8 November 2022 in European Journal of Surgical Oncology
This study aimed to evaluate the prognostic value of lymph node dissection (LND) in node-negative intrahepatic cholangiocarcinoma (ICC) and identify the appropriately total number of lymph nodes examined (TNLE). Data from node-negative ICC patients who underwent curative intent resection in ten Chinese hepatobiliary centers from January 2010 to December 2018 were collected. Overall survival (OS), relapse-free survival (RFS) and postoperative complications were analyzed. Propensity score matching (PSM) was performed to reduce the bias due to confounding variables in LND group and non-lymph node dissection (NLND) group. The optimal TNLE was determined by survival analysis performed by the X-tile program using the enumeration method. A total of 637 clinically node-negative ICC patients were included in this study, 74 cases were found lymph node (LN) positive after operation. Among the remaining 563 node-negative ICC patients, LND was associated with longer OS but not RFS before PSM (OS: 35.4 vs 26.0 months, p = 0.047; RFS: 15.0 vs 15.4 months, p = 0.992). After PSM, patients in LND group had better prognosis on both OS and RFS (OS: 38.0 vs 23.0 months, p < 0.001; RFS: 15.0 vs 13.0 months, p = 0.029). There were no statistically differences in postoperative complications. When TNLE was greater than 8, OS (48.5 vs 31.1 months, p = 0.025) and RFS (21.0 vs 13.0 months, p = 0.043) were longer in the group with more dissected LNs. Routinely LND for node-negative ICC patients is recommended for it helps accurate tumor staging and associates with better prognosis. The optimal TNLE is more than 8.
16 September 2021 in Annals of Surgical Oncology
22 July 2025 in IEEE Transactions on Visualization and Computer Graphics
Point cloud sampling plays a crucial role in reducing computation costs and storage requirements for various vision tasks. Traditional sampling methods, such as farthest point sampling, lack task-specific information and, as a result, cannot guarantee optimal performance in specific applications. Learning-based methods train a network to sample the point cloud for the targeted downstream task. However, they do not guarantee that the sampled points are the most relevant ones. Moreover, they may result in duplicate sampled points, which requires completion of the sampled point cloud through post-processing techniques. To address these limitations, we propose a contribution-based sampling network (CS-Net), where the sampling operation is formulated as a Top-$k$ operation. To ensure that the network can be trained in an end-to-end way using gradient descent algorithms, we use a differentiable approximation to the Top-$k$ operation via entropy regularization of an optimal transport problem. Our network consists of a feature embedding module, a cascade attention module, and a contribution scoring module. The feature embedding module includes a specifically designed spatial pooling layer to reduce parameters while preserving important features. The cascade attention module combines the outputs of three skip connected offset attention layers to emphasize the attractive features and suppress less important ones. The contribution scoring module generates a contribution score for each point and guides the sampling process to prioritize the most important ones. Experiments on the ModelNet40 and PU147 showed that CS-Net achieved state-of-the-art performance in two semantic-based downstream tasks (classification and registration) and two reconstruction-based tasks (compression and surface reconstruction). CS-Net also achieved high average precision for objection detection on the KITTI LiDAR point cloud dataset, demonstrating its effectiveness in three-dimensional object detection.