with lower CBX6 expression levels, and multivariate analyses confirmed that increased CBX6 expression was an independent unfavorable prognostic factor for HCC patients. Functional study demonstrated that CBX6 profoundly promoted HCC cell growth both in vitro and in vivo, and mechanistic investigation revealed that the S100A9/NF-κB/MAPK pathway was essential for mediating CBX6 function. In conclusion, our results represent the first evidence that CBX6 contributes to tumor progression and indicate that the protein may serve as a novel prognostic biomarker for HCC and as a therapeutic target in the treatment of the disease.
Background
Portal vein tumor thrombosis (PVTT) in hepatocellular carcinoma (HCC) is a sign of advanced stage disease, which is associated with poor prognosis. Liver resection (LR) may provide better prognosis in selected patients. In the present study, we aimed to assess information from HCC patients with PVTT who died within 3 months or 2 years after LR in order to identify preoperative factors correlated to short-term or long-term survival, by which inappropriate selection of patients for LR might be avoided in the future.
Methods
A retrospective cohort study consisting of 487 consecutive cases of HCC patients with PVTT was performed from 2008 to 2010 at Eastern Hepatobiliary Surgery Hospital. Medical records, including laboratory values, imaging results and treatment information, were obtained from participants. Study endpoints were survival at 3 months and 2 years post-hepatectomy. Logistic regression analysis was utilized to determine the significant pre-operative factors influencing short-term or long-term survival.
Results
In multivariable analysis, α-fetoprotein, total bilirubin and radiologic ascites were significantly associated with short-term survival, while α-fetoprotein level, clinical significant portal hypertension, extent of PVTT and tumor differentiation were factors significantly associated with long-term survival.
Conclusions
The independent risk factors of poor short-term survival were the liver function-associated, such as factors radiologic ascites and total bilirubin, while tumor differentiation indicating the tumor biology was associated with longer-term survival. In addition, α-fetoprotein was a risk factor associated with both short-term and longer-term survivals.
Electronic supplementary material
The online version of this article (10.1186/s13578-019-0285-z) contains supplementary material, which is available to authorized users.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.